Peptide Antimicrobial Agents - Clinical Microbiology Reviews

21 Pages·2006·0.49 MB·English

CLINICALMICROBIOLOGYREVIEWS,July2006,p.491–511 Vol.19,No.3
0893-8512/06/$08.00(cid:1)0 doi:10.1128/CMR.00056-05
Copyright©2006,AmericanSocietyforMicrobiology.AllRightsReserved.
Peptide Antimicrobial Agents
Håvard Jenssen, Pamela Hamill, and Robert E. W. Hancock*
CentreforMicrobialDiseasesandImmunityResearch,UniversityofBritishColumbia,LowerMallResearchStation,
232-2259LowerMall,Vancouver,BritishColumbiaV6T1Z4,Canada
INTRODUCTION.......................................................................................................................................................491
NATURALDISTRIBUTIONANDACTIVITIESOFANTIMICROBIALHOSTDEFENSEPEPTIDES.........492
ANTIVIRALACTIVITY.............................................................................................................................................494
StructuralRequirementsforAntiviralPeptides................................................................................................495
ModeofActionofAntiviralPeptides...................................................................................................................496 D
Blockingofviralentrybyheparansulfateinteraction.................................................................................496 o
w
(i)Blockingofcell-to-cellspread.................................................................................................................497
n
Blockingofviralentrybyinteractionwithspecificcellularreceptors.......................................................497 lo
Blockingofviralentrybyinteractionwithviralglycoproteins....................................................................497 a
Membraneorviralenvelopeinteraction.........................................................................................................497 d
e
(i)Viralenvelopeinteraction........................................................................................................................497 d
(ii)Cellularmembraneinteraction..............................................................................................................497 f
r
Intracellulartargetsandhostcellstimulation...........................................................................................................498 o
m
ANTIBACTERIALACTIVITY...................................................................................................................................498

StructuralRequirementsforAntibacterialPeptides.........................................................................................498 h
t
ModeofActionofAntibacterialPeptides...........................................................................................................499 tp
Membrane-permeabilizingpeptides.................................................................................................................500 :/
/
Peptidesthatdonotactbymembranepermeabilization.............................................................................500 c
m
ANTIFUNGALACTIVITY.........................................................................................................................................502
r
StructuralRequirementsforAntifungalPeptides.............................................................................................502 .a
ModeofActionofAntifungalPeptides................................................................................................................503 s
m
ANTIPARASITICACTIVITY....................................................................................................................................503
.
DEVELOPMENTOFANTIMICROBIALPEPTIDESFORCLINICALAPPLICATIONS.............................503 o
r
CONCLUSION............................................................................................................................................................505 g
/
ACKNOWLEDGMENTS...........................................................................................................................................505 o
REFERENCES............................................................................................................................................................505 n

N
o
v
INTRODUCTION proteases,polyvalentanionssuchasglycosaminoglycans(e.g., e
heparan sulfate), and low local peptide concentrations. Con- m
Awidevarietyoforganismsproduceantimicrobialpeptides b
verselythesepeptidesareimportanteffectormoleculesofthe
as part of their first line of defense (90). Antimicrobial pep- e
innate immune system (24, 30). They are able to enhance r
tidesaretypicallyrelativelyshort(12to100aminoacids),are 2
positivelycharged(netchargeof(cid:1)2to(cid:1)9),areamphiphilic, psehpatgicoceyffteocsitss,osftiLmPuSla,tperopmroostteagrleacnrduiintmreenletaasne,dnaecucutrmaluizleatitohne 5,
andhavebeenisolatedfromsingle-celledmicroorganisms,in- 2
of various immune cells at inflammatory sites (58, 266), 0
sects and other invertebrates, plants, amphibians, birds, fish, 1
promote angiogenesis (129), and induce wound repair (36). 8
andmammals,includinghumans(159,257).Todate,hundreds
Peptidesofmammalianoriginhavealsobeendemonstrated b
of such peptides have been identified (88), indicating their to have an active role in the transition to the adaptive y
importanceintheinnateimmunesystem(89).Theexpression g
immune response by being chemotactic for human mono- u
of these antimicrobial peptides can be constitutive or can be
cytes (246) and T cells (40) and by exhibiting adjuvant and e
inducible by infectious and/or inflammatory stimuli, such as s
polarizing effects in influencing dendritic cell development t
proinflammatory cytokines, bacteria, or bacterial molecules
(47).Althoughsuchpeptidesmayhaveadirecteffectonthe
that induce innate immunity, e.g., lipopolysaccharides (LPS)
microbe,suchasbydamagingordestabilizingthebacterial,
(44,86).Someofthesepeptidesarepotentantimicrobials.In
viral, or fungal membrane or acting on other targets, they
contrast, the direct antimicrobial activity of others is largely
appear to be broadly involved in the orchestration of the
evident in dilute media, and direct microbe killing is almost
innateimmuneandinflammatoryresponses(89).Thus,they
certainlypreventedbyphysiologicalconditions,includinghigh
are increasingly being referred to as host defense peptides.
monovalent or moderate divalent cation concentrations, host
Forexample(cid:2)-defensinsarealmostcertainlybactericidalat
the high (mg/ml) concentrations found in neutrophil gran-
ules, but they probably act primarily as immunomodulators
*Correspondingauthor.Mailingaddress:CentreforMicrobialDis- at the lower concentrations released by degranulation at
easesandImmunityResearch,UniversityofBritishColumbia,Lower
inflammatory sites.
MallResearchStation,232-2259LowerMall,Vancouver,BritishCo-
lumbiaV6T1Z4,Canada.Phone:[phone].Fax:[phone]. Despitetheirsimilargeneralphysicalproperties,individual
E-mail:[email protected]. cationicpeptideshaveverylimitedsequencehomologiesanda
491
492 JENSSEN ET AL. CLIN.MICROBIOL.REV.
D
o
w
n
lo
a
d
e
d

f
r
o
m

h
t
t
p
:
/
/
c
m
r
.
a
s
FIG. 1. Structuralclassesofantimicrobialpeptides.(A)Mixedstructureofhuman(cid:3)-defensin-2(PDBcode1FQQ)(216);(B)loopedthanatin m
(PDB code 8TFV) (156); (C) (cid:3)-sheeted polyphemusin (PDB code 1RKK) (202); (D) rabbit kidney defensin-1 (PDB code 1EWS) (165); .o
(E) (cid:2)-helical magainin-2 (PDB code 2MAG) (76); (F) extended indolicidin (PDB code 1G89) (212). The disulfide bonds are indicated in r
g
yellow,andtheillustrationshavebeenpreparedwithuseofthegraphicprogramMolMol2K.1(132). /

o
n

N
o
v
wide range of secondary structures with at least four major NATURALDISTRIBUTIONANDACTIVITIESOF e
themes. The most prominent structures are amphiphilic pep- ANTIMICROBIALHOSTDEFENSEPEPTIDES m
tides with two to four (cid:3)-strands, amphipathic (cid:2)-helices, loop b
e
structures, and extended structures (21, 87) (Fig. 1; Table 1). Antimicrobialpeptidesareauniversalfeatureofthedefense r
2
Thisreviewprovidesanoverviewofthe(direct)antimicrobial systemsofvirtuallyallformsoflife,withrepresentativesfound 5
functions of these peptides, with an emphasis on antiviral ac- inorganismsrangingfrombacteriatoplantsandinvertebrate ,
2
tivity and an update on antibacterial, antifungal, and antipar- andvertebratespecies,includingmammals.Theyformpartof 0
1
asiticactivities. the ancient, nonspecific innate immune system, which is the 8

b
y

g
u
e
TABLE 1. Someexamplesofthediverseprimarysequencecompositionsofantimicrobialpeptides s
t
Peptide Primaryaminoacidsequencea Reference(s)
Rabbitkidneydefensin MPCSC KKYCDPWEVIDGSCGLFNSKYIC CREK 165
1 2 3 2 3 1
Human(cid:3)-defensin-2 GIGDPVTCLKSGAICHPVFCPRRYKQIGTCGLPGTKC C KKP 216
1 2 3 2 1 3
Magainin2 GIGKFLHSAKKFGKAFVGEIMNS 76
Indolicidin ILPWKWPWWPWRR 212
Polyphemusin1 RRWCFRVCYRGFCYRKCR 202
1 2 2 1
Thanatin GSKKPVPIIYCNRRTGKCQRM 156
1 1
BuforinII TRSSRAGLQFPVGRVHRLLRK 187
CecropinA1 GWLKKIGKKIERVGQHTRDATIQGLGVAQQAANVAATAR 205
Melittin GIGAVLKVLTTGLPALISWIKRKRQQ 84
Humanlactoferricin GRRRRSVQWCAVSQPEATKC FQWQRNMRRVRGPPVSCIKRDSPIQCIQA 17,107,118
1 2 2 1
Bovinelactoferricin FKCRRWQWRMKKLGAPSITCVRRAFA 17,107,118
1 1
LL-37 LLGDFFRKSKEKIGKEFKRIVQRIKDFLRNLVPRTES 81
aAminoacidsequencesaregiveninone-lettercode.Cysteinesformingdisulfidebondsarenumberedwithsubscriptstoindicatetheirpairings.Boldfaceindicates
cationicaminoacidresidues.
VOL.19,2006 ANTIMICROBIAL HOST DEFENSE PEPTIDES 493
principal defense system for the majority of living organisms. peptidesanditsvirulence.Sofar,onlypeptideswitha(cid:3)-sheet
In many cases, their primary role is in the killing of invading globularstructurehavebeenidentifiedinplants,withthetwo
pathogenic organisms, and this is the focus of this review; major and best-studied groups being thionins and defensins
however,itisincreasinglyrecognizedthattheymayalsofunc- (reviewedinreference72).Physiologicallyrelevantconcentra-
tion as modulators of the innate immune response in higher tionsofthioninsareactiveagainstbacteriaandfungiinvitro,
organisms(23,220,265,271).Collectively,theydisplaydirect and studies utilizing transgenic plants have shown that heter-
microbicidal activities toward bacteria, fungi, and some para- ologous expression of thionins can confer protection against
sitesandviruses,althoughtheimportanceoftheseactivitiesin bacterialchallenge(35,59).Plantdefensinsdisplayantibacte-
contributing to host defense may vary between different sites rial and antifungal activities in vitro (245). Consistent with a
within a particular organism and also between different types defensive role, they are found in leaves, flowers, seeds, and
oforganisms.Antimicrobialpeptidesmaybeexpressedconsti- tubers.
tutively in some cases or may be inducibly expressed in re- Sinceinvertebrateslacktheadaptiveimmunesystemfound
sponsetopathogenicchallenge.Inmulticellularanimals,they invertebratespecies,theyarereliantsolelyupontheirinnate
D
may be expressed systemically (for example, in insect hemo- immune systems to counteract invading pathogens. Consider- o
lymphorvertebrateimmunecells)and/orlocalizedtospecific ing the extraordinary evolutionary success of this group of w
n
cell or tissue types in the body most susceptible to infection, organisms,itisevidentthatinvertebrateinnateimmunemech- lo
suchasmucosalepitheliaandtheskin.Thefollowingisabrief anisms are extremely effective. This has prompted intense a
d
overviewofthedistributionofantimicrobialpeptidesinnature studies of invertebrate species such as the arthropod fruit fly,
e
andtheirrolesindefense. Drosophilamelanogaster,whichhasbecomeamodelsystemfor d

Antimicrobial peptides produced by bacteria were among the study of innate immunity and has led to the discovery of fr
o
thefirsttobeisolatedandcharacterized(163).Whiletheydo immunesystemstrategies,suchaspathogenrecognitionrecep- m
not protect against infection in the classical sense, they con- tors (Toll-like receptors), that are conserved in higher organ-
h
tribute to survival of individual bacterial cells by killing other isms, including mammals. Numerous antimicrobial peptides tt
p
bacteriathatmightcompetefornutrientsinthesameenviron- havenowbeenidentifiedininvertebrates,andtheyarerecog- :
/
ment. Bacterial antimicrobial peptides, also called bacterio- nized as playing a key role in protection from pathogenic /c
m
cins, are thought to be produced by many or most bacteria organisms. Indeed, the role of antimicrobial peptides and the
r
(128, 206) and are generally extremely potent compared with regulationoftheirexpression,includingthesignalingcascades .a
most of their eukaryotic counterparts. Their activities may be involved, is well understood for Drosophila (111). Antimicro- s
m
eithernarroworbroadspectrum,capableoftargetingbacteria bialpeptidesarefoundinthehemolymph(plasmaandhemo- .
o
within the same species or from different genera. The bacte- cytes), in phagocytic cells, and in certain epithelial cells of r
g
riocinsconstituteastructurallydiversegroupofpeptides,and invertebrates.Theycanbeexpressedconstitutively,forexam- /

o
itwasrecentlyproposedthattheybeclassifiedintotwobroad ple, in the hemocytes of marine arthropods such as shrimp,
n
categories: lanthionine containing (lantibiotics) and non-lan- oyster,andhorseshoecrab(11,114),orinducedinresponseto N
thioninecontaining(43).Lantibioticsarecharacterizedbythe pathogenrecognition,suchasantifungalpeptidesinDrosoph- o
v
inclusion of the unusual amino acid lanthionine and the ne- ila (149). Among some of the prototypic invertebrate antimi- e
cessity for posttranslational processing to acquire their active crobial peptides are the (cid:2)-helical cecropins (fly hemolymph) m
forms. The most extensively studied lantibiotic is nisin, pro- andmelittin(beevenom)aswellasthe(cid:3)-hairpin-likepeptides b
e
duced by Lactococcus lactis, which has been commonly used tachyplesin and polyphemusin (horseshoe crab). The horse- r
2
for nearly 50 years as a food preservative without significant shoe crab-derived peptides possess some of the most potent 5
developmentofresistance.Itisalsoextremelypotent,display- antibacterial and antifungal activities observed, with MICs of ,
2
ingactivityagainstavarietyofgram-positivebacteriaatMICs (cid:4)2 (cid:5)g/ml (280). Interestingly, polyphemusin also displays ac- 0
1
inthelownanomolarrange.Thesepropertieshaveprompted tivity against human immunodeficiency virus (HIV) (160). 8
intense study of the mechanism of action of nisin, which is However, the most abundant group of antimicrobial peptides b
y
discussed later in this review. Other lantibiotics have also re- ininvertebratesarethedefensins,whichareopen-endedcyclic
g
ceivedattentionduetotheirpossibleapplicationsinthetreat- peptideswiththreeorfourdisulfidebridges.Theactivitiesof u
ment of bacterial species which have developed antibiotic re- invertebrate defensins can be divided according to whether e
s
sistance.Mersacidin,atetracyclicpeptidethatisproducedby theirprincipalbiologicalactivityisdirectedtowardbacteriaor t
Bacillusspp.(38,39),displaysbactericidalactivityagainstmeth- fungi(33).
icillin-resistant Staphylococcus aureus that is comparable to Antimicrobialpeptideshavebeenisolatedfromawiderange
that of vancomycin, but without the development of cross- of vertebrate species, including fish, amphibians, and mam-
resistance(135). mals, indicating that, even in the presence of an adaptive im-
In plants, it is widely believed that antimicrobial peptides muneresponse,thesepeptideshaveanimportantroleinhost
play an important and fundamental role in defense against defense. Direct microbicidal activity is associated with verte-
infection by bacteria and fungi. Observations to support this brate antimicrobial peptides to various degrees under physio-
role include the presence and expression of genes encoding logical conditions, and these activities likely contribute to the
antimicrobialpeptidesinawidevarietyofplantspeciesinves- first line of defense, especially where they are found in very
tigatedthusfar,demonstrationsoftheirbactericidalandfun- highconcentrations,suchasinthegranulesofphagocyticcells
gicidal activity in vitro, and correlations between expression or the crypts of the small intestine (23, 25, 220, 265, 266).
levelsofpeptidesandsusceptibilitytoagivenpathogenorthe However,itisincreasinglyrecognizedthatinadditiontodirect
extentofresistanceofaparticularbacteriumtoplant-derived microbicidal activity, small cationic peptides perform critical
494 JENSSEN ET AL. CLIN.MICROBIOL.REV.
immunomodulatoryfunctionsandmaybeinvolvedinthecon- skin(54),aswellasconditionsinwhichadeficiencyofLL-37
trolofinflammation,whichservestorecruitavarietyofother leadstochronicperiodontaldisease(204).Inadditiontodirect
microbicidalmechanisms(24,265,266).Consistentwiththeir microbicidal activity, LL-37 has important additional roles in
roleindirectandindirectantimicrobialdefenses,antimicrobial hostdefense,includingchemotacticpropertiesandmodulation
peptides in vertebrates are found at sites that routinely en- ofinflammatoryresponses(24,271).
counter pathogens, such as mucosal surfaces and the skin, as A second prominent group of mammalian antimicrobial
wellaswithinthegranulesofimmunecells(24,265,266). peptides is the defensins (70, 221), cyclic peptides which are
Amphibian skin glands have proven to be a rich source of categorizedintothreesubfamiliesonthebasisofthedisulfide
antimicrobial peptides, with approximately 500 having been pairingsbetweentheirsixconservedcysteineresidues((cid:2)-and
described to date as originating from this source. This repre- (cid:3)-defensins)ortheirmacrocyclicnature((cid:7)-defensins).Aswith
sentsalargeproportionofthetotalnumberofreportedanti- cathelicidins, vertebrate defensins are synthesized as prepep-
microbialpeptides(207;http://www.bbcm.univ.trieste.it/(cid:6)tossi tides which require proteolytic processing to their active pep-
/pag1.html). The (cid:2)-helical magainins (272) are the prototypic tide forms. The (cid:2)- and (cid:3)-defensins are widely distributed in
D
amphibianantimicrobialpeptides,withstrongmembrane-per- vertebratespecies,whereas(cid:7)-defensinshavesofarbeeniden- o
meabilizing activity towards gram-positive and -negative bac- tifiedonlyinOldWorldmonkeysandapparentlyonlyinneu- w
n
teria, fungi, yeasts, and viruses. Structure-function relation- trophilsandmonocytessofar(244).Dependingonthespecies, lo
ships and the mechanism of action of magainin have been (cid:2)- and (cid:3)-defensins are found in the granules of neutrophils, a
d
extensively studied, and these peptides have subsequently macrophages, NK cells, intestinal Paneth cells and epithelial
e
served as the template for development of the first (although tissues,theskin,certainmucosalsurfacessuchastherespira- d

ultimately unsuccessful) clinical antibacterial peptide treat- tory passage and urinogenital tract, and many bodily fluids. fr
o
ment (74, 137). The broad antibacterial and antifungal activi- Expression of the defensins may be constitutive, such as for m
tiesofdermaseptins,isolatedfromtheskinofSouthAmerican human (cid:3)-defensin-1 (hBD-1) in most tissues, or inducible,
h
frogs,havealsobeenwidelystudied.Inadditiontotheirpres- such as for hBD-2, the expression of which in monocytes is tt
p
ence in the skin, amphibian antimicrobial peptides are pro- upregulatedfollowingexposuretobacteriaorLPS(56,62).In :
/
ducedinthemucosaofthestomach,indicatingaroleinpro- vitro studies demonstrate that collectively, defensins possess /c
m
tection from ingested pathogens. The best-characterized generally weak microbicidal activities towards bacteria, fungi,
r
examples are the Asian toad peptides buforin and buforin II, and some viruses. While the bactericidal activities of most (cid:2)- .a
which are generated by cleavage of the nucleosome protein and(cid:3)-defensinsareantagonizedbyincreasingsaltconcentra- s
m
histone 2A. A number of excellent reviews have covered this tions(e.g.,100mMmonovalentand/or2mMdivalentcations, .
o
largegroupofantimicrobialpeptides(33,207,224). whicharefoundatmanybodysites),thehighconcentrationsof r
g
Cathelicidins are a large and diverse group of vertebrate (cid:2)-defensins that are found in some locations, particularly in /

o
antimicrobial peptides. They are characterized by a well con- the granules in phagocytic cells and in intestinal crypts, are
n
served N-terminal segment (the cathelin domain) that is pro- thoughttobesufficienttoresultinkillingdespiteantagonism N
teolyticallycleavedtogeneratethemature,activepeptidecon- by salts (10, 71). (cid:7)-Defensins and hBD-3, on the other hand, o
v
tained within the C terminus. Hence, most cathelicidins are retain their bactericidal activities in physiological salt condi- e
stored in an inactive propeptide state, mostly within granules tionsandalsodisplayantiviralactivityininvitrostudiesusing m
b
ofcirculatingimmunecells.Neutrophilsecretorygranulesare HIV (42, 113). Transgenic and knockout experiments with
e
thepredominantsourceofcathelicidins,buttheymayalsobe micehaveindicatedacriticalrolefordefensinsinhostdefense. r
expressedinmucosalsurfacesinthemouth,lung,andgenito- MMP-7-nullmice,whichlackallmature(cid:2)-defensinsduetothe 25
urinary tract and in skin keratinocytes in inflammatory disor- lossoftheproteaserequiredforproteolyticcleavage,displaya ,
2
ders, as is the case with human cathelicidin LL-37 (hCAP18) reduced clearance of Escherichia coli and higher mortality 0
1
(66).BeyondthecommonNterminus,thestructureofmature rates upon challenge with Salmonella enterica serovar Typhi- 8
cathelicidins is diverse, with (cid:2)-helical, (cid:3)-hairpin, and proline/ murium, indicating a important role in intestinal immunity b
y
arginine-richpeptidesallrepresented.Thestructuraldiversity (258). Conversely, knock-in of human HD-5 defensin into
g
withinthecathelicidinfamilyisalsoindicativeoftheirappar- mousePanethcellsconferredimmunitytooralchallengeusing u
entlydistinctfunctions,andtheyexhibitadiversespectrumof Salmonella enterica serovar Typhimurium (214). Importantly, e
s
microbicidal and immunomodulatory activities. Cathelicidins in each case a correlation was observed between the antibac- t
have been isolated from many mammalian species, such as terial activity of the altered Paneth cell products in vitro and
mice, rabbits, sheep, horses, and humans. In some mammals, theirprotectiveabilitiesinvivo.
such as cattle, multiple cathelicidins are found in the body,
indicating that they likely perform varied biological roles in
ANTIVIRALACTIVITY
hostdefense.Oneofthebestcharacterizedbovineantimicro-
bial peptides is BMAP-28, an (cid:2)-helical peptide which rapidly Representatives from all four structural classes of the cat-
permeabilizesthemembranesofabroadspectrumofbacteria ionic host defense peptides have shown the ability to inhibit
and fungi at moderate concentrations in vitro (229), whereas viral infection. The spectra of viruses that are affected com-
the proline-rich bovine peptide Bac 5 shows selectivity for prise primarily enveloped RNA and DNA viruses, with the
gram-negative bacteria under the same conditions (75). In exceptionofnonenvelopedadenovirus(15,102),felinecalici-
contrast, only one cathelicidin is expressed in humans: LL-37 virus (164), and echovirus 6 (199). The antiviral activity of
(hCAP18). Evidence in support of an early defensive role for antimicrobialpeptidesoftenappearstoberelatedtotheviral
LL-37includesitsupregulationinresponsetoinfectioninthe adsorptionandentryprocess(16)orisaresultofadirecteffect
VOL.19,2006 ANTIMICROBIAL HOST DEFENSE PEPTIDES 495
TABLE 2. Selectedexamplesofantiviralpeptides
Peptide Structure Source(s) Virus Proposedantiviralmechanism Reference(s)
Magainin (cid:2)-Helix Frog HSVa Cellulartarget 1,3
HIV Suppressesviralgeneexpression
Cecropin (cid:2)-Helix Insect Juninvirus Suppressesviralproteinsynthesis 3,255
HSV Cellulartarget
HIV Suppressesviralgeneexpression
Mellitin (cid:2)-Helix Bee HSV Cellulartarget 3,255,269
Juninvirus Cellulartarget
LL-37 (cid:2)-Helix Human HSV Weakviralinactivation 269
Brevinin-1 (cid:2)-Helix Frog HSV Viralinactivation 269 D
o
(cid:7)-Defensin Cyclic(cid:3)-sheet Primate,human HIV Bindsglycosylatedgp120 42,270 w
n
HSV BindsgBandblocksviralattachment lo
a
Defensin (cid:3)-Sheet Human,rabbit HSV InteractswithHSVmembrane/glycoproteinand 15,45,80,178, d
cellulartargetbutnotheparansulfate 225,270 e
d
IAV Inactivatesviralparticle
f
HCMV Inactivatesviralparticle r
o
VSV Inactivatesviralparticle m
HIV Cellulartarget
h
Adenovirus Unknown
t
t
p
Dermaseptin (cid:3)-Sheet Frog HIV Viralmembranedisruption 16,153 :/
/
HSV Activityatvirus-cellinterface c
m
Tachyplesin (cid:3)-Sheet Horseshoecrab HIV Virus-cellfusion 171,174,269 r.
a
HSV Viralinactivation s
VSV Viralenvelope m
IAV Viralenvelope .
o
r
Protegrin (cid:3)-Sheet Human,porcine HIV Unknown 236,269 g
/
HSV Viralinactivation o
n
Polyphemusin (cid:3)-Sheet Horseshoecrab HIV Bindsgp120andCD4 177,242 N
o
Lactoferricin (cid:3)-Turn Bovine,human HCMV Activityatvirus-cellinterface 4,6,55,120 ve
HIV Unknown m
HSV Blocksheparansulfate,butasecondaryeffect b
hasalsobeenindicated e
r
Papillomavirus Activityatvirus-cellinterface
2
5
Indolicidin Extended Bovine HIV Inhibitintegrase 3,134,208 ,
HSV Targetsviralmembrane/glycoprotein 2
0
1
aHSVindicateseitherHSV-1orHSV-2orbothtypes. 8

b
y

g
u
on the viral envelope (1, 208). However, it appears to be StructuralRequirementsforAntiviralPeptides e
s
impossibletopredictantiviralactivitybasedprimarilyonsec- t
ondarystructuresofpeptides(Table2).Forexample(cid:2)-helical Syntheticanaloguesofseveralnaturallyoccurringantimicro-
peptides such as cecropins, clavanins, and the cathelicidin bialpeptideshavebeenmadeinanattempttoidentifyimpor-
LL-37havebeenshowntocauseminimalornoherpessimplex tant structural features contributing to the antiviral activity.
virus(HSV)inactivation(18,185,269),while(cid:2)-helicalmagai- Differentstrategiesfordesignofsuchpeptideshavebeenpur-
nins(Fig.1),dermaseptin,andmelittinhaveshownquitepo- sued.Severalgroupshavelookedattheimportanceofcharged
tent anti-HSV activity (Table 2) (1, 3, 16, 269). Conversely, and aromatic amino acids, since antiviral peptides are often
(cid:3)-sheetpeptidessuchasdefensins,tachyplesin,andprotegrins highlycationicandamphiphilic(45,119,241,269).Thehydro-
aswellasthe(cid:3)-turnpeptidelactoferricinhaveallshownhigh phobic character of the peptides has been investigated for a
activitytowardsHSV(Table2)(4,45,120,148,225,269,270). hybridpeptideofcecropinAandmagainin-2(144),whilethe
Itshouldbenotedthatwithinthedifferentpeptidesubclasses, substitutionofD-orL-aminoacidshasbeenstudiedonasetof
activity may vary considerably. For example, protegrin ana- (cid:7)-defensins(270).
logues lacking one or both disulfide bridges vary from highly Thecreationofaseriesoflactoferricinanaloguesandstudy
activetoinactiveagainstHSVinfections(269). oftheiractivitytowardsHSVrevealedarelationshipbetween
496 JENSSEN ET AL. CLIN.MICROBIOL.REV.
the peptide net charge and its antiviral activity (120, 121). ferent viral infections varies considerably. It has been
However, the spatial positioning of the charged amino acids demonstrated that recombinant cells lacking heparan sulfate
seemed to be more important for antiviral activity than the and chondroitin sulfate expression demonstrate an 80% and
actual net charge (120). For lactoferricin the nature of the 60%reduction,respectively,insusceptibilitytoHSVinfection
aromatic amino acid appeared to be of minor importance for (158).Enzymaticremovalofcellularheparansulfateandchon-
theantiviralactivity,althoughitscontributiontothesecondary droitinsulfatehasledtotheