J. C. Slootweg et al. / Bioorg. Med. Chem. 22 (2014) 5345–5353
5347
with nisin hybrid 6. The sharp contrast between antimicrobial
activity and absence of pore-formation is explained by the dual
mode of action of nisin, since nisin analogs without pore-formation
activity are still bacteriostatic agents caused by lipid II binding and
therefore capable of inhibiting the bacterial cell-wall synthesis.28
Another lantibiotic, and a close family-member of nisin, subtilin,
has a high structural homology as well as similar biological activity
2
9
compared to nisin. Parisot et al. showed that a truncated analog,
subtilin (1–29), containing only Lys29 at its C-terminal sequence,
displayed both antimicrobial and pore-formation activity.30 There-
fore, it was decided to extend the synthetic DE-ring with a C-termi-
nal lysine residue. The detailed synthesis of the extended dicarba
1
7
DE-ring 12 (vide infra) was described previously, and is shown
in Scheme 2.
Another modification that was introduced into nisin hybrid
mimic 6 was the triazole containing moiety, which replaced the
native peptide bond between Met21–Lys22. Since propargylamine
was used as the amino alkyne, leading to a glycine-derived isoste-
re, a possibly important hydrophobic side chain, as present in the
Met21 residue in native nisin, was also removed. However, this
hydrophobic residue is conserved among the six known natural
nisin variants and is present either in methionine (nisin A, Z and
F) or in leucine (nisin Q, U and U2). Furthermore, structure-activ-
ity-relationship studies of nisin underlined the importance of this
hydrophobic residue since a nisin analog with a methione21 to gly-
cine21 (Met21 ? Gly21) mutation had no pore-formation activ-
Figure 1. Nisin-induced leakage (at 1 nM) in a competitive assay after incubation
with different concentrations (in M) of nisin (1–20) fragment 2 (red trace) and
l
nisin hybrid 6 (blue trace). Unilamellar vesicles were loaded with carboxyfluores-
cein as the fluorophore, while the vesicle was spiked with lipid II as the target of
nisin. In the absence of peptide 2 and 6, nisin binds lipid II and induces membrane
leakage by pore formation which is set to ꢀ80% compared to Triton X-100 as a
unspecific membrane-disrupting detergent which leads to 100% leakage. Addition
of peptide 2 or 6, prior to the addition of nisin, will lead to a competition for the
lipid II binding site resulting in a lower leakage % depending on the affinity of both
peptides for lipid II. Thus, peptide 2 displays approximately four times higher
affinity toward lipid II than peptide 6 (0.2 vs 0.8 lM, respectively).
3
1
ity. These data suggested that it was crucial to synthesize nisin
analogs which included a hydrophobic residue on position 21 to
retain pore-formation activity. In this stage of the design process,
it was decided to focus on the chemically stable leucine residue,
since the methionine thioether is prone to oxidation and might
negatively interfere with pore formation. For this purpose, the leu-
cine-derived amino alkyne was synthesized via conversion of the
corresponding amino aldehyde using the Bestmann–Ohira reagent
chemo-selectively conjugated via Cu(I)-catalyzed click chemistry
in the presence of CuSO /sodium ascorbate under microwave irra-
diation in DMF/H O as the solvent system. Gratifyingly, the liga-
tion proceeded very well and after preparative HPLC purification,
nisin mimic 6 could be obtained in 56% yield.
4
26
2
The bioactivity of this nisin hybrid was tested in a model mem-
1
8
32,33
brane experiment, to determine the ability of this nisin deriva-
tive to form pores like native nisin. For this purpose, large
unilamellar vesicles (LUVs) composed of 1,2-dioleoyl-sn-glycero-
7, which was synthesized according to a literature procedure
(see Supplementary data, Scheme SI-1).
Boc-protected leucine 8 was converted to its corresponding
Weinreb amide 9 via a BOP-mediated coupling with N,O-dim-
ethylhydroxylamine, in excellent yield, as shown in Scheme 3. Sub-
sequently, the Weinreb amide was reduced to the corresponding
Boc-protected amino-aldehyde in the presence of DIBAL-H, imme-
diately followed by a reaction with the Bestmann–Ohira reagent 7,
affording leucine-derived alkyne 10 in a modest yield of 40%.
The synthesis of the optimized nisin hybrid 13, as shown in
Scheme 4, started with a TFA treatment to remove the Boc group
of amino alkyne 10, and after a simple workup, the amine was used
in the BOP-mediated coupling step. Via a similar procedure as
described to obtain alkyne 3, the leucine-derived amino alkyne
was coupled to nisin-fragment (1–20) 2 using an excess of 25 equiv
to avoid a presumably difficult purification of 11. This coupling
reaction proceeded rather slowly compared to propargylamine
and only trace amounts of alkyne 11 were observed after a reaction
time of 15 min. Increasing the reaction time to 30 min did improve
product formation and alkyne 11 could be isolated in an overall
yield of 15%. However, also the formation of a side product was
observed. This side product corresponded to a molecule with a
mass 18 amu lower than alkyne 11 (HPLC and ESI-MS, see Support-
ing information section). This side product could not explained by
3
-phosphocholine (DOPC) and containing 0.2% of lipid II as the nat-
ural target of nisin were loaded with carboxyfluorescein (CF) as a
fluorophore, and peptide-membrane interaction was measured
by monitoring the release of CF by fluorescence spectroscopy. It
turned out that mimic 6 did not show any pore-formation activity
up to 1000-fold higher concentration than native nisin (data not
shown). However, mimic 6 was still able to bind lipid II, as shown
in a competitive binding assay (as shown in Fig. 1) since it was able
to compete with native nisin for the binding-site of lipid II. As a
control, also nisin-fragment (1–20) 2 was tested, which is known
to compete with native nisin in this experimental set-up. It was
observed that mimic 6 was competitive to native nisin comparable
to 2 (Fig. 1), an indication that although mimic 6 had no pore-for-
mation activity, the lipid II binding activity of the N-terminal part
of mimic 6 was retained. Thus, although the synthesis of a nisin
hybrid 6 was successfully performed, the C-terminal part (com-
prising the DE-rings) clearly needed to be optimized in order to
display pore-formation activity.
1
4
2
.2. Optimization of the nisin hybrid
The synthetic DE fragment as used in nisin hybrid 6 represents
assuming lactam formation between the N-terminal
the -amine of lysine12 and the C-terminal carboxylate. A more
reasonable explanation would be a dehydration of the aspara-
gine20 side chain. It is known from the literature that carbodiim-
ide-based coupling reactions with C-terminally unprotected
asparagine residues result in dehydration of the amide side chain
to the corresponding cyano moiety. It was expected that BOP-med-
iated couplings were devoid of this side reaction since the highly
a-amine or
nisin-fragment (22–28), while the sequence of native nisin is seven
residues longer: ꢀSer-Ile-His-Val-Dha-Lys-OH (nisin 29–34). Trun-
cation studies of nisin showed that C-terminally truncated variants
were slightly less active as an antimicrobial agent.24a,27 However,
some truncations led to a complete loss of pore-formation activity
as was recently described for nisin-fragment (1–28).27 This insight
might explain the lack of pore-formation activity as was observed
e
3
4