Design of antimicrobial compounds based on peptide structures

Article abstract of DOI:10.1016/j.bmcl.2007.01.075

New antimicrobial compounds are of major importance because of the growing problem of bacterial resistance and antimicrobial peptides have been gaining a lot of interest. Their mechanism of action, however, is often obscure. Here a set of non-peptidic compounds with antimicrobial activity are presented that have been designed based on criteria derived from three-dimensional structures of antimicrobial peptides. Even though only a small set of compounds has been designed, the activity immediately matches that of the original peptides, supporting the proposed criteria for activity, i.e. not the peptidic nature of antimicrobial peptides is responsible for their activity but rather the proper arrangement of the relevant functional groups.

Full text of DOI:10.1016/j.bmcl.2007.01.075

Bioorganic & Medicinal Chemistry Letters 17 (2007) 2334–2337
Design of antimicrobial compounds based on peptide structures
Christian Appelt, Anna K. Schrey, J. Arvid Soderhall and Peter Schmieder*
¨
¨
Leibnizinstitut fu¨r Molekulare Pharmakologie (FMP), Robert-Ro¨ssle-Str. 10, D-13125 Berlin, Germany
Received 20 December 2006; revised 16 January 2007; accepted 17 January 2007
Available online 27 January 2007
Abstract—New antimicrobial compounds are of major importance because of the growing problem of bacterial resistance and anti-
microbial peptides have been gaining a lot of interest. Their mechanism of action, however, is often obscure. Here a set of non-pep-
tidic compounds with antimicrobial activity are presented that have been designed based on criteria derived from three-dimensional
structures of antimicrobial peptides. Even though only a small set of compounds has been designed, the activity immediately match-
es that of the original peptides, supporting the proposed criteria for activity, i.e. not the peptidic nature of antimicrobial peptides is
responsible for their activity but rather the proper arrangement of the relevant functional groups.
Ó 2007 Elsevier Ltd. All rights reserved.
Bacterial resistance to existing drugs is a constantly grow-
ing problem that, combined with a decline in the develop-
ment of new antibiotics, presents a significant threat to
human health.1–3 The identification of new antimicrobial
agents is therefore of considerable importance. In recent
years, antibacterial and antifungal peptides have gained
a lot of interest, due to their potential use as a new gener-
ation of therapeutic agents.4,5 These peptides exhibit
activity against a broad spectrum of microbes, albeit with
fairly low activity. However, resistance against them has
rarely been reported, even though they are evolutionary
ancient weapons of higher animals.4 Unfortunately, their
mechanism of action is still not clear and its elucidation
would form a sound basis for the further development
of pharmaceutical compounds.
the guanidine groups form contacts to charged lipid
head groups, and the backbone faces the outside of
the membrane. When tryptophan is replaced by tyrosine
or arginine is replaced by lysine the structure of the
peptide does not change, the backbone forms similar
b-turns positioning the amino acid side chains in
similar directions. The activity, however, is changed
considerably.9 On the other hand, scrambling of the
original sequence in a way that the three aromatic side
chains are next to each other (cyclo(RRWWFR) and
cyclo(RRWFWR)) changes the backbone structure
but does not affect the activity of the peptides [to be
published]. From those findings we conclude that the
peptide backbone merely presents the scaffold for the
orientation of the side chains of the amino acids and
that the antimicrobial activity requires a sufficient num-
ber of indole rings and guanidinium groups. It should
then be possible to design an antimicrobial compound
with similar activity as the original peptide by using a
simpler scaffold that is capable of positioning the amino
acids in a proper manner.
In two recent papers we have determined the structure
of the antimicrobial peptide cyclo-(Arg-Arg-Trp-Trp-
Arg-Phe) and several analogues using solution NMR
spectroscopy and have described their potential interac-
tions with a biological membrane using extensive molec-
ular dynamic simulations.6–8 In a membrane mimicking
environment, the peptide exhibits an amphipathic struc-
ture that differs considerably from the one found in
aqueous solution. The hydrophobic part is formed by
the aromatic side chains while the hydrophilic part is
made up of the peptide backbone. The aromatic side
chains protrude into the lipid chains of the membrane,
To test this hypothesis, we have chosen to synthesize a
small set of compounds based on trimesic acid as the
scaffold with indole rings and guanidinium groups at-
tached (Scheme 1). The structural rationale behind the
design of the compounds is shown in Figure 1 where
the structure of the peptide bound to detergent micelles
as determined by NMR is overlaid with a simple model
of the new compound. As can be seen the trimesic acid is
capable of positioning the relevant functional groups in
a way comparable to that in the peptide. To assess the
Keywords: Antimicrobial peptides; Synthesis; Design.
*
Corresponding author. Tel.: +49 30 94793 227; fax: +49 30 94793
230; e-mail: schmieder@fmp-berlin.de
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.01.075

