Antibacterial activity of glycine betaine analogues: Involvement of osmoporters

Article abstract of DOI:10.1016/j.bmcl.2004.02.045

Glycine betaine (GB) analogues were obtained using solid phase organic synthesis and assayed for their toxic activity against 15 Gram positive and Gram negative bacteria. Four benzyl derivatives of GB were selected to determine their effect on bacterial g

Full text of DOI:10.1016/j.bmcl.2004.02.045

Bioorganic & Medicinal Chemistry Letters 14 (2004) 2061–2065
Antibacterial activity of glycine betaine analogues: involvement
of osmoporters
Anne Cosquer,a Morgane Ficamos,b Mohamed Jebbar,a Jean-Charles Corbel,b
a
Gwenaelle Choquet, Catherine Fontenelle, Philippe Uriac and Theophile Bernard
a
b
a,
*
ꢀ €
ꢀ
aUniversitꢀe de Rennes I, UMR CNRS 6026, Interactions Cellulaires et Molꢀeculaires, Equipe Osmoadaptation chez les Bactꢀeries,
Campus de Beaulieu, 35042 Rennes, France
bInstitut de Chimie de Rennes, Synthꢁese et Extraction de Molꢀecules ꢁa visꢀee Thꢀerapeutique, UPRES 2234, Facultꢀe de Pharmacie,
2 Avenue du Pr L. Bernard, 35043 Rennes, France
Received 22 December 2003; revised 9 February 2004; accepted 12 February 2004
Abstract—Glycine betaine (GB) analogues were obtained using solid phase organic synthesis and assayed for their toxic activity
against 15 Gram positive and Gram negative bacteria. Four benzyl derivatives of GB were selected to determine their effect on
bacterial growth. Bacteriostatic and lethal effects were observed for compound 1 and compound 2, respectively. The importation of
the two GB analogues into bacterial cells appeared strictly dependent on the presence of the powerful betaine membrane osmo-
porters; their capacity to be amassed intracellularly at molar levels from extremely dilute solutions might constitute a basis to design
a new class of antimicrobial agents.
Ó 2004 Elsevier Ltd. All rights reserved.
1. Introduction
Such properties suggest that betaine porters could be
very efficient ways to deliver toxic GB analogues, at a
lethal level, into pathogenic bacteria, particularly those
invading an hyperosmotic environment like the urinary
tract,7 salted foods, plant leaves and animal skin sur-
faces. Interestingly it has already been reported that
structural analogues of GB exert a toxic activity against
the Gram negative soil bacterium and plant microsym-
biont Sinorhizobium meliloti.8
Numerous species of bacteria cope with elevated
osmolality by importing compatible solutes (osmopro-
tectants) from their environment.1–3 Glycine betaine
(N,N,N-trimethylglycine, GB), the most powerful
osmoprotectant identified so far, is internalised through
unspecific but highly efficient membrane carriers
(osmoporters) and can be amassed intracellularly up to
molar levels2 (i.e., 103–109 times that of the medium)
depending on the degree of the osmotic constraint and
the concentration of the osmoprotectant in the med-
ium.4 Examples of the most studied osmoporters are
provided by ProU and ProP in the Gram negative
models Escherichia coli and Salmonella typhimurium1;5
and by OpuA, OpuC and OpuD in the Gram positive
model Bacillus subtilis.6 All these membrane carriers
display a high affinity towards glycine betaine allowing
concentration as low as one nanomolar to be effective in
osmoprotection;4 moreover these carriers are strikingly
activated when bacterial cells have to face hyperosmotic
conditions.
In this paper we report upon seven novel GB analogues
obtained by solid phase synthesis and bearing a substi-
tuted benzyl moiety in the place of one methyl group.
Two of these GB analogues were recently shown to
express a cytotoxic activity against pathogenic strains of
E. coli.9 Thus, in order to perform a more complete
biological study, we selected four compounds, which
were prepared on a larger scale, using a classic method
for synthesis, and assayed for their antimicrobial activ-
ities against 15 Gram negative and Gram positive bac-
terial species.
2. Chemistry
Keywords: Betaine; Toxic analogues; Bacteria.
