Báo cáo khoa học: nNOS inhibition, antimicrobial and anticancer activity of the amphibian skin peptide, citropin 1.1 and synthetic modifications pot

nNOS inhibition, antimicrobial and anticancer activity of theamphibian skin peptide, citropin 1.1 and synthetic modifications The solution structure of a modified citropin 1.1 Jason Doyl

nNOS inhibition, antimicrobial and anticancer activity of the

amphibian skin peptide, citropin 1.1 and synthetic modifications

The solution structure of a modified citropin 1.1

Jason Doyle1, Craig S Brinkworth2, Kate L Wegener2, John A Carver3, Lyndon E Llewellyn1,

Ian N Olver4, John H Bowie2, Paul A Wabnitz2and Michael J Tyler5

1 Australian Institute for Marine Science, Townsville MC, Queensland, Australia; 2 Department of Chemistry,

The University of Adelaide, Australia; 3 Department of Chemistry, University of Wollongong, Wollongong, Australia;

4 Oncology Department, Royal Adelaide Hospital and Department of Medicine, The University of Adelaide,

South Australia, Australia;5Department of Environmental Biology, The University of Adelaide, South Australia, Australia

A large number of bioactive peptides have been isolated

from amphibian skin secretions These peptides have a

variety of actions including antibiotic and anticancer

acti-vities and the inhibition of neuronal nitric oxide synthase

We have investigated the structure–activity relationship of

citropin 1.1, a broad-spectrum antibiotic and anticancer

agent that also causes inhibition of neuronal nitric oxide

synthase, by making a number of synthetically modified

analogues Citropin 1.1 has been shown previously to form

an amphipathic a-helix in aqueous trifluoroethanol The

results of the structure–activity studies indicate the terminal

residues are important for bacterial activity and increasing

the overall positive charge, while maintaining an amphi-pathic distribution of residues, increases activity against Gram-negative organisms Anticancer activity generally mirrors antibiotic activity suggesting a common mechanism

of action The N-terminal residues are important for inhi-bition of neuronal nitric oxide synthase, as is an overall positive charge greater than three The structure of one of the more active synthetic modifications (A4K14-citropin 1.1) was determined in aqueous trifluoroethanol, showing that this peptide also forms an amphipathic a-helix

Keywords: citropin; antibacterial; anticancer; nNOS activity

Amphibians have rich chemical arsenals that form an

integral part of their defence systems, and also assist with

the regulation of dermal physiological action In response to

a variety of stimuli, host defence compounds are secreted

from specialized glands onto the dorsal surface and into the

gut of the amphibian [1–4] A number of different types of

bioactive peptides have been identified from the glandular

skin secretions of Australian anurans of the Litoria genus,

including (a) smooth muscle active neuropeptides of the

caerulein family [5–8], and (b) wide-spectrum antibiotics,

e.g., the caerin peptides from green tree frogs of the genus

Litoria [6–8], the citropins from the tree frog, L citropa

[9,10], and the aureins from the bell frogs, L aurea and

L raniformis[11] Among the most active of the antibiotic

peptides are caerin 1.1, citropin 1.1 and aurein 1.2:

caerulein 1.1 pEQGY(SO3)TGWMDF-NH2; caerin 1.1

GLLSVLGSVAKHVLPHVVPVIAEHL-NH2; citropin 1.1 GLFDVIKKVASVIGGL-NH2; aurein 1.2 GLFDIIKKI AESF-NH2

Aurein 1.2 contains only 13 amino acid residues and is the smallest peptide from an anuran reported to have significant antibiotic activity The aurein peptides have also been shown to exhibit modest anticancer activity in tests carried out by the National Cancer Institute (Washington,

WA, USA) [12]

The solution structures of the antibiotic (and anticancer active if appropriate) peptides shown above have been investigated by NMR spectroscopy In d3-trifluoroethanol/ water mixtures, caerin 1.1 adopts two well-defined helices (Leu2–Lys11 and Val17–His24) separated by a hinge region

of less-defined helicity and greater flexibility, with hydro-philic and hydrophobic residues occupying well defined zones [13] The central hinge region is necessary for optimal antibiotic activity [13] Similar NMR studies of citropin 1.1 [9] and aurein 1.2 [11] show that these peptides adopt conventional amphipathic a-helical structures, a feature commonly found in membrane-active agents [1–4,8] Inter-action occurs at the membrane surface with the charged, and normally basic peptide adopting an a-helical confor-mation and attaching itself to charged, and normally anionic sites on the lipid bilayer This ultimately causes disruption of normal membrane function leading to lysis of the bacterial or cancer cell [14–16]

Many Australian anurans that we have studied conform

to the model outlined above in that they have a variety of

Correspondence to: J H Bowie, Department of Chemistry,

The University of Adelaide, South Australia, Australia.

Fax: + 61 8303 4358, Tel.: + 61 88303 5767,

E-mail: john.bowie@adelaide.edu.au

Abbreviations: MIC, minimum inhibitory concentration; NADPH,

nicotinamide adenine nucleotide phosphate, reduced form;

eNOS, endothelial nitric oxide synthase; iNOS, inducible NOS;

nNOS, neuronal NOS; RMD, restrained molecular dynamics;

SA, simulated annealing.