C. Appelt et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2334–2337
2335
Scheme 1. Structure of the non-peptidic analogs of antimicrobial peptides together with the structure of cyclo(RRWWRF) and the ‘side chain’
compounds CAB5 and CAB6.
importance of the guanidinium groups, we synthesized
and investigated analoges carrying amino instead of
guanidinium groups. To allow for variation in the
positioning of the aromatic rings and the charged
groups relative to each other, two different chain lengths
were used for the attachment of the charged groups. The
synthesis of the four compounds is exemplified in
Scheme 2 by the synthesis of CAB1 and CAB2 contain-
ing ethane linkers. CAB3 and CAB4 contained pentane
chains and were synthesized in a similar way.
The biological activity of the new compounds was tested
on two bacterial strains, Escherichia coli and Bacillus
subtilis, as representatives of gramnegative and gram-
positive bacteria, respectively. CAB4 possesses three
guanidinium moieties attached to pentane linkers. It
turned out to be the most potent compound, having
minimal inhibitory concentrations in the low micromo-
lar range against both E. coli and B. subtilis (Table 1).
This activity is similar to the one of the cyclo-(Arg-
Arg-Trp-Trp-Arg-Phe) peptide. The compound CAB3
is the corresponding triamine. It is less effective in inhib-
iting bacterial growth and is therefore comparable to the
peptide cyclo-(Lys-Lys-Trp-Trp-Lys-Phe).
Figure 1. Comparison of a model of CAB4 (thick) and the NMR
structure of cyclo(RRWWRF) (thin)6. The model was created by
In order to check whether fragments of these compo-
orienting the ‘side chains’ in the same way as in the peptide followed by
nents display any antimicrobial activity, CAB5, CAB6,
and trimesic acid were also tested (Table 1). They were
an energy minimization. This shows that the indoles (green) can adopt
a similar orientation as the aromatic side chains of the peptide.
completely inactive demonstrating that the linkage of
three tryptophans with guanidines or amines is essential
for the activity.
Likewise, the guanidine moieties can be oriented in the same way as the
arginine side chains. A structure determination of CAB4 bound to the
micelle has so far been precluded by the symmetry of the compound.

2336
C. Appelt et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2334–2337
Scheme 2. Synthesis of CAB1 and CAB2. TMACl, trimesic acid chloride. PCA, pyrazole carboxamidine.11
Table 1. Antimicrobial activities and erythrocyte lysis of the developed compounds
R
n
MIC E. coli (lM)
MIC B. subtilis (lM)
Erythrocyte lysis at 100lM (%)
CAB1
CAB2
CAB3
Amino
Guanidino
2
2
5
5
–
–
–
–
–
>125
>125
62.5
15.6
6.3
31.3
3.9
1.2
5.1
0.4
22.6
24
Amino
Guanidino
31.3
2.0
CAB4
cyclo(RRWWRF)a
cyclo(KKWWKF)a
CAB5
–
–
–
–
–
3.1
25
25
10
>250
>250
>250
>250
>250
>250
<1
<1
<1
CAB6
Trimesic acid
a Reproduced with permission from.8 CAB5, (S)-2-amino-N-(2-aminoethyl)-3-(1H-indol-3-yl)propanamide; CAB6, (S)-2-amino-N-(5-aminopentyl)-
3-(1H-indol-3-yl)-propanamide.
A different behavior was seen for CAB1 and CAB2
which have their amino and guanindino moieties
coupled to an ethyl linker. Like CAB3 and CAB4
they were active against B. subtilis. However, they were
inactive against E. coli. This indicates that the distance
between the charged groups and the aromatic rings has
an impact on the activity but given the small set of com-
pounds tested it is not possible to explain the observed
selectivity. Expanding the set of compounds to overcome
these limitations is in progress. Additionally, the symme-
try of the compounds has so far precluded the determina-
tion of their micelle-bound structures. It was, however,
possible to obtain information on the orientation of the
peptides using methods described previously.6 The
compounds orient parallel to the micelle surface with
the aromatic rings pointing into the micelle and the

C. Appelt et al. / Bioorg. Med. Chem. Lett. 17 (2007) 2334–2337
2337
guanidinium groups on the surface. Together with the
References and notes
size of the compounds this is a clear indication that no
channel formation is taking place and that the
compounds rather act according to the carpet model.10
1. Nathan, C. Nature 2004, 431, 899.
2. Wenzel, R. P. N. Engl. J. Med. 2004, 351, 523.
3. Projan, S. J. Curr. Opin. Microbiol. 2003, 6, 427.
4. Zasloff, M. Nature 2002, 415, 389.
5. Hancock, R. E.; Lehrer, R. Trends Biotechnol. 1998,
16, 82.
6. Appelt, C.; Wessolowski, A.; So¨derha¨ll, J. A.;
Dathe, M.; Schmieder, P. Chembiochem 2005, 6,
1654.
In conclusion, we have been able to design non-peptidic
compounds with antimicrobial activity based on princi-
ples derived from the structure of antimicrobial peptides
determined using solution NMR-spectroscopy. Even
though the design was simple, the activities of the new
compounds immediately match those of the original
peptides.
7. Appelt, C.; Eisenmenger, F.; Kuhne, R.; Schmieder, P.;
¨
So¨derha¨ll, J. A. Biophys. J. 2005, 89, 2296.
8. PDB-entries: 1QVK (cyc-(RRWWRF)), 1SKI (cyc-
(RRYYRF)), 1SKK (cyc-(KKWWKF)), 1SKL (cyc-
(RRNalNalRF)).
9. Dathe, M.; Nikolenko, H.; Klose, J.; Bienert, M. Bio-
chemistry 2004, 43, 9140.
Supplementary data
Supplementary data associated with this article can
be found, in the online version, at doi:10.1016/j.bmcl.
2007.01.075.
10. Shai, Y. Biochim. Biophys. Acta 1999, 1462, 55.
11. Bernatowicz, M. S.; Wu, Y. L.; Matsueda, G. R. J. Org.
Chem. 1992, 57, 2497.