*Corresponding author. Tel.: +33-2-2323-6853; fax: +33-2-2323-6775;
e-mail: tbernard@univ-rennes1.fr
In our previous work dedicated to synthesis of GB
analogues,8 alkylation of Wang resin supported
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.02.045

2062
A. Cosquer et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2061–2065
O
bacterial growth (data not shown) led to the selection of
H C
3
C
HOBT
the four most active compounds: 1, 2, 3 and 4. In
addition, a MIC distribution of 1 and 4 against 82
strains of E. coli isolated from infected urinary tracts has
already been reported.9
1)
2)
N
CH
2
CH3
N
H C
3
OH
O
C
O
CH2
CH2
, X
Z
CH3
CH2
X
Z
Fifteen bacterial strains (listed in Table 1) were assayed
for their growth in the presence of these compounds.
The experiments were conducted aerobically in classic
defined media (M63, M9, SMM or LAS), using NaCl
to impose hyperosmotic conditions (0.3–0.5 M final
concentration, depending on the degree of tolerance of
the bacterium). Growth yield for each culture was
expressed as the absorbance (optical density) of the
suspension at 570 nm (A570). GB and its analogues were
supplied at 1 mM each.
TFA
CH2Cl2
CH3
N
HO
C
O
CH2
CH2
, X
Z
CH3
(2) : X = Br, Z = 4-NO2
;
(3) : X = Br, Z = 2-NO2
;
(1) : X = Cl or Br, Z = H ;
(4) : X = Cl, Z = 3-OCH3 ; (5) : X = Br, Z = 3-I ;
(6) : X = Br, Z = 4-CN ;
O
OH
(7 ,8) : X = Br, Z =
4-
85 % ,
15%
B
B
4-
X = Br, Z =
O
OH
On the basal medium, without added NaCl, except for
S. enterica treated with compounds 3 and 4, no signifi-
cant growth inhibition was observed (data not shown).
Under osmotically stressing conditions (Table 1), addi-
tion of compound 2 to the culture medium was followed
by a severe loss of biomass production for a large
majority (12 out of 15) of the species assayed; almost no
growth (more than 92% inhibition) was observed for
A. hydrophila, C. freundii, E. chrysanthemi, E. coli,
S. enterica and S. marcescens. Only B. linens, E. faecalis
and S. meliloti have their growth either not affected or
rather slightly improved in the presence of 2. The level
of bacterial production, following treatment with 3, was
almost identical to that resulting from treatment with 2.
It should be mentioned that only B. subtilis was insen-
sitive to that GB analogue.
Scheme 1.
bromoacetic acid with tertiary amines was used. Here,
another synthetic pathway was employed consisting of
alkylation of dimethylglycine10 linked to Wang resin by
an ester (Scheme 1). Complete linkage of dimethylgly-
cine to Wang resin (assessed using FTIR and 13C gel
NMR) was achieved after activation as HOBT ester,
which also enabled dimethylglycine dissolution. Then,
diversity was introduced with selected benzylhalides,
which were either commercially available or prepared
according to the literature.11 During the cleavage, per-
formed using a dichloromethane solution of trifluoro-
acetic acid, the bromide (or chloride) anion was not
exchanged with trifluoroacetate anion, as demonstrated
by MS (negative mode). Structure and purity12 were
checked by H NMR, 13C NMR, high-voltage electro-
1
Strains for which the growth was inhibited by 1 are also
those affected by 2 except B. subtilis, C. freundii, E.
cloacae and S. enterica. In contrast, compound 1 exerted
a beneficial effect on the growth of C. freundi and
S. meliloti with 21% and 57% yield improvement,
respectively. Among the four analogues assayed, com-
pound 4 showed the lowest toxicity. It appeared more or
less effective (more than 27% growth inhibition) against
A. hydrophila, C. glutamicum, E. chrysanthemi, E. coli,
K. pneumoniae and S. marcescens.
phoresis on Whatman 3 MM paper13 and chromatog-
raphy (TLC and HPLC). This method14 allowed us to
prepare seven GB analogues (1–7) with very good yields
(>90%) and with a high level of purity (>95%). The
preparation of larger amounts of compounds 1–4 for the
biological studies was performed by alkylation of the
dimethylglycine ethyl ester according to the literature.15
In addition, 14C radiolabelled compounds 1r and 2r were
synthesised (Scheme 2) by direct alkylation of radio-
labelled 14C [carboxy]bromoacetic acid in DMF, using
an excess of tertiary amines 9 and 10. The amine 10 was
prepared by alkylation of dimethylamine by 4-nitro-
benzylbromide.16 Purification of 1r and 2r was achieved
using preparative high-voltage electrophoresis.13
Thus, for most of the strains submitted to treatment,
compounds 1 and 2, and to a lesser extent compound 3
caused the most inhibitory effect. Only S. meliloti and E.
faecalis were not responsive to any of the GB analogues.
To determine whether an active transport, through
betaine carriers, was required to cause bacterial toxicity,
mutants were used. E. coli wild-type MC4100 (WT) and
its derivatives BK 32 (proP), GM50 (proU) and MKH13
(proU proP) affected in osmoporters ProU and/or ProP
synthesis were grown on the minimal medium M63 with
glucose as the carbon and energy sources. Elevation of
medium osmolality was made by addition of NaCl to a
final concentration of 0.3 M. Betaine analogues were
supplied at 1 mM each, as required.
O
CH3
DMF
C*
CH2
OH
+
1r, 2r
Z
CH2
N
RT
CH3
Br
(9): Z = H ; (10): Z = NO2
Scheme 2.
3. Biological evaluation
A preliminary broad screening carried out to determine
the effect of the seven synthesised GB analogues on
Addition of salt to the culture medium altered the
growth rate significantly in all the E. coli strains assayed

A. Cosquer et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2061–2065
2063
Table 1. Effect of structural analogues of glycine betaine (GB) on bacterial growth under stressing conditionsa
Bacteria
Betaines
0
1
2
3
4
Acinetobacter baumannii
Aeromonas hydrophila
Bacillus subtilis
1.14
1.65
3.20
1.76
1.09
0.87
1.25
1.16
1.97
1.24
1.80
1.24
1.23
1.03
1.46
0.93
0.16
3.20
0.88
1.32
0.43
1.05
1.24
0.22
0.12
0.46
0.06
1.24
0.21
2.30
0.19
0.06
0.18
1.74
0.03
0.33
0.24
1.23
0.06
0.09
0.23
0.16
0.07
0.07
1.85
0.72
0.20
3.06
1.52
0.06
0.26
0.46
1.30
0.11
0.08
0.40
0.11
0.23
0.09
1.90
1.16
1.05
3.20
1.45
1.23
0.63
1.11
1.12
0.25
0.18
0.70
1.07
1.11
0.48
1.70
Brevibacterium linens
Citrobacter freundii
Corynebacterium glutamicum
Enterobacter cloacae
Enterococcus faecalis
Erwinia chrysanthemi
Escherichia coli
Klebsiella pneumoniae
Pseudomonas aeruginosa
Salmonella enterica
Serratia marcescens
Sinorhizobium meliloti
GB derivatives (1, 2, 3 and 4) were supplied at 1 mM each. Bacterial yield was expressed as absorbance (OD) at 570 nm.
a Defined classic culture media: M63 with glucose as carbon source except for Enterococcus (M9 medium), S. meliloti, P. aeruginosa and B. linens
(LAS medium) and B. subtilis (SMM). NaCl concentration was adapted to each strain to allow a significant growth yield (0.5 M for Bl, Ef and Pa,
0.3 M for Ec, and 0.4 M for all other species). Bold numbers indicate the most prominent growth inhibiting activities. Data represent the average of
at least three experiments; deviation was within 10%.
(Fig. 1). About a 50% decrease (from 0.85 to 0.46 gen-
eration/hour without and with added NaCl, respec-
tively) was observed. Supplying 1 and 2 to their
non-stressing medium (M63 0 M NaCl) did not affect
the behaviour of either MC4100 or that of the three
mutants (data not shown). An elevation of medium
salinity (osmolality) in the presence of the two GB
analogues led to a dramatic slow down of the growth
rate of MC4100, BK32 and GM50; in fact only weak
growth was observed in the presence of 1 and no growth
in the presence of 2. Interestingly, MKH13, the mutant
deprived of both osmoporters appeared insensitive to 1
and 2 treatments. Identical conclusions were drawn
from experiments carried out on E. chrysanthemi double
mutant (ousA, ous B) and B. subtilis mutant (opuC); all
these strains are known to be affected in betaine trans-
port. Moreover, withdrawing compounds 1 and 2 from
the salted medium permitted bacterial growth to resume
for cells treated with 1; in contrast those treated with
compound 2 did not show any more growth.
In conclusion we infer that an active transport, through
osmoporters, is required for the toxic activity of com-
pounds 1 and 2 and that both growth inhibitors display
a differential effect, with only that of 2 being lethal.
To understand why 1 and 2 exerted a differential effect,
cultures in mid exponential phase of E. coli MC4100
grown in M63 and M63-0.3 M NaCl were supplied with
radiolabelled GB analogues 1r and 2r;17 bacterial cells
were harvested by centrifugation and concentrated to an
OD570 ¼ 10. A rapid screening of other bacteria grown in
their corresponding defined medium was also carried
out.
Compounds 1r and 2r, 14C-labelled on their C1 carbon
were actively imported by the bacterial cells and, as
expected, their transport activity was severely enhanced
by salt addition to the assay medium. For E. coli,
analysis of chromatographic data from compound 1
clearly showed that only one labelled solute was present
in the bacterial extract (Fig. 2); moreover, no radioac-
tivity was found either in evolved CO2 or in the insol-
uble pellet, even after 10 h of incubation. That means the
totality of absorbed 1r remained unchanged in E. coli. In
contrast, compound 2r was rapidly metabolised, leading
to labelled compounds, among which only dimethyl-
glycine, also known as a demethylated catabolite of
glycine betaine, was identified by co-chromatography
(Fig. 2). The other part of the molecule, not labelled,
produced a yellow colour in the culture medium and
could be attributed to 4-nitrobenzaldehyde. The degra-
dation level of 2r was time-dependent; almost all the
substrate had disappeared after 1 h of incubation. Data
from other strains clearly demonstrated a behaviour of
Figure 1. Growth rate of E. coli strains on M63 medium without NaCl
(M1), with added 0.3 M NaCl (M2) and with added 0.3 M
NaCl + compound 1 (M3). WT (wild type MC4100), BK32 (mutant
proP), GM50 (mutant proU), MKH13 (double mutant proUproP).

2064
A. Cosquer et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2061–2065
Figure 2. Electronic autoradiography of a 2D-TLC analysis of the soluble fraction of E. coli MC4100 cells treated with compounds 1r (60 min) and 2r
(15 and 60 min). Solvent a: n-butanol–acetic acid–water (12:3:5); solvent b: phenol–ethanol–ammonia (50:50:0.5). X: unknown radiolabelled com-
pound.
4. Cosquer, A.; Pichereau, V.; Pocard, J. A.; Minet, J.;
Cormier, M.; Bernard, T. Appl. Environ. Microbiol. 1999,
65, 3304.
5. Koo, S. P.; Booth, I. R. Microbiology 1994, 140, 617.
6. Kempf, B.; Bremer, E. Arch. Microbiol. 1998, 170, 319.
7. Chambers, S. T.; Lever, M. Nephron 1996, 74, 1.
both radiolabelled betaines quite similar to that ob-
served in E. coli. Interestingly, all the strains sensitive to
the treatment must possess at least one transport system
equivalent to those of E. coli.
Together, our data demonstrate a powerful inhibitory
effect of glycine betaine analogues against several bac-
terial strains belonging to diverse taxonomic groups.
The toxic activity is manifest mainly for compounds 1
and 2. The former displays a bacteriostatic effect
whereas compound 2 seems lethal, probably because of
the release of a nitrobenzyl aldehyde resulting from its
metabolism. Efficient specific transport systems are re-
quired for importation of compounds 1 and 2; they are
strikingly activated by elevation of medium osmolality
and do allow a high internal accumulation of the toxic
analogues, as they do for glycine betaine.
ꢀ
8. Cosquer, A.; Pichereau, V.; Le Mee, D.; Le Roch, M.;
Renault, J.; Carboni, B.; Uriac, P.; Bernard, T. Bioorg.
Med. Chem. Lett. 1999, 9, 4954.
9. Queau, M.; Renault, J.; Uriac, P.; Corbel, J. C.; Travert,
M. F.; Bernard, T.; Donnio, P. Y. Pathol. Biol. 2003, 51,
516.
10. Lloyd, A. W.; Smith, G.; Oliff, C. J.; Rutt, K. J. J. Pharm.
Pharmacol. 1992, 44, 507.
ꢀ
11. Takeuchi, M.; Mizuno, T.; Shinmori, U.; Nakashima, M.;
Shinkai, S. Tetrahedron 1996, 52, 1195.
12. Structural data for compound 1. HRMS (LSIMS)
[M + H]þ C11H16NO2 calcd 194.1181 found: 194.118
negative mode [M + Cl]ꢀ: 228. 1H NMR (500 MHz,
D2O) 3.26 (s, 6H), 4.03 (s, 2H), 4.75 (s, 2H), 7.55 (m,
5H). 13C NMR (125 MHz, D2O) 51.36; 61.08; 68.48;
127.16; 129.69; 131.38; 133.21; 167.70.
13. Electrophoresis and chromatography. High voltage paper
electrophoresis: a sheet of Whatman 3 MM paper,
moistened with 3% formic acid was submitted to an
electric field value of 30 V/cm. Extract subsamples were
spotted at the anode side and migration was maintained
for about 60 min.
Indeed, such efficient systems could be a useful means to
promote penetration of drugs into recalcitrant bacteria.
Curiously, very few reports dealing with successful
attempts using betaine analogues are available so far.18
Unfortunately, the mechanism of inhibition exerted by
betaine analogues still remains unknown; preparation of
resistant mutants (Tn5 insertion) would help to deter-
mine the actual target of these inhibitors. In addition,
synthesis of new molecules, structurally related to gly-
cine betaine or other osmoprotectants, should be
encouraged.
2-D TLC on silica gel 60 F254 Merck using as solvent 1, n-
butanol–acetic acid–water (12:3:5 v/v) and as solvent
2, phenol (80% in water)–ethanol–ammonia (50:50:0.5
v/v).
14. Typical experimental procedure. Anchorage of dimethyl-
glycine: dimethylglycine (0.67 g; 6.5 mmol) and HOBT
(0.77 g; 6.5 mmol) were stirred for 5 min in anhydrous
DMF (5 mL). Then DCI (0.82 g; 6.5 mmol) was added.
This solution, after stirring for 1 h at 0 °C and 1 h at rt, was
added to a suspension of Wang resin (1 g; 0.65 mmol) in
anhydrous DMF (10 mL) in a 25 mL syringe. Washing was
achieved as following: DMF, THF, THF/water (50/50),
water, THF, anhydrous ether and then the resin was dried
under reduced pressure.
Acknowledgements
We thank M. Villarejo and E. Bremer for providing
E. coli and B. subtilis strains.
Alkylation: to a suspension of a resin in anhydrous DMF
(10 mL) 5 equiv (3.25 mmol) of alkylating agent in 2 mL of
DMF was added. After stirring at rt for 24 h the same
washing procedure was used.
Cleavage: Wang resin (1 g) was allowed to cleave in 6 mL
of a 50:50 mixture of TFA and CH2Cl2. After evaporation
of the solvent, the crude residue was stirred for 30 min in a
mixture of 1.5 mL of water and 1.5 mL of CH2Cl2. The
References and notes
1. Csonka, L. N.; Hanson, A. D. Annu. Rev. Microbiol. 1991,
45, 569.
2. Booth, I. R.; Higgins, C. F. FEMS Microbiol. Rev. 1990,
6, 239.
3. Galinski, E. A. Adv. Microbiol. Physiol. 1995, 37, 272.

A. Cosquer et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2061–2065
2065
aqueous layer was washed twice using 2 mL of CH2Cl2
and then evaporated. The solid residue was dried under
reduced pressure.
mented with the radioactive substrate. 14CO2 was trapped
in a concentrated solution of KOH (5 M); the whole
soluble metabolites were extracted with 80% ethanol and
the insoluble material was gathered in the pellet after
centrifugation. The soluble fraction was analysed on a 2D
thin-layer, as in Ref. 13. Detection of the radiolabelled
compounds was performed using the InstantImager Pack-
ard (electronic autoradiography).
15. (a) Ziegler, E.; Wittmann, H. Monatsh. Chem. 1985, 821;
(b) Du Pont de Nemours & Co, U.S. Patent 2359863,
1942.
16. Crist, D. R.; Jordan, G. J.; Moore, D. W.; Hashmall, J.
A.; Borsetti, A. P.; Turujman, S. A. J. Am. Chem. Soc.
1983, 105, 4136.
17. Radiolabelling assays: typically, a diluted suspension
(OD ¼ 1) in either salted or unsalted medium was supple-
18. Peddie, B. A.; Wood, J. E.; Lever, M.; Happer, D. A.; De
Zwart, F.; Chambers, S. T. Antonie van Leeuwenhoek
2003, 83, 175.