Albumin binding of short cationic antimicrobial micropeptides and its influence on the in vitro bactericidal effect

Article abstract of DOI:10.1021/jm0703542

The interactions between a range of small cationic antibacterial tripeptides and bovine and human serum albumin in a buffered aqueous solution at 25°C have been studied using isothermal titration calorimetry. Results from the binding study indicate a sing

Full text of DOI:10.1021/jm0703542

3334
J. Med. Chem. 2007, 50, 3334-3339
Albumin Binding of Short Cationic Antimicrobial Micropeptides and Its Influence on the in
Vitro Bactericidal Effect
Johan Svenson,*,† Bjørn-Olav Brandsdal,‡ Wenche Stensen,†,§ and John S. Svendsen†,§
Department of Chemistry, UniVersity of Tromsø, N-9037 Tromsø, Norway, The Norwegian Structural Biology Centre, Department of Chemistry,
UniVersity of Tromsø, N-9037 Tromsø, Norway, and Lytix Biopharma AS, P.O. Box 6447, Tromsø Science Park, N-9294 Tromsø, Norway
ReceiVed March 27, 2007
The interactions between a range of small cationic antibacterial tripeptides and bovine and human serum
albumin in a buffered aqueous solution at 25 °C have been studied using isothermal titration calorimetry.
Results from the binding study indicate a single binding site on albumin with a dissociation constant between
4.3 and 22.2 µM for the different peptides. In a theoretical mouse model, a dissociation constant in this
range corresponds to 95% albumin binding. The effect of this albumin interaction on the antibacterial capacity
of the peptides against Staphylococcus aureus, strain ATCC 25923 was studied by including albumin in the
assays at a 0.55 mM concentration. Presence of albumin induced a 10-fold increase of the minimal inhibitory
concentration for the bulk of the peptides. Albumin itself has no effect on the bacterial growth and this
increase is entirely ascribed to a strong competing protein binding. Collectively these results indicate that
these antibacterial peptides do bind to albumin and that this binding strongly reduces the effective
concentration of peptides available to combat bacteria.
Introduction
in a series of dipeptides displaying adequate lipophilic bulk and
cationic charge required by the pharmacophore. Generally, two
units of bulk, each similar to a phenyl group in size and polarity
and two cationic charges are sufficient, even though there exist
differences among the motifs between different target bacteria.8
Incorporation of bulky, nongenetically coded amino acids can
in some instances be beneficial for the activity while simulta-
neously allowing for a wider range of possible structural
modifications.9 Initial studies on the ability of CAPs to destroy
bacteria (both antibiotic sensitive and multi resistant strains) in
Vitro have been most promising and suggest further efforts.
CAPs are believed to exert their antibacterial action by interact-
ing with the negatively charged bacterial surface.11 Once close
enough, the CAPs are able to insert their hydrophobic section
into the cytoplasmic membrane, resulting in permeabilization,
leakage, and cell death.12 The specific mechanisms behind the
cell death is however both bacteria and peptide dependent.
Strong binding to plasma proteins like albumin could decrease
the effective concentration of CAPs at the bacterial surface, and
therefore it is of interest to evaluate what effect possible protein
binding would have on the bactericidal potency. In this paper
we describe the use of isothermal titration calorimetry (ITC) to
determine the binding of a range of CAPs to albumin (bovine
and human) and the influence of this binding on the antibacterial
activity of the peptides. The SAR of the CAP library with regard
to antibacterial activity, as well as protein binding, is also
discussed.
Albumin, a 66 kDa monomer forming three helical domains,
is the most abundant protein in plasma. The protein is present
in humans at around 0.6 mM, and it plays an important role as
it binds and transports a range of endogenous molecules such
as fatty acids, hemin, and bilirubin.1 The multiple binding sites
can also bind exogenous compounds displaying structural
features similar in size and polarity, and therefore it also binds
hydrophobic drugs like Propofol, Diazepam, and Ibuprofen to
mention but a few.2-4 While a protein binding may aid in the
solubilization of hydrophobic drugs it may also, at high binding
affinities, prevent the drugs from reaching their target in a
sufficient concentration to exert their efficacy.5 Determination
of the binding to albumin is thus important for the development
of any new drug.
