Design and Synthesis of Potent Sensitizers of Gram-Negative Bacteria Based on a Cholic Acid Scaffolding

Full text of DOI:10.1021/ja973881r

J. Am. Chem. Soc. 1998, 120, 2961-2962
Design and Synthesis of Potent Sensitizers of
2961
Gram-Negative Bacteria Based on a Cholic Acid
Scaffolding
Chunhong Li,† Adam S. Peters,† Erik L. Meredith,†
Glenn W. Allman,‡ and Paul B. Savage*,†
Departments of Chemistry and Biochemistry
and Microbiology
Brigham Young UniVersity, ProVo, Utah 84602
ReceiVed NoVember 12, 1997
Figure 1. Polymyxin B2 (1), polymyxin B nonapeptide (2), and
polymyxin B heptapeptide (3).
The outer membrane of Gram-negative bacteria provides a
protective barrier against proteases, lysozymes, and many types
of antibiotics.1 Consequently, numerous antibiotics that are active
against Gram-positive bacteria are much less active against Gram-
negative strains. Lipid A is a primary component of the outer
membrane of Gram-negative bacteria and plays an essential role
in cell wall integrity. Compounds known to associate strongly
with lipid A disrupt the organization of the outer cell wall and
thereby sensitize Gram-negative bacteria to antibiotics.2 These
compounds have generally been derivatives of polymyxin B (1,
Figure 1), the most common member of a family of related peptide
antibiotics. Derivatives of polymyxin B such as polymyxin B
nonapeptide (2, Figure 1), and polymyxin B heptapeptide (3,
Figure 1) sensitize bacteria to antibiotics without causing toxic
effects that limit the use of polymyxin B.3,4 However, compounds
such as 2 and 3 are difficult to prepare and purify.2c,i We have
prepared simple, nonpeptide mimics of polymyxin B that act as
potent sensitizers of Gram-negative bacteria to antibiotics.
Design of the new sensitizers was based upon modeling of the
lipid A binding domain of polymyxin B and consideration of the
conserved residues found in polymyxin B and related antibiotics.
Through the pioneering work of Vaara and co-workers,2c the lipid
A binding domain of polymyxin B has been identified as the
macrocyclic portion of the molecule (3). Molecular modeling5
of 3 with constraints derived from reported NOESY experiments6
and predicted peptide turn formation7 provided a low-energy
structure in which the three amine groups derived from diami-
nobutyric acid residues are oriented on one face of the molecule
(Figure 2). Since these diaminobutyric acid groups are conserved
among the related antibiotics polymyxins A, B1, B2, D1, E1, and
E2, circulin A, and octapeptins A1, A2, A3, B1, B2, B3, and C1, we
included three primary amines in the design of our sensitizers.
The fact that simple polyamines and linear versions of 3 do not
sensitize Gram-negative bacteria to antibiotics8 suggests that a
Figure 2. Conformations of 3 and 4 predicted using molecular mechanics
(MM3 parameters using the program Spartan). Primary amine groups
are indicated. Hydrogens and the phenyl group of 4 have been omitted
for clarity.
specific arrangement of amine groups is required for sensitization
activity. Our modeling demonstrated that appropriate function-
alization of cholic acid results in orientation of three amine groups
in a conformation comparable to that predicted for 3 (Figure 2).
Our hypothesis was that if association of 3 with lipid A were
mediated by the diaminobutyric acid side chains, then the cholic
acid derivatives would bind to lipid A and sensitize bacteria to
antibiotics.
Preparation of the cholic acid derivatives is shown in Scheme
1. Tethers of either two or three carbons were used between the
steroid and the amine groups. Also, because guanidine groups
were expected to interact strongly with phosphates on lipid A,9 4
and 5 were converted to the tris(guanidine) analogues 6 and 7.10
The ability of 4-7 to sensitize bacteria to antibiotics and to
inhibit the growth of Gram-negative bacteria was assayed by
measuring minimum inhibition concentration (MIC) values.11 To
distinguish between bacteristatic and bactericidal activity, we also
measured minimum bactericidal concentration (MBC) values.12
Prior to measuring sensitizing activity of 4-7, we determined
the effects of these compounds alone on bacterial growth.
Unexpectedly, 4-7 exhibited bacteristatic and bactericidal activ-
ity. We first measured MIC values of 4-7 with Escherichia coli
strain ATCC 10798, and the results are shown in Figure 3. The
† Department of Chemistry and Biochemistry.
‡ Department of Microbiology.
(1) (a) Labischinski, G.; Barnickle, G.; Bradaczek, H.; Naumann, D.;
Rietschel, D. T.; Giesbrecht, P. J. Bacteriol. 1985, 162, 9. (b) Nikaido, H.;
Vaara, M. Microbiol. ReV. 1985, 49, 1. (c) Hancock, R. E. W. Annu. ReV.
Microbiol. 1984, 38, 237.
(2) (a) Morris, C. M.; George, A.; Wilson, W. W.; Champlin, F. R. J.
Antibiot. 1995, 48, 67. (b) Ofek, I.; Cohen, S.; Rahmani, R.; Kabha, K.;
Tamarkin, D.; Herzig, Y.; Rubenstein, E. Antimicrob. Agents Chemother. 1994,
38, 374. (c) Vaara, M. Antimicrob. Agents Chemother. 1993, 37, 354. (d)
Kimura, Y.; Matsunaga, H.; Vaara, M. J. Antibiot. 1992, 45, 742. (e) Vaara,
M. Drugs Exp. Clin. Res. 1991, 17, 437. (f) Peterson, A. A.; Hancock, R. E.
W.; McGroarty, E. J. J. Bacteriol. 1985, 164, 1256. (g) Vaara, M.; Vaara, T.
Antimicrob. Agents Chemother. 1983, 24, 107. (h) Vaara, M.; Vaara, T.
Antimicrob. Agents Chemother. 1983, 24, 114. (i) Vaara, M.; Vaara, T. Nature
1983, 303, 526.
(8) Vaara, M. Drugs Exp. Clin. Res. 1991, 17, 437.
(9) Hsieh, H.-P.; Muller, J. G.; Burrows, C. J. J. Am. Chem. Soc. 1994,
116, 12077.
(10) Kim, K.; Lin, Y.-T.; Mosher, H. S. Tetrahedron Lett. 1988, 29, 3183.
(11) An MIC measurement consists of incubating a known concentration
of bacteria for 24 h in a nutrient broth containing incrementally varied
concentrations of the compound of interest followed by determination of
bacterial growth via cell counting and turbidity measurements. The MIC value
is the concentration of the studied compound at which the number of bacteria
remains constant or decreases during incubation. (For example, see: Pitt, W.
G.; McBride, M. O.; Lunceford, J. K.; Roper, R. J.; Sagers, R. D. Antimicrob.
Agent Chemother. 1994, 38, 2577.) In our experiments, each MIC determi-
nation was repeated three times with MIC values varying by less than 10%.
(12) An MBC value is the concentration at which fewer than 0.1% of
bacteria survive incubation with the compound of interest.
(3) For example, see: Vinnicombe, J.; Stamey, T. A. InVest. Urol. 1969,
6, 505.
(4) Conte, J. E., Jr. Manual of Antibiotics and Infectious Diseases; Williams
& Wilkins: Philadelphia, PA, 1995.
(5) Using MM3 parameters with the program Spartan.
(6) Bhattacharjya, S.; David, S. A.; Mathan, V. I.; Balaram, P. Biopolymers
1997, 41, 251.
(7) Perkins, S. J.; Radda, G. K.; Richards, R. E. Eur. J. Biochem. 1978,
82, 551.
S0002-7863(97)03881-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 03/12/1998

2962 J. Am. Chem. Soc., Vol. 120, No. 12, 1998
Communications to the Editor
Scheme 1
with erythromycin. With 4, the MIC of erythromycin was
lowered by a factor of 70 and the MIC of 4 by a factor of 10. To
compare 4-7 with known bacterial sensitizers, we used the same
conditions to measure synergism with 2, a compound described
as a potent sensitizer of Gram-negative bacteria.2 The amount
required to lower the MIC of erythromycin to 1 µg/mL was over
50 µg/mL.
To demonstrate that the activity of the cholic acid derivatives
was not antibiotic dependent, we repeated the sensitization
experiments with novobiocin, an antibiotic unrelated to erythro-
mycin. The MIC of novobiocin with E. coli (ATCC 10798) is
>500 µg/mL. Concentrations of 4 µg/mL of 4 or 2 µg/mL of 5
lowered the MIC of novobiocin to 1 µg/mL.