(Received 23 September 2002, revised 28 November 2002,

accepted 15 January 2003)

host defence peptides in the skin (and gut) glands including a

neuropeptide that acts on smooth muscle and at least one

powerful wide-spectrum antibiotic and/or anticancer active

peptide like those described above [8] However there are

some species of anuran that divert markedly from this

scenario For example, the Australian stony creek frog

(L lesueuri) [17] and the giant tree frog (L infrafrenata) [18]

both produce the neuropeptide, caerulein, but lack any

wide-spectrum antimicrobial peptide The major peptides in the

skin secretions of these two Litoria species have been named

lesueurin and frenatin 3, respectively: their sequences are

shown below: Lesueurin GLLDILKKVGKVA-NH2;

Fren-atin 3 GLMSVLGHAVGNVLGGLFKPKS-OH Neither

lesueurin nor frenatin 3 show any significant antibiotic or

anticancer activity, but in tests carried out at the Australian

Institute of Marine Science (Townsville, Queensland,

Australia), both peptides were shown to inhibit the

forma-tion of nitric oxide by the neuronal isoform of nitric oxide

synthase (nNOS) with IC50values at lM concentrations [17]

Further nNOS testing on other peptides isolated from tree

frogs of the Litoria genus showed that each species has at

least one major skin peptide that inhibits nNOS and that

there are (at least) three groups of peptides that inhibit

nNOS Inhibitor group 1 includes citropin type peptides

(that are also antimicrobial and anticancer agents); for the

sequence of citropin 1.1 see above The second group

comprises peptides with sequence similarity to frenatin 3:

these peptides show no significant antimicrobial or

antican-cer activity The third inhibitor group includes the caerin 1

peptides (see the sequence of caerin 1.1 above): these peptides

also show powerful antimicrobial and antifungal activity

The three nitric oxide synthases, namely neuronal,

endo-thelial (eNOS) and inducible (iNOS), are highly regulated

enzymes responsible for the synthesis of the signal molecule,

nitric oxide They are among the most complex enzymes

known (e.g., for nNOS see [19,20]) By a complex sequence

involving binding sites for a number of cofactors including

heme, tetrahydrobiopterin, FMN, FAD and NADPDH,

nNOS converts arginine to citrulline, releasing the

short-lived but reactive radical NO [21,22] Nitric oxide synthases

are composed of two domains: (a) the catalytic oxygenase

domain that binds heme, tetrahydrobiopterin and the

substrate arginine, and (b) the electron supplying reductase

domain that binds NADPH, FAD and FMN

Communi-cation between the oxygenase and reductase domains is

determined by the regulatory protein calmodulin which

interacts at a specific site between the two domains In the

cases of nNOS and eNOS isoforms, but not for iNOS,

calmodulin is regulated by intracellular Ca2+[23–26]

Dime-rization of the oxygenase domain is necessary for catalytic

activity [21,22] The amphipathic amphibian peptides inhibit

nNOS by interacting with Ca2+-calmodulin, changing the

shape of the regulatory enzyme, thus impeding its

inter-action at the calmodulin binding site on nNOS [17] There

are other examples of small helical peptides inhibiting

nNOS in this way [27,28]

The amphibian may have two possible uses for a peptide

that inhibits nNOS First, on attack by a predator, the

amphibian may use the nNOS inhibitor to regulate its own

physiological state The second scenario is that the nNOS

inhibitors are front-line defence compounds A predator

ingesting even a small amount of the nNOS inhibitor could

be seriously affected if only part of its NO messenger capability is reduced All animals produce NOS isoforms, and it has been reported that bacteria also produce NOS [29–32]

The citropin 1 group of peptides has significant anti-biotic, anticancer and nNOS activity, despite being com-prised of only 16 amino-acid residues In this paper we describe our investigations into the structure/activity rela-tionships for the amphibian peptide citropin 1.1 The activities of citropin 1.1 are compared with those of a number of synthetically modified citropins 1 and other related molecules to gain insight into the sequence require-ments for activity The 3D solution structure of one of the most potent of the synthetically modified citropins has been determined using 1H-NMR procedures This structure is compared with that already determined for citropin 1.1 [9] Methods

Preparation of synthetic peptides All peptides listed in Tables 1 and 4 were synthesized (by Mimotopes, Clayton, Victoria, Australia) using L-amino acids via the standard N-a-Fmoc method (full details including protecting groups and deprotection have been reported recently [33]) Synthetic versions of naturally occurring peptides were shown to be identical to the native form by electrospray mass spectrometry and HPLC Bioactivity assays

Bioactivity testing was carried out on citropin 1.1,

D-citropin 1.1 and A4K14-citropin of both 95% and 80% purities The activities were the same range for each pair of samples Activity tests on all other synthetic modifications were performed with samples which had
80% purity as adjudged by HPLC

Antimicrobial testing Synthetic peptides were tested for antibiotic activity by the Microbiology Department of the Institute of Medical and Veterinary Science (Adelaide, Australia) by a standard method [34] The method involved the measurement of inhibition zones (produced by the applied peptide) on a thin agarose plate containing the microorganisms listed in Table 2 Concentrations of peptide tested were 100, 50,

25, 12.5, 6, 3 and 1.5 lgÆmL)1 The maximum error in the antibiotic results listed in Table 2 is ± 1 dilution factor: e.g., if the MIC is 3 lgÆmL)1, the maximum possible range

is 1.5–6 lgÆmL )1 Anticancer activity testing Synthetic peptides were tested in the human tumour line testing program of the US National Cancer Institute [12] All compounds were tested initially against three tumour lines (breast, lung and CNS cancers), and if activity was indicated, the peptide was then tested in vitro against 60 human cell lines If a particular peptide failed the first stage

of the test program it is indicated as inactive (even though it may have shown some activity) Full test data are not

provided in this paper The summary data recorded in

Table 3 indicate the particular groups of cancers tested, the

average IC50concentration of the peptide against that group

of cancers and the number of tumours, out of 60 tested, that

were affected by the particular peptide For details of how

the IC50value is determined from graphical data see [12]

Neuronal nitric oxide synthase inhibition Inhibition of nNOS was measured by monitoring the conversion of [3H]arginine to[3H]citrulline In brief, this involved incubation of a homogenate of rat cerebella (which had endogenous arginine removed by ion exchange chro-matography) in a reaction buffer (33 mMHepes, 0.65 mM

EDTA, 0.8 mM CaCl2, 3.5 lgÆmL)1 calmodulin, 670 lM

b-NADPH, 670 lM, dithiothreitol, pH 7.4) containing

20 nM [3H]arginine (NEN Life Sciences, Boston, MA, USA) The nNOS inhibitor, Nx-nitro-L-arginine (1 mM) was used to measure background radioactivity Reactions were terminated after 10 min with 50 lL of 0.3MEGTA

An aliquot (50 lL) of this quenched reaction mixture was transferred to 50 lL of 500 mMHepes (pH 5.5) AG50W-X8 (Na+ form) resin (100 lL) was added to separate [3H]arginine from [3H]citrulline After repeated vortexing, this slurry was centrifuged at 1200 g for 10 min, and 70 lL

of supernatent was removed and the [3H]citrulline measured

by scintillation counting Peptides selected for further examination to determine the mechanism of inhibition were assayed in the same reaction buffer as used for initial screening except that it contained 30 nM [3H]arginine supplemented with 0.3–13.3 mMarginine

Data analysis for nNOS studies Peptide inhibition curves were fitted using the curve-fitting routine of SIGMAPLOT(SPSS, Chicago, IL, USA) using a variation of the Hill equation: fmols [3H]citrulline produc-tion¼ 1/(1 + [inhibitor]/ICn

50), where IC50is the concen-tration at which the peptide causes 50% inhibition and n is the slope of the curve and can be considered as a pseudo Hill coefficient [35] Lineweaver–Burk plots [36] were generated using SIGMAPLOT (SPSS, Chicago, IL, USA) The mean error in the IC50results listed in Table 4 is ± 1.3%

NMR spectroscopy of citropin synthetic modification (A4K14-citropin 1.1)

NMR experiments were performed on a solution of 5.7 mg

of A4K14-citropin 1.1 dissolved in a mixture of water (0.35 mL) and d3-trifluoroethanol (0.35 mL), that had a final concentration of 4.9 mMand a measured pH of 4.12 NMR spectra were acquired on a Varian Inova-600 NMR spectrometer at a 1H frequency of 600 MHz and 13C

Table 1 Citropin 1.1 and synthetic modifications Modifications are

shown in bold.

Relative molecular mass

a,b

1297 Modified peptide

1 GlfdvikkvasviGGl-NH 2 1614

2 GL A DVIKKVASVIGGL-NH 2b 1537

3 GLF A VIKKVASVIGGL-NH 2 1570

4 GLFDVI A KVASVIGGL-NH 2b 1557

5 GLFDVIK A VASVIGGL-NH 2a 1557

6 GLFDVI AA VASVIGGL-NH 2

a,b

1500

7 GLFDVIKKVA A VIGGL-NH 2 1599

8 GLFDVIKKVASVIGG A -NH 2b 1572

9 GLF E VIKKVASVIGGL-NH 2

b

1628

10 GLFDVIKKVAS K IGGL-NH 2b 1643

11 GLFDVIKKVASVI K GL-NH 2 1685

12 GLFDVIKKVAS K I K GL-NH 2

b

1714

13 GLFDVIKKVASVI KK L-NH 2 1756

14 GLFDVI A KVASVI KK L-NH 2 1699

15 GLF A VIKKVASVI K GL-NH 2 1655

16 GLF A VIKKVASVI KK L-NH 2 1712

17 GLF A VIKKVA A VI KK L-NH 2 1696

18 GLF A VIKKVA A VI RR L-NH 2 1752

19 GLF A VIKKVA K VI KK L-NH 2 1753

20 K LF A VIKKVA A VIGGL-NH 2b 1625

21 K LF A VIKKVA A VI RR L-NH 2b 1823

22 GLF K VIKKVASVIGGL-NH 2 1627

23 GLF K VIKKVA K VI KK L-NH 2 1810

Retro

1.1 LGGIVSAVKKIVDFLG- NH 2 1614

a

These compounds show no antibiotic activity against the listed

bacteria in Table 2 at MIC ¼ 100 lgÆmL)1 b Compounds so

marked failed the initial NCI tests against three cancer types Many

of these compounds do show activity, but not below concentrations

of 10)4M For NCI test results, see Table 3.