Antimicrobial peptides represent a versatile class of promising
molecules to combat the growing problem of antibiotic resis-
tance.6 Recent decades have seen a boom in the cases of
antibiotic resistance among different bacteria which unfortu-
nately has not been followed by a similar rise in new classes of
antibiotics to combat them. Our group, has over the years
worked toward simplifying these complex molecules in an
attempt to bring antimicrobial peptides closer to clinical use.7-10
One step has been the identification of the minimum antibacterial
motif in cationic antibacterial peptides (CAP)a using derivatives
of lactoferricin.7 These minimal peptides can be surprisingly
small, and a pronounced antibacterial activity has been shown
Results and Discussion
* To whom correspondence should be addressed. Phone: + 47 776
45505. Fax: + 47 776 44765. E-mail: johan.svenson@chem.uit.no.
† Department of Chemistry, University of Tromsø.
All the peptides used in this study are based on the same
“scaffold” consisting of a nongenetically coded biphenyl alanine
(Bip) moiety flanked by two arginines. Bip has previously been
successfully used in the preparation of CAPs, and it offers a
convenient way of increasing the hydrophobic bulk of a small
peptide.8,9 Boc-Bip is commercially available and easily incor-
porated into the growing peptide chain. A range of other,
similarly bulky and hydrophobic, derivatives are available, but
Bip was used throughout the study for simplicity and to aid the
interpretation. Diversity is instead obtained by modifying the
‡ The Norwegian Structural Biology Centre, Department of Chemistry,
University of Tromsø.
§ Lytix Biopharma AS.
a Abbreviations: Cationic antibacterial peptide, CAP; isothermal titration
calorimetry, ITC; bovine serum albumin, BSA; human serum albumin, HSA;
diisopropylethylamine, DIEA; 1-hydroxybenzotriazole, 1-HOBt; chlorot-
ripyrrolidinophosphonium hexafluorophosphate, PyCloP; O-(benzotriazol-
1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate, HBTU; reversed
phase HPLC, RP-HPLC; biphenyl alanine, Bip; minimal inhibitory con-
centration, MIC.
10.1021/jm0703542 CCC: $37.00 © 2007 American Chemical Society
Published on Web 06/15/2007

Albumin Binding of Antimicrobial Micropeptides
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 14 3335
Figure 2. Experimental data from the ITC runs, in this case 20
sequential additions of CAP 3 to BSA at 25 °C. Each peak represents
the generated heat upon addition. The first peak is small due to slow
syringe leakage during the initial equilibration step.
close to that of the heats of dilution after ∼11 injections (same
amount of peptide and BSA in the reaction cell), representing
a 1:1 complex even though the stoichiometry generated from
the binding isotherm, as can be seen in Table 1, is slightly lower,
averaging 0.8:1 (CAP:BSA). The data were, after subtraction
of the heats of dilution, fitted to a one-site binding model
developed using the BindWorks software. A correlation between
the model and the data is shown in Figure 3. The dissociation
constant (KD, the inverse association constant (1/Ka)) for the
interactions was determined to be between 4.3 and 22.2 µM
(Table 1) using the binding isotherm for each peptide
(Figure 3). Table 1 also presents the thermodynamic parameters
for binding of the peptides to BSA and confirms a similar
binding mechanism for the peptides. The free energy of binding
(∆G) varies from -26.7 to -30.6 kJ/mol with both favorable
enthalpic and entropic contributions (Table 1).
While albumin preferably binds anionic drugs, there are
examples of binding of a range of cationic drugs to albumin
with similar binding affinities as reported here.17,18 As a range
of peptides, including the arginine-containing nonapeptide
vasopressin, also bind to albumin,19 it is not entirely surprising
that the different CAPs associate with BSA to such an extent
as in our experiments. The similarity in binding strength among
the different peptides suggests a binding mode independent of
the variable Y-groups used (Table 1). It remains unclear,
however, whether the binding occurs in any of the two major
binding sites, site I and site II, responsible for most exogeneous
compound binding described in the early work by Sudlow,3 but
as the ligands for these sites usually bear one or two negative
charges, such a binding seems less likely. Many other ligands
bind to albumin in areas other than the well studied site I and
II,18 and we expect this to be the case for the library of Arg-
Bip-Arg amides used in this study. Whether the binding is
mainly governed by hydrophobic interactions via the Bip moiety
or electrostatic interactions between the two Arg and acidic
functionalities on BSA is still under investigation. A likely
binding mode could be through a combination of both, as the
albumins offer both deep hydrophobic binding pockets and an
anionic surface.4 This is further supported by ITC data
(Table 1) which suggest that the binding is both enthalpy and
entropy driven, indicating favorable electrostatic as well as
hydrophobic interactions in the complexes. The isoelectric point
(Ip) of BSA is between 4.7 and 4.9, yielding a net negative
charge under the experimental conditions employed, well suited
for ionic interactions with the two Arg groups.1 The saturation
at stoichiometric amounts and the excellent fit to the theoretical
Figure 1. General structure of the peptide scaffold (Arg-Bip-Arg-Y)
used in the binding studies. “Y” indicates the different C-terminal
modifications.