To verify that the effects of the cholic acid derivatives were
not strain or species dependent, we measured MIC and sensitiza-
tion data with an additional strain of E. coli and with a strain of
Pseudomonas aeruginosa. The MIC values and sensitization
properties of 4-7 with E. coli (ATCC 25922) proved to be very
similar to those measured with E. coli (ATCC 10798). P.
aeruginosa strains are typically more resistant to antibiotics than
E. coli,4 and the MIC values of erythromycin, 4, and 5 with P.
aeruginosa (ATCC 27853) were 240, 45, and 25 µg/mL,
respectively. The sensitization properties of our cholic acid
derivatives also decreased, although the trends remained the same
(e.g., 12 µg/mL of 4 or 7 µg/mL of 5 lowered the MIC of
erythromycin to 5 µg/mL). These results demonstrate that the
cholic acid derivatives are capable of sensitizing multiple strains
and species of Gram-negative bacteria to antibiotics.
To probe the requirements for a spatial arrangement of amine
groups in bacterial sensitization, we prepared the C-3-epimer of
4 by inverting14 the alcohol group at C-3 on cholic acid and
following the procedures outlined in Scheme 1 to give 11. With
11, we found a significant rise in the MIC (as compared to 4)
and a decrease in sensitization properties of this compound (Figure
3). These results suggest the importance of a specific arrangement
of amines in the cholic acid derivatives.
Bactericidal and bacteristatic activity of polymyxin B (1) can
be eliminated while preserving sensitization properties by removal
of the fatty acid and any number of the exocyclic amino acids
from the antibiotic giving compounds such as 2 and 3.2 We
hypothesized that there might be an equivalent effect from similar
variations of the group extending from C-24 in the cholic acid
derivatives. To address this issue, we prepared 12 (compound 5
with R′ ) OH) and 13 (compound 5 with R′ ) -O(CH2)7CH3).
We expected that if the group at C-24 were involved in
bacteristatic activity, the MIC of 12 would increase compared to
that of 5 and that the MIC of 13 would decrease. Indeed, these
effects were observed (Figure 3). However, even more striking
is that the abilities of the two compounds to sensitize bacteria
were nearly indistinguishable (Figure 3). These data suggest that
sensitization properties can be separated from bacteristatic activity
in a fashion similar to that observed with polymyxin B and its
derivatives.
Figure 3. Hatched bars: MIC values for compounds 4-7 and 11-13
with E. coli (ATCC 10798). Solid bars: concentrations of compounds 2,
4-7, and 11-13 required to lower the MIC of erythromycin from 70 to
1 µg/mL with E. coli (ATCC 10798).
data suggest that longer tethers between the steroid backbone and
the primary amines increase activity as well as replacement of
amines with guanidine groups. MBC values were also measured
for 4-7 and were found to be approximately 1.5 times as high
as the MIC values.
We used erythromycin to measure the ability of 4-7 to
sensitize bacteria to other antibiotics. Erythromycin is not as
active against Gram-negative bacteria as it is against Gram-
positive strains due to the inability of the large, partially
hydrophobic antibiotic to traverse the outer membrane of the
former.1 The MIC of erythromycin with E. coli (ATCC 10798)
is 70 µg/mL, and our bench mark was to lower the MIC of
erythromycin to 1 µg/mL.13 Cultures containing 1 µg/mL of
erythromycin were incubated with incrementally varied concen-
trations of 4-7. The results are shown in Figure 3. All four
compounds displayed a synergistic inhibition of bacterial growth
We have demonstrated that appropriate arrangements of
primary amines or guanidine groups on a steroid scaffolding can
produce potent sensitizers of Gram-negative bacteria. These
sensitizers may allow use of antibiotics against Gram-negative
bacteria that have traditionally been used primarily against Gram-
positive bacterial infections.
Acknowledgment. We thank Brigham Young University for financial
support, Cynthia Burrows and Morris Robins for helpful suggestions,
and William Pitt for help initiating MIC measurements.
Supporting Information Available: Experimental data including
synthetic details for compounds 4-7 and 11-13 and representative
procedures for MIC and MBC measurements (30 pages, print/PDF). See
any current masthead page for ordering information and Web access
instructions.
JA973881R
(14) Mitsunobu, O. Synthesis 1981, 1.
(13) Many clinically useful antibiotics against Gram-negative bacteria have
MIC values of ca. 1 µg/mL (see ref 4).