Table 2 Antibiotic activites of Citropin 1.1 and synthetic analogues [MIC values (lgÆmL)1)] The absence of a figure means the activity is

> 100 lgÆmL)1 For error range see Methods.

a Gram-negative organism.

frequency of 150 MHz All NMR experiments were

acquired at 25C 1H-NMR resonances were referenced

to the methylene protons of residual d3-trifluoroethanol

(3.918 p.p.m) The13C (F1) dimensions of the heteronuclear

single-quantum coherence (HSQC) and heteronuclear

mul-tiple-bond correlation (HMBC) spectra were referenced to

the 13CD2 (60.975 p.p.m) and 13CF3 (125.9 p.p.m)

reso-nances of d3-trifluoroethanol, respectively

Double-quantum-filtered correlation spectroscopy

(DQF-COSY) [37]; total correlation spectroscopy (TOCSY)

[38]; and nuclear Overhauser effect spectroscopy

(NOESY) [39]; were all collected in the phase-sensitive

mode using time proportional phase incrementation [40]

in t1 Two hundred and fifty-six t1 increments were used

for each experiment Thirty-two scans were time averaged

for each increment in the TOCSY and NOESY

experi-ments, while 16 scans were averaged in the DQF-COSY

experiment The free induction decay in t2 consisted of

2048 data points over a spectral width of 5555.2 Hz The

transmitter frequency was centred on the water resonance

and conventional low power presaturation from the same

frequency synthesizer was applied during a 1.5-s

relaxa-tion delay to suppress the large water signal in the

TOCSY and NOESY spectra Gradient methods for

water suppression were used in the DQF-COSY spectrum

[41] The TOCSY spectrum was acquired with the pulse

sequence used by Griesinger et al., 1988 [42] which

minimizes cross relaxation effects, employing a 70-ms

MLEV-17 spin-lock NOESY spectra were acquired with

mixing times of 80, 150 and 250 ms

An HSQC experiment [43] was performed to assign the

a-13C resonances via correlations to their attached protons

The interpulse delay was set to 1/2JCH(3.6 ms

correspond-ing to JCH¼ 140 Hz) Two hundred and fifty-six t1

increments, each comprising 64 time averaged scans, were

acquired over 2048 data points and 5555.2 Hz in the directly

detected (1H, F2) dimension The spectral width in the13C

(F1) dimension was 24133 Hz An HMBC spectrum [44]

was collected to assign the carbonyl-13C resonances via

correlations through two and three bonds to a, b and NH

protons (with an interpulse delay of 1/2JCH¼ 62.5 ms for

JCH¼ 8 Hz) For this experiment, 400 t1increments, each

comprising 64 scans, were acquired over 4096 data points

and 5555.2 Hz in the1H (F2) dimension The spectral width for the13C (F1) dimension was 36216 Hz

All 2D NMR spectra were processed on a Sun Micro-systems Ultra Sparc 1/170 workstation usingVNMRsoftware (version 6.1 A) The data matrices were multiplied by a Gaussian function in both dimensions, then zero-filled to

2048 data points in F1 prior to Fourier transformation (4096 data points for the HMBC) Final processed 2D NMR matrices consisted of 2048· 2048 or 4096 · 4096 real points

Structural restraints Cross-peaks in the NOESY (mixing time¼ 250 ms) spec-trum were assigned using the programSPARKY(version 3.98) [45] The cross-peak volumes were converted to distance restraints using the method of Xu et al., 1995 [46] Briefly, in this procedure, the weakest and strongest peaks are calibra-ted at 5.0 and 1.8 A˚, respectively, in order to calculate intensity-dependent proportionality factors These factors were then used to determine the upper bound restraints for the remaining peaks To be conservative, the final restraints were increased by 10 percent from these calculated values All lower bound restraints were set to 1.8 A˚ For each symmetric pair of cross-peaks, the peak of smaller volume was used This procedure generated 264 distance restraints, including

115 intraresidue restraints, 52 sequential (i,i + 1) restraints and 65 medium range restraints (from 2–4 residues distant) Thirty-two additional restraints were ambiguous.3JNHCaH

values were measured from a 1D1H NMR spectrum, where the free induction decay had been multiplied by a sine-bell window function to enhance the resolution Dihedral angles were restrained as follows: 3JNHCaH65 Hz, /¼)60 ± 30; 5 <3JNHCaH66 Hz, /¼)60 ± 40 Where3JNHCaH> 6 Hz, phi angles were not restrained A total of 13 dihedral angle restraints were used in the structure calculations

Structural calculations Structures were generated on a Sun Microsystems Sparc 1/

170 workstation using X-PLOR software (version 3.851) [47,48] The restrained molecular dynamics (RMD) and

Table 3 Anticancer activites of citropin 1.1 and synthetic analogues (IC 50 values) Averaged concentration for a particular group of cancers, e.g 5 means 10)5M The number on the bottom line (total) indicates to how many human cancers (out of the test number of 60) that peptide is cytotoxic.

dynamical simulated annealing (SA) protocol was used [49],

which included the use of floating stereospecific assignments

[50] Sum-averaging was employed to take care of the

ambiguous restraints The all hydrogen distance geometry

(ALLHDG) force field (version 4.03) was employed for all

calculations [51] Initially, a family of 60 structures was

generated with random / and w dihedral angles These

structures were subjected to 6500 steps (19.5 ps) of high

temperature dynamics at 2000 K The Knoeand Krepelforce

constants were increased from 1000–5000 kcalÆmol)1Ænm)2

and 200–1000 kcalÆmol)1Ænm)4, respectively This was

followed by 2500 steps (7.5 ps) of cooling to 1000 K with

Krepelincreasing from 1000–40000 kcalÆmol)1Ænm)4and the

atomic radii decreased from 0.9 to 0.75 times those in the

ALLHDG parameter set The last step involved 1000 steps

(3 ps) of cooling from 1000–100 K Final structures were

subjected to 200 steps of conjugate gradient energy mini-mization The 20 structures produced with the lowest potential energies were selected for analysis 3D structures were displayed using INSIGHTII software (version 95.0, MSI) and the programMOLMOL[52]

Results Biological testing The antibiotic activities [as minimum inhibitory concentra-tion (MIC) values in lgÆmL)1] of two natural citropins (1.1 and 1.1.2) and 23 synthetic modifications of citropin 1.1, against nine pathogens, are listed in Table 2; summarized in Table 3 are the IC50values of the same peptides in in vitro anticancer tests against 60 human tumour lines as

Table 4 nNOS activities of citropin peptides, citropin synthetic modifications, and some related peptides IC 50 mean error ± 1.3% Citropin 1.1 modification 6 has a charge of zero, is hydrophobic,and shows minimal solubility in water thus testing was carried out in dimethyl sulfoxide as solvent, and is not reproducible Three tests gave IC 50 values of 29.6, 33.7 and 39.5 lgÆm L )1 , hence we give the IC 50 range as 30–40 lgÆm L )1 Qualitatively, this compound shows less nNOS inhibition than modifications 4 and 5 Modifications are shown in bold.

Relative molecular mass

IC 50

Hill

Modified peptide

Retro

Lesueurin

Modified peptide

determined by the National Cancer Institute The NCI lists

anticancer activities in molar concentrations and these are

the units used here In Table 3 (anticancer activities), the

numbers 5 and 6 refer to 10)5and 10)6M, respectively Ten

of the peptides failed the first stage of the anticancer testing

program and are specified as inactive: essentially this means

that no anticancer activity is noted at peptide concentrations

less than 1· 10)4M

Table 4 lists the data for nNOS inhibition by 32 peptides

Twenty-five of these peptides are citropin 1.1 and

synthe-tically modified analogues The other seven peptides are related to citropin 1.1, but have fewer residues These include lesueurin [17], dahleins 1.1 and 1.2 [53] and some synthetic modifications of lesueurin

The solution structure of citropin 1.1 synthetic modification (A4K14-citropin 1.1)

The solution structure of the basic peptide citropin 1.1, as determined by 2D NMR, is that of a well defined a-helical and amphipathic peptide [9] A number of synthetically modified citropin peptides have significantly greater anti-cancer and antibacterial activity (and also nNOS activity) than citropin 1.1 itself We have chosen to investigate the

7.40 7.60 7.80 8.00 8.20 8.40 8.60 8.80

7.40

7.60

7.80

8.00

8.20

8.40

8.60

8.80

A10 V12

V5

K8

I13

D4

I6 K7

F3

Sll V9

L2 G14

G15 F3

D4

V5

V9 K8

A10

I13

G14

F2

(ppm)

F1(ppm)

V12 Sll K7 I6

Fig 1 NH to NH region of the NOESY spectrum (mixing time ¼

250 ms) of A4K41-citropin 1.1 in 50% (v/v) d 3 -trifluoroethanol in water.

NOEs between sequential NH protons are indicated.

Fig 2 Summary of NOEs used in structure calculations for

A4K14-citropin 1.1 in 50% (v/v) d 3 -trifluoroethanol in water The thickness of

the bars indicates the relative strength of the NOEs (strong < 3.1 A˚,

medium 3.1–3.7 A˚ or weak > 3.7 A˚) Grey shaded boxes represent

NOEs that could not be assigned unambiguously The3J NHaCH values

obtained are also shown The error here is ± 0.5 Hz A cross-hatch (#)

indicates the coupling constant could not be determined reliably due to

overlap Due to overlap with the diagonal, the d NN (i,i + 1) NOE

between I6 and K7 could not be determined with certainty, and is not

included in this figure.

Fig 3 Deviation from random coil chemical shifts [59] (A) 1 H a-CH resonances, (B)13C a-C resonances, and (C)1H NH resonances Solid line, A4K14-citropin 1.1 (GLFAVIKKVASVIKGL-NH 2 ) Dotted line, citropin 1.1 (GLFDVIKKVASVIGGL-NH 2 ) A negative chemi-cal shift difference indicates an upfield chemichemi-cal shift compared to random coil, while a positive chemical shift difference indicates a downfield shift Deviation values for the a-CH resonances were smoothed over a window of n ¼ ±2 residues [60].

structure of one of the more active synthetic modifications

of citropin 1.1 – A4K14-citropin 1.1 (number 15 in

Tables 1–4) – by CD and NMR spectroscopy in order to

see whether there is any major difference between the

solution structure of this peptide and that of citropin 1.1

NMR spectroscopy

NMR experiments were performed on the synthetically

modified citropin analogue in which the Asp4 residue was

replaced with Ala and the Gly14 residue was replaced with

Lys (A4K14-citropin 1.1) NMR studies were performed

using a 50% d3-trifluoroethanol/H2O solution of

A4K14-citropin 1.1 as the parent peptide A4K14-citropin 1.1 has maximal

helicity in this solvent system, as judged by circular

dichroism [9] d3-Trifluoroethanol is widely thought of as

a helix-inducing solvent, however, So¨nnichsen et al., 1992

[54] found that for peptides in trifluoroethanol/H2O

solu-tions, helical structure was only observed where there was a

helical propensity in the sequence In addition, examples of

b-turn [55] and b-sheet [56,57] structures have been observed

in aqueous trifluoroethanol mixtures, demonstrating that

trifluoroethanol does not enforce helical structure but

merely enhances it if the propensity exists Thus

trifluoro-ethanol/H2O was deemed a suitable solvent system for

structural studies on the citropin 1.1 peptides The NMR

experiments were carried out at the same temperature as

that used for the experiments on citropin 1.1 [9] The

NMR sample of A4K14-citropin 1.1 had a pH of 4.1,

compared to pH 2.3 for citropin 1.1 The difference in pH

value was not expected to have an effect on the final

structures as both peptides were fully protonated at their

respective pH values

The 1H-NMR resonances were assigned using the

sequential assignment procedure of Wu¨thrich [58], which

involved the combined use of DQF-COSY, TOCSY and

NOESY spectra The a-13C resonances were assigned from

the one-bond correlations to the assigned a-1H resonances,

recorded in the HSQC spectrum Similarly, an HMBC

spectrum was employed to make the carbonyl-13C

assign-ments from the two- and three-bond correlations to the

assigned aH, bH and NH1H resonances Table 5 lists all

the assignments for the1H and a-13C resonances

A qualitative indication of the peptide structure can be

obtained from an examination of the observed NOEs and

chemical shifts The NH region of the A4K14-citropin 1.1

NOESY spectrum (mixing time¼ 250 ms), shown in Fig 1,

reveals a series of sequential NH–NH NOEs [dNN(i,i + 1)]

that occur along the length of the peptide A series of weaker

dNN(i,i + 2) NOEs can also be observed at a lower contour

level in this region The various types of NOEs observed for

A4K14-citropin 1.1 are summarized in Fig 2 Here it can

be seen that, in addition to the NOEs mentioned above, a

number of weak sequential daN(i,i + 1) NOEs occur as well

as a series of NOEs from residues three and four amino

acids apart [daN(i,i + 3), dab(i,i + 3) and daN(i,i + 4)]

Taken together, the observed NOEs and their intensities are

consistent with A4K14-citropin 1.1 having a helical

struc-ture along the majority of its sequence The pattern of NOE

connectivity is also similar to that found for the parent

peptide, citropin 1.1 [9] However, the patterns extend over

more residues for A4K14-citropin 1.1 This is particularly

noticeable for the daN(i,i + 1) NOEs that cease at residue

10 in citropin 1.1, but continue over the length of the peptide for A4K14-citropin 1.1 Similarly, the daN(i,i + 3) NOEs extend right up to residue 16 in A4K14-citropin 1.1 but stop at residue 14 for the parent peptide Thus, from an examination of the NOE data, it would seem the modified citropin peptide has the greater a-helical character beyond residue 10

A helical structure for A4K14-citropin 1.1 is also indica-ted from an examination of the deviation from random coil chemical shift values of the a-1H and a-13C resonances determined in water [58,59] Smoothed over a window of

n¼ ± 2 residues [60], the plot for the a-CH1H resonances shows a distinct upfield shift (Fig 3A), while those for the

13C resonances show a distinct downfield shift (Fig 3B) The directions of these deviations from random coil chemical shift values are consistent with the peptide having

a helical structure along its length, with maximal helicity in its central region and less well-defined structure at its N- and C-termini [61–63] For comparison, Fig 3A,B also show the deviations from random coil chemical shift for the1H and

13C a-CH resonances of citropin 1.1 [9] Both peptides have very similar plots over the central region of the peptide (from residues 4–10), i.e., where there is no difference in amino acid sequence between the two peptides and they both have the greatest helicity However, from approxi-mately residue Ala10 onwards, the 1H and 13C chemical shifts of A4K14-citropin 1.1 are consistently upfield and downfield, respectively, of those of the parent peptide These differences suggest that A4K14-citropin 1.1 forms a more stable a-helix than citropin 1.1 in the C-terminal region The small differences at the extreme N-terminal region (first three residues) for the plots of the 1H and 13C a-CH resonances are opposite in directional trend for structural conclusions to be drawn This may reflect the poorly defined nature of the first turn of the a-helix due to the lack of hydrogen bonds to their NH protons

Comparison of the observed NH chemical shifts of A4K14-citropin 1.1 with the corresponding random coil

NH chemical shifts [59] revealed a periodic distribution such that those from hydrophobic residues were shifted down-field with respect to the random coil values and those from hydrophilic residues were shifted upfield (Fig 3C) This behaviour is characteristic of amphipathic a-helices [64,65] and is due to differences in backbone hydrogen bond length

on either face of the peptide, which lead to slight curvature

of the helix The curvature may not be significant for A4K14-citropin 1.1, as it consists of only 16 residues, however, the periodic distribution of NH shifts is consistent with A4K14-citropin 1.1 forming an amphipathic a-helix Furthermore, Fig 3C also shows that the periodicity of the

NH chemical shifts is very similar between the parent and modified peptides

Structural analysis The conclusions derived from an examination of the NMR data were confirmed when the NOE data were used as input for structural calculations Sixty structures were generated by restrained molecular dynamics (RMD) and dynamical SA calculations and the 20 structures of lowest potential energy were selected for close examination

Table 6 Structural statistics of A4K14-citropin following RMD/SA calculations <SA> is the ensemble of the 20 final structures (SA) is the mean structure obtained by best-fitting and averaging the coordinates of backbone N, a-C and carbonyl-C atoms of the 20 final structures (SA) r is the representative structure obtained after restrained energy minimization of the mean structure Well-defined residues are those with angular order parameters (S) > 0.9 For A4K14-citropin 1.1, residues Leu2 to Gly15 are well-defined.

RMSD from mean geometry (A˚)

X - PLOR energies (kcalÆmol)1)

Table 5. 1H and13C NMR chemical shifts for A4K14-citropin in 50% trifluoroethanol in water (by volume), at a measured pH of 4.12 at 25 °C Data are shown in p.p.m Assignments for all the 1 H NMR resonances are shown whereas only the a- 13 C and carbonyl- 13 C resonances are presented;

NO, not observed.

Residue

Chemical shift of

13

CO

d-CH 3 0.99, 0.92

H3,5 7.31 H4 7.26

c-CH 3 0.96 d-CH 3 0.88

d-CH 2 1.68 e-CH 2 2.94

NH 3 +

n.o.

d-CH 2 1.81, 1.73 e-CH 2 2.96

NH 3 +

n.o.

c-CH 3 0.99 d-CH 3 0.88

d-CH 2 1.62 e-CH 2 3.04

NH 3+n.o.

d-CH 3 0.95 CONH 2 7.24, 6.77

Some statistics for the 20 final structures are given in

Table 6

The superimposition of the 20 structures over the

backbone N, aC and carbonyl-C atoms shows that

A4K14-citropin 1.1 forms a regular a-helix along its

entire length (Fig 4A) Analysis of the angular order

parameters (S, / and w) [66] of these structures indicated

that, except for the N- and C-terminal residues (Gly1

and Leu16), all residues were well defined (S > 0.9 for

both their / and w angles) A Ramachandran plot [67]

of the average / and w angles of the well-defined

residues reveals these angles are distributed within the

favoured region for a-helical structure (not shown) The

most energetically stable of the 20 final structures is

displayed in Fig 4B and from this representation it is

apparent that A4K14-citropin 1.1 forms an amphipathic

a-helical structure with well-defined hydrophobic and

hydrophilic faces

Discussion

Citropin 1.1 is the major wide-spectrum antibiotic peptide

in the secretion of the skin glands of L citropa [9] It is one

of the most potent membrane-active antibiotic peptides

isolated from amphibians and is particularly effective

against Gram-positive organisms [8] Citropin 1.1 is a 16

residue peptide and is one of a number of amphibian

antibiotic peptides containing the characteristic Lys7-Lys8

pattern: a group which includes lesueurin (from L lesueuri)

[17], the aureins (from L aurea and L raniformis) [11] and

the uperins (from toadlets of the genus Uperoleia) [68]

Citropin 1.1 does not cause lysis of red blood cells at a

concentration of 100 lgÆmL)1, but lysis is complete at

1 mgÆmL)1 (B C S Chia & J H Bowie, unpublished

results) Citropin 1.1 is thought to be stored in an inactive

form (spacer peptide – citropin 1.1) in the skin glands, but

when the frog is stressed, sick or attacked, an endoprotease

cleaves off the spacer peptide and the active citropin 1.1 is

released onto the skin Citropin 1.1 must be cytotoxic to the

frog as after about 10 min of exposure on the skin a further

endoprotease removes the first two residues of the peptide

destroying the antibiotic (and anticancer) activity [9]

The solution structure of citropin 1.1 is shown in Fig 5;

this should be compared with that of the synthetically

modified A4K14-citropin 1.1 depicted in Fig 4B The

NMR studies reported here indicate that both peptides

adopt amphipathic a-helices, but that the helicity is more

pronounced for A4K14-citropin 1.1 Each peptide has well

defined hydrophobic and hydrophilic regions However,

chemical shift and NOE connectivity data suggest that the

C-terminal region of the a-helix may be more stable in the

modified citropin This is due probably to the replacement

of Gly14 with Lys14 Gly is more conformationally mobile

than other residues, due to its lack of a side chain, and is a

well-known breaker of helical structure [69] The Lys residue

would therefore be expected to stabilize a helical structure in

this region In addition, the positively charged side-chain of

Lys would stabilize a C-terminal helix due to its

inter-action with the negative end of the helix dipole [69] The

replacement ofAsp4 with Ala4 does not have a significant

effect on the structure of the peptide This may be because

removal of the negatively charged Asp4, which would

stabilize the N–terminal helix by interaction with the positive end of the helix dipole [69], is compensated by the introduction of Ala, which has a high helical propensity Finally, we believe it is likely that all of the peptides listed in Tables 1 and 4 adopt such structures when interacting with either bacterial or cancer cell membranes

The antibiotic and anticancer activities of peptides of this type are due to the disruption of the cell membrane

by the peptide In order to span the lipid bilayer of bacterial and cancer cells, the peptide needs to have at least 20 amino acid residues [4,14,70] Citropin 1.1 has only 16 residues and thus is unable to fully span the lipid bilayer Amphipathic peptides of this type are thought

to operate via the carpet mechanism, which involves aggregation of the helical peptides on the surface of the membrane by interaction of the positively charged sites of the peptide with negatively charged sites on the membrane surface The peptides then insert into the lipid membrane, weakening the bilayer and making it susceptible to osmotic lysis [4,24,70] From the work reported herein, the greated helicity of A4K14-citropin 1.1 in its C-terminal region may be responsible for its enhanced antimicrobial activity

Antibacterial and anticancer activity Synthetic modifications of citropin 1.1, shown in Table 1, were made to investigate the relationship between activity and sequence The first point to be made is that the natural

L-citropin 1.1 has, within experimental error (± 1 dilution factor), the same spectrum of antibiotic activities as the synthetic allD-citropin 1.1 This is a feature of membrane active peptides [4,13] Other synthetic modifications were made to the following plan: (a) to successively replace the hydrophilic residues (to ascertain the effect of a particular hydrophilic residue on the bioactivity), and some hydro-phobic residues (certain hydrohydro-phobic residues, particularly terminal residues are often vital for good activity) with Ala, and (b) to change Gly and some hydrophilic residues to Lys (to determine the effect on activity of an increase in the positive charge of the peptide) The spectrum of antibiotic activities for each synthetic modification is recorded in Table 2 The following observations can be made Replace-ment of the following residues with Ala show (a) little change in activity for Asp4 and Ser11, and (b) significant reduction in the activity for Phe3, Lys7, Lys8 and Leu16; replacement of the following residues with Lys show (a) reduction in activity for Gly1 and Val12 and (b) significant increases in activity against Gram-negative organisms for Gly14 and Gly15 The conclusions from this study are that (a) modification of either of the terminal residues reduces the activity, and (b) the activity against Gram-positive organisms is not significantly improved (in comparison with citropin 1.1) by synthetic modification, but increasing the number of basic Lys residues in the hydrophilic zone of the amphipathic peptide markedly increases the activity against Gram-negative organisms like E coli

Apart from particular detail, the trends in anticancer activities of the modified citropins 1.1 mirror those outlined above for the antibiotic activities (Table 3) The citropin 1 peptides are generally cytotoxic toward the majority of the

60 cancers tested in the NCI regime: IC values are

generally in the moderate 10)5M range, with synthetic

modification 3 (Asp4 to Ala4) showing the strongest

cytotoxicity (in the 10)6M range) As was the case with

antibiotic activity, L- and D-citropins 1.1 show almost

identical activity

The trends observed for antibiotic activity are more

marked when considering anticancer activity For example,

some synthetic modifications which decrease antibiotic

activity, often destroy the anticancer activity, e.g., the

modifications Gly1 to Lys1, Phe3 to Ala3, Lys7 to Ala7,

Val12 to Lys12, and Leu16 to Ala16 The conclusions from

this study are, that for best anticancer activity of

citro-pin 1.1 type molecules, (a) the residues Gly1, Phe3, Ala4,

Lys7 and Leu16 are essential, and (b) the charge needs to be

P+2 The close correlation between the broad-spectrum

anticancer and antibacterial activity of membrane active peptides, suggests that the anticancer activity is also due to penetration and disruption of the membranes of the cancer cells The selectivity of these peptides for cancer over normal cells may be due to the significantly higher levels of anionic phospholipids present in the outer leaflet of cancer cells [71–74]

nNOS activity

We have already reported that citropin 1.1 causes the inhi-bition of nNOS by forming a complex with the regulatory protein Ca2+-calmodulin, thus impeding the attachment of this enzyme at the calmodulin binding site on nNOS [17] The actual nature of the complex is not known, but NMR

GLY1

LEU16

H3N+

+NH

3

H3N+ OH

O

NH2

+NH

3

A

B

Fig 4 Most stable structures of A4K14-citroprin 1.1 (A) Superimposition of the 20 most stable structures of A4K14-citropin 1.1 along the backbone atoms (N, a-C and carbonyl C) (prepared with the program MOLMOL [52]) and (B) the most stable calculated structure of A4K14-citropin 1.1 A ribbon is drawn along the peptide backbone in (B).

H3N+

+NH 3

OH

O

H2N

H3N+

-O O

Fig 5 The most stable calculated structure of citropin 1.1 This figure was originally published by Wegener et al [9] in Eur J Biochem 265, 627–635.