C-terminus in various ways, as shown above in Figure 1,
yielding a library of secondary and tertiary amides differing in
both bulk, polarity, and stereochemistry and known to have
antibacterial effect while displaying variation in stability toward
degradation (unpublished experiments; manuscript to be submit-
ted to Biochemistry) which is an important property for these
compounds. It should be noted that the Y-group diversity has
been limited to small alterations in order to maintain a high
antibacterial activity.
ITC allows for the determination of the heat generated during
a binding event, and data collected during analysis can give
detailed information about the forces involved in the inter-
action.13-15 Stepwise additions of ligand to a stirred protein
solution generate heat effects based on both dilution and ligand-
protein interactions. Thermal power, i.e., heating or cooling, is
applied to the reaction cell to keep it at the same constant
temperature as a reference cell containing only water, and this
thermal power is proportional to the heat of reaction. The
technique, which is simple, rapid, and versatile, has previously
been used in a range of studies of the interactions between
protein and drugs.16 Thermodynamic descriptors such as the
association constant, enthalpy, and the stoichiometry of a given
interaction can be determined with high accuracy. Initial
binding studies were performed solely with bovine serum
albumin (BSA, g99%). The repeated injections of CAP to the
stirred BSA solution resulted in an exothermic release of energy
represented by the peaks shown in Figure 2. The size of the
peaks decreased with the number of additions, as less binding
sites became available on the BSA molecules upon increased
CAP concentrations.
Data from the ITC studies suggest the formation of a 1:1
complex between the CAPs and BSA. The peak size becomes

3336 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 14
SVenson et al.
Table 1. Data from the ITC Runs and the Theoretical Mouse Model
molecular mass
calcd (obsd)
fraction protein
bound (%)d
peptide
KD (µM)a
na,b
∆G (kJ/ mol)a,c
∆H (kJ/mol)a,c
T∆S (kJ/mol)a
CAP 1
CAP 2
CAP 3
CAP 4
CAP 5
CAP 6
CAP 7
CAP 8
CAP 9
670.9 (670.5)
670.9 (671.5)
642.8 (642.5)
656.8 (657.4)
656.8 (657.4)
670.9 (670.5)
732.9 (732.8)
670.9 (670.5)
594.8 (594.8)
12.2
12.4
4.3
11.1
6.7
17.3
10.0
22.2
14.1
0.9
0.8
0.9
0.7
0.9
0.7
0.8
0.7
0.8
-28.1
-28.0
-30.6
-28.3
-29.6
-27.2
-28.6
-26.7
-27.7
-14.6
-12.5
-10.1
-16.3
-13.7
-15.1
-12.5
-15.2
-8.7
13.5
15.5
20.5
12.0
15.9
12.1
16.1
11.5
19.0
96.5
96.4
98.6
96.7
97.9
95.2
97.3
93.8
95.6
a Data from the ITC experiments and the binding isotherms. b Stoichiometry of the interaction, experimental error (0.1. c Fitting error (15%. d Calculated
using eqs 1-3.
Figure 3. Binding isotherm for CAP 7 and fit to the theoretical 1:1 binding model. The data from the first injection (1) is not included due to slow
peptide diffusion from the syringe during the equilibration step prior to the first injection. STD: 6.94.
Table 2. Data from the Antibacterial Activity Assaya
and [Ptot] represent the total concentrations of BSA (0.55 mM
in this model due to dilution from the large injection volume)
and peptide. Ka is the association constant obtained from ITC
analysis.
MIC
(µg/mL)
MIC with BSA
peptide
(µg/mL)b
CAP 1
CAP 2
CAP 3
CAP 4
CAP 5
CAP 6
CAP 7
CAP 8
CAP 9
4 ( 0
4 ( 0
5 ( 1
4 ( 0
7 ( 1
5 ( 1
2 ( 0
2 ( 0
47 ( 3
31 ( 15
31 ( 15
47 ( 3
[MP]
Ka )
(1)
(2)
47 ( 3
[M]‚[P]
38 ( 13
38 ( 13
13 ( 7
[MP]
Ka )
25 ( 0
110 ( 10
(([Mtot] - [MP])‚([Ptot] - [MP]))
a Generated using standardized microdilution techniques. Experiments
performed in triplicate. b Present at a concentration of 0.55 mM.
1
Ka
[MP]2 - [Mtot] + [Ptot] +
‚[MP] + [Mtot]‚[Ptot] ) 0
(
)
model point toward a single binding site with affinity for the
peptides. The fact, that the peak size does not equal that of the
heats of dilution upon saturation is merely an indication of the
presence of weak nonspecific interactions in addition to this
binding site, which is only to be expected, considering the size
of albumin and the many potential interaction points on its
surface.
A 20 g mouse with 1.67 mL of blood injected with 0.28 mg
of CAP (14 mg/kg, resulting in a 2.3-fold excess of BSA
compared to CAP) in a total volume of 200 µL was used as
theoretical model to calculate the amount of BSA bound CAP.
The sequences for bovine, human, and murine serum albumin
align with 70-76% identity (see Supporting Information for
sequence alignment analysis), making comparisons between
these systems feasible. The extent of complexation was calcu-
lated using eq 3, derived from eqs 1 and 2, where [M], [P], and
[MP] represent the concentrations of macromolecule ([M], BSA
in this case), peptide [P], and complex [MP], respectively. [Mtot]
(3)
As is shown in Table 1, the theoretical binding to albumin
was very similar, roughly ranging from 94 to 98%. It is clear
that the small difference in dissociation constant within the
library here plays little role for the extent of protein binding.
Staphylococcus aureus (antibiotic resistant strains as well as
sensitive ones) are highly sensitive to this class of molecules,
and the minimal inhibitory concentrations (MIC) for the different
CAPs against strain ATCC 25923, determined through a
standard micro dilution technique,20 are shown in Table 2.
While the hydrophobicity offered by the Bip side chain
together with the two positively charged arginines is sufficient
for some bactericidal activity, it is obvious that incorporation
of additional bulky hydrophobic elements will further improve
this activity. Incorporation of an extra phenyl group on the
C-terminus is enough to generate a tenfold decrease in MIC.
While the different Y-groups hardly seemed to influence

Albumin Binding of Antimicrobial Micropeptides
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 14 3337
Table 3. Binding and Antibacterial Data for the Peptides on Different Albumin of Different Purity
peptide
albumina
KD (µM)b
n b,c
∆G (kJ/mol)b,d
∆H (kJ/mol)b,d
T∆S (kJ/ mol)b
MIC with albumin (µg/ mL)e
CAP 3
HSA I
HSA II
BSA I
BSA II
HSA I
HSA II
BSA I
BSA II
HSA I
HSA II
BSA I
BSA II
15.7
15.5
4.3
0.8
0.8
0.9
1.1
0.7
0.7
0.8
0.8
0.8
0.7
0.8
0.8
-27.4
-27.4
-30.6
-27.9
-28.4
-27.4
-28.6
-28.0
-28.3
-27.1
-27.7
-27.6
-24.2
-25.1
-10.1
-11.4
-25.8
-20.2
-12.5
-12.9
-16.4
-7.1
3.4
2.3
40 ( 9
80 ( 20
47 ( 3
20.5
16.5
2.6
13.0
10.4
15.9
10.0
12.3
10.8
18.2
14.1
14.5
30 ( 10
6 ( 1
CAP 7
CAP 9
7.2
11 ( 8
16.1
15.1
11.9
20.0
19.0
11.9
13 ( 7
6 ( 0
113 ( 10
100 ( 9
110 ( 10
80 ( 11
-8.7
-15.7
a Albumin purities: HSA I ∼99%, HSA II g96%, BSA I g99%, BSA II g96%. b Data from the ITC experiments and the binding isotherms. c Stoichiometry
of the interaction, experimental error (0.1. d Fitting error (15%. e Generated using standardized microdilution techniques. Experiments performed in duplicate.
Albumin is present at a concentration of 0.55 mM.
albumin binding, its role in killing bacteria is obvious, as CAP
9, lacking any bulky hydrophobic moieties, displays the highest
MIC value, while CAP 7 with its two phenyl groups displays
a 25 times lower MIC value (Table 2). It has previously been
shown that a balance between charge and hydrophobic bulk
exists for this group of molecules, and a clear distinction is seen
here for the peptides with aromatic Y-groups even though size,
orientation, and stereochemistry seem to play little role in this
case.8 Their BSA complexation is similar, and one can assume
a similar behavior for the binding to bacteria even though the
albumin lacks the anionic lipopolysaccarides of the bacteria
responsible for the initial binding.11,12 This indicates that binding
only is not enough for killing bacteria, adding further support
to the theory that insertion of a hydrophobic element is a crucial
step for cytoplasmic membrane permeabilization.
A 10-fold increase in MIC is seen when BSA is included in
the bactericidal assay at physiological concentration as can be
seen in Table 2. This corresponds well with the roughly ∼95%
fraction that is bound to BSA in the theoretical model under
similar conditions, thereby reducing the effective concentration
of free CAP ten times. The concentration of the CAPs at MIC
are some three times lower than in the theoretical model (data
not shown) so a slightly higher protein binding could be
expected in those cases. Albumin binding of CAP 9 with a
Y-group differing in both size and hydrophobicity from the rest
of the peptides did not increase the MIC to the same extent,
suggesting a slightly different behavior in the presence of
bacteria. It should be noted that the experimental determination
of Ka used to calculate the theoretical bound fraction, for
practical reasons, is performed in the absence of competing
bacteria. Keeping a constant concentration of bacteria during
the course of the binding study would indeed be very difficult.
The increased osmolality caused by BSA itself has no effect
on the growth of S. aureus, and control experiments without
CAPs (with and without BSA) showed comparable growth rates
(data not shown).
ranging from 4.3 to 18.2 µM. All peptides displayed nearly
identical ∆G values. The pure, essentially fatty acid free
albumins BSA I and HSA I associated with the peptides to the
same extent as those of less purity (g96%), BSA II and HSA
II. The binding contribution to albumin would be difficult to
evaluate with a protein sample of much lower purity. This
similarity in binding implies that the role of associated endog-
enous fatty acids is of little significance for the association, at
least within the frame of these experiments. It further suggests
that the peptide binding site is different from the fatty acid
binding sites on the albumin surface. Of more interest is perhaps
the difference in T∆S seen for HSA compared to BSA for CAP
3 and CAP 7. These peptides both have bulky, lipophilic
Y-groups, and the contribution to the overall binding for these
seems to differ slightly between the different classes of albumins
although they bind to the same extent. CAP 9 with its C-terminal
isopropyl modification does not display this difference. Further
studies are underway to investigate the detailed mode of binding
for this class of molecules, using peptides displaying more
diverse structural features which may impair the antibacterial
effect. The extent of the increase in MIC was similar for the
different albumins. Binding of CAP 3 induces a larger increase
than does CAP 7 and CAP 9 for both HSA and BSA.
It may be perceived that the interactions between the peptides
and the anionic bacterial surface may be strong enough to cause
the peptides to dissociate from albumin, preventing the protein
from inhibiting the bactericidal effect. However, since albumin
itself is anionic at physiological pH, the bacterial surface needs
to offer a superior binding environment to displace the bound
peptide. For the peptides with a bulky Y-group, no evidence of
such competitive displacement is detected though, suggesting
that the charged guanidine groups, responsible for the initial
bacterial interaction, become unavailable upon binding to the
binding site. CAP 9, lacking a bulky Y-group, is however easier
to displace by the bacteria, and the increase in MIC is lower
for this peptide. Unfortunately, this peptide is the least efficient
bactericidal compound in the library.
To establish if these observations were solely associated with
BSA of certain purity, complementary experiments were
performed on HSA as well, with three peptides diverse in
structure and antimicrobial activity (CAP 3, CAP 7, CAP 9).
Four albumins, two human, and two bovine of different purities
were selected for this study. The purities were different (see
the Experimental Section for details about the different albu-
mins) to allow investigation of the potential role of associated
fatty acids on peptide binding. The results from these experi-
ments are summarized below in Table 3.
As the protein binding seems to lower the effective concen-
tration of CAPs, it will be important to take it into consideration
for future studies in blood and for in ViVo testing of these and
similar molecules. Rational drug design could be a way of
approaching the problem with albumin binding and lowered in
ViVo efficacy, but that requires detailed information about the
albumin/peptide complex.18,21 The lowered effect of binding of
CAP 9 illustrates that it should be feasible to rationally design
peptides of this class that lower the unwanted increase in MIC
caused by albumin complexation and thus still maintain a high
antibacterial activity. It is also important to bear in mind that
The binding of the three peptides to the different albumins
was generally very similar to that presented in Table 1
(corresponding to BSA I in Table 3) with dissociation constants

3338 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 14
SVenson et al.
brine before the organic solvent was removed under vacuum.
Complete removal of the Boc protecting group of the resulting
peptide was achieved through the use of 95% TFA or acetyl chloride
in anhydrous methanol. This approach was also used for incorpora-
tion of the different amines forming the C-terminal amides (apart
from the capping with N,N-dibenzylamine which was prepared
according to the procedure below) yielding the Y-group diversity.
Procedure for Solution-Phase Amide Formation Using Py-
CloP.22 Boc-Arg-OH (1 equiv), N,N-dibenzylamine (1.1 equiv), and
PyCloP (1 equiv) were dissolved in a mixture of dry DCM and
DMF (2:1 v/v). The solution was cooled on ice, and DIEA (2 equiv)
was added under stirring. The solution was stirred for 1 h at room
temperature. The reaction mixture was evaporated, redissolved in
ethyl acetate, and washed with citric acid, sodium bicarbonate, and
brine. The solvent was removed under vacuum and the Boc
protecting group removed in the dark using 95% TFA.
the parts of the molecule responsible for protein binding seem,
to a large extent, to be the same as those needed for bactericidal
effect, leaving little room for large molecular alterations. Co-
administration of a displacer compound is also a potential
strategy to increase the effect of the bound peptides which, just
like the previously mentioned drug design approach, requires a
more detailed knowledge about the chemistry of the binding
site.18 One way to minimize the influence of protein binding
could be dosage modification. It is however difficult to discuss
dosage issues before proper knowledge is obtained about how
these compounds behave in such a complex matrix as blood.
More studies are thus needed and are underway.
Conclusion
To conclude, ITC has been used to study the interactions
between a range of active antibacterial peptides and bovine and
human serum albumin. It was shown that the different peptides
bind to albumins to a similar extent, with a dissociation constant
of around 10 µM. The mode of binding is unclear at the moment,
but all the peptides seem to bind to a single binding site. In a
theoretical mouse model this corresponds to ∼95% protein
binding. Possessing a net negative charge at physiological pH,
it is likely the albumins compete with the bacterial surface for
peptide binding. The bound peptides do not seem to be able to
participate in destroying bacteria, as a tenfold increase in MIC
is seen when albumin is included in the antibacterial assay in
physiological concentrations. Collectively the results indicate
that these antibacterial peptides bind to albumin and that this
binding effectively lowers the effective concentration of peptides
available to combat bacteria. This drop in activity upon protein
binding is not desired, but can hopefully be circumvented by
proper dosage allowing the peptides to effectively continue to
destroy bacteria. While lowering the effective concentration of
CAPs in Vitro, a high protein binding may not necessarily be a
problem considering the complexity in ViVo with regards to the
pharmacokinetics of the compounds, i.e., volume of distribution,
mode of clearance and half-life.
CAP 3: 1H NMR (D2O), δ 0.96 (d, J ) 6.9 Hz, 3H), 1.13-
1.28 (m, 2H), 1.34-1.49 (m, 2H), 1.52-1.64 (m, 2H), 1.81-1.94
(m, 2H), 2.61-2.70 (m, 1H), 2.92-3.05 (m, 5H), 3.15-3.20 (m,
3H), 3.96 (t, J ) 7.1 Hz, 1H), 4.02 (t, J ) 6.5 Hz, 1H), 4.61 (dd,
J ) 9.7, 6.7 Hz, 1H), 7.06 (d, J ) 7.2 Hz, 2H), 7.22 (dd, J ) 8.5,
6.1 Hz, 1H), 7.30 (dd, J ) 16.9, 7.9 Hz, 4H), 7.42 (t, J ) 7.4 Hz,
1H), 7.52 (t, J ) 7.7 Hz, 2H), 7.72 (dd, J ) 18.3, 7.8 Hz, 4H);
ESMS: calcd for C36H50N10O3: 670.9, found 670.5; Purity
determined by HPLC: Retention time: 13.80 min, purity 99.7%.
See Supporting Information for data on remaining compounds.
Peptide Purification and Analysis. The peptides were purified
using reversed phase HPLC (RP-HPLC) on a Delta-Pak (Waters)
C18 column (100 Å, 15 µm, 25 × 100 mm) employing a mixture
of water and acetonitrile (both containing 0.1% TFA) as the eluent.
The peptides were further analyzed by RP-HPLC using an analytical
Delta-Pak (Waters) C18 column (100 Å, 5 µm, 3.9 × 150 mm).
Positive ion electrospray mass spectrometry on a VG Quattro
quadrupole mass spectrometer (VG Instruments Inc., Altringham,
UK) ensured isolation and identification of the correct products.
ITC Studies. Heats of interaction were determined using a CSC
5300 NanoIsothermal Titration Calorimeter III with a 1 mL cell
volume (Calorimetry Sciences Corporation, Lindon, UT). In a
typical titration experiment, the peptide (2.1 mM) was added in 20
aliquots (5 µL) to a stirred (200 rpm) solution of albumin (0.11
mM) in an aqueous buffer at 25 °C. The buffer used was a 50 mM
Tris-HCl, 10 mM CaCl2 at pH 8.2. An interval of 400 s between
the injections was allowed to reach equilibrium between the
interacting species. The heats of dilution were determined in a
similar fashion where the peptides were added to a stirred buffer
solution without albumin. Subtraction of the dilution heat yielded
the heat of interaction, and a binding isotherm from which the
association constant and complex stoichiometry was calculated
using BindWorks analysis software.
Experimental Section
Chemicals. Essentially fatty acid and globulin free bovine serum
albumin g99% (BSA I) (A0281), bovine serum albumin g96%
(BSA II) (A9085), essentially fatty acid free human serum albumin
∼99% (HSA I) (A3782), and human serum albumin g96% (HSA,
HSA II) (A4327) were supplied by Sigma-Aldrich. Boc-Arg-OH
and Boc-p-phenyl-Phe-OH were supplied by Bachem. Benzylamine,
isopropylamine, 2-phenylethylamine, 3-phenylpropylamine, (R)-2-
phenylpropylamine, (S)-2-phenylpropylamine, N,N-methylbenzy-
lamine, N,N-ethylbenzylamine, and N,N-dibenzylamine providing
the C-terminal diversity were purchased from Fluka. Diisopropy-
lethylamine (DIEA), 1-hydroxybenzotriazole (1-HOBt), chlorot-
ripyrrolidinophosphonium hexafluorophosphate (PyCloP), and O-(ben-
zotriazol-1-yl)-N,N,N′,N′ tetramethyluronium hexafluorophosphate
(HBTU) were purchased from Fluka. Other reagents and solvents
were supplied by Sigma-Aldrich and were used without further
purification with the exception of DCM, which was filtered through
alumina.
Microbiological Studies. MIC determinations and growth stud-
ies, with and without albumins, on S. aureus, strain ATCC 25923,
were performed by Toslab AS using standard methods.20
Acknowledgment. Hilde Ulvatne Marthinsen, Toslab AS,
is acknowledged for supplying the MIC data and for performing
all the bacterial assays. Financial support from the Research
Council of Norway and Lytix Biopharma is also gratefully
acknowledged. The Norwegian Structural Biology Centre (Nor-
Struct) is supported by the Functional Genomics Program
(FUGE) of the Research Council of Norway.
General Procedure for Solution-Phase Peptide Coupling
Using HBTU. The peptides were prepared in solution by stepwise
amino acid coupling using Boc protecting strategy according to
the following general procedure.7 The C-terminal peptide part with
a free amino group (1 equiv) and the Boc protected amino acid
(1.05 equiv) and 1-HOBt (1.8 equiv) were dissolved in DMF (2-4
mL/mmol of amino component) before addition of DIEA (4.8
equiv). The mixture was cooled on ice, and HBTU (1.2 equiv) was
added in portions before it was shaken at ambient temperature for
1-2 h on a rocking table. The reaction mixture was diluted with
ethyl acetate and washed with citric acid, sodium bicarbonate, and
Supporting Information Available: Experimental and 1H NMR
spectroscopic data for CAP 1-9 in conjunction with LC traces
and structures; sequence alignment analysis for human, bovine, and
murine serum albumin. This material is available free of charge
via the Internet at http://pubs.acs.org.
References
(1) Peters, T. Serum Albumin. AdV. Protein Chem. 1985, 37, 161-245.
(2) He, X. M.; Carter, D. C. Atomic structure and chemistry of human
serum albumin. Nature 1992, 358, 209-215.

Albumin Binding of Antimicrobial Micropeptides
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 14 3339
(3) Sudlow, G.; Birkett, D. J.; Wade, D. N. Characterization of two
specific drug binding sites on human serum albumin. Mol. Pharmacol.
1975, 11, 824-832.
(13) Leavitt, S.; Freire, E. Direct measurement of protein binding
energetics by isothermal titration calorimetry. Curr. Opin. Struct. Biol.
2001, 11, 560-566.
(14) Thomson, J. A.; Ladbury, J. E. Isothermal titration calorimetry: a
Tutorial. In Biocalorimetry 2; Ladbury, J. E.; Doyle, M. L., Eds.;
Wiley & Sons Ltd.: Chicester, 2004; pp 37-58.
(15) Perozzo, R.; Folkers, G.; Scapozza, L. Thermodynamics of protein-
ligand interactions: History, presence, and future aspects. J. Rec.
Sign. Trans. 2004, 24, 1-52.
(4) Ghuman, J.; Zunszain, P. A.; Petitpas, I.; Bhattacharya, A. A.; Otagiri,
M.; Curry, S. Structural basis of the drug-binding specificity of human
serum albumin. J. Mol. Biol. 2005, 353, 38-52.
(5) Kratochwil, N. A.; Huber, W.; Muller, F.; Kansy, M.; Gerber, P. R.
Predicting plasma protein binding of drugs: a new approach.
Biochem. Pharmacol. 2002, 64, 1355-1374.
(6) Hancock, R. E. W. Host defense (Cationic) peptides. What is their
future clinical potential? Drugs 1999, 57, 469-473.
(16) Ward, W. H. J.; Holdgate, G. A. Isothermal titration calorimetry in
drug discovery. Prog. Med. Chem. 2001, 38, 309-376.
(17) Kragh-Hansen, U. Molecular aspects of ligand binding to serum
albumin. Pharmacol. ReV. 1981, 33, 17-53.
(7) Strøm, M. B.; Haug, B. E.; Skar, M. L.; Stensen, W.; Stiberg, T.;
Svendsen, J. S. The pharmacophore of short cationic antibacterial
peptides. J. Med. Chem. 2003, 46, 1567-1570.
(18) Kragh-Hansen, U.; Chuang, V. T. G.; Otagiri, M. Practical aspects
of the ligand-binding and enzymatic properties of human serum
albumin. Biol. Pharm. Bull. 2002, 25, 695-704.
(8) Haug, B. E.; Stensen, W.; Stiberg, T.; Svendsen, J. S. Bulky
nonproteinogenic amino acids permit the design of very small and
effective cationic antibacterial peptides. J. Med. Chem. 2004, 47,
4159-4162.
(19) Menezo, Y.; Khatchadourian, C. Peptides bound to albumin. Life
Sci. 1986, 39, 1751-1753.
(9) Haug, B. E.; Skar, M. L.; Svendsen, J. S. Bulky aromatic amino
acids increase the antibacterial activity of 15-residue bovine lacto-
ferricin derivatives. J. Pept. Sci. 2001, 7, 425-432.
(10) Strøm, M. B.; Rekdal, Ø.; Svendsen, J. S. Antimicrobial activity of
short arginine- and tryptophan-rich peptides. J. Pept. Sci. 2002, 8,
431-437.
(11) Vorland, L. H.; Ulvatne, H.; Rekdal, Ø.; Svendsen, J. S. Initial binding
sites of antimicrobial peptides in Staphylococcus aureus and Es-
cherichia coli. Scand. J. Infect. Dis. 1999, 31, 467-473.
(12) Zasloff, M. Antimicrobial peptides of multicellular organisms. Nature
2002, 415, 389-395.
(20) Amsterdam, D. Susceptibility testing of antimicrobials in liquid media.
In Antibiotics in laboratory medicine, 4th ed.; Lorian V, Ed.; Williams
and Wilkins.: Philadelphia, 1996; pp 75-78.
(21) Mao, H.; Hajduk, P. J.; Craig, R.; Bell, R.; Borre, T.; Fesik, S. W.
Rational design of diflunsial analogues with reduced affinity for
human serum albumin. J. Am. Chem. Soc. 2001, 123, 10429-10435.
(22) Coste, J.; Fre´rot, E.; Jouin, P. Coupling N-methylated amino acids
using PyBroP and PyCloP halogenophosphonium salts: Mechanism
and fields of application. J. Org. Chem. 1994, 59, 2437-2446.
JM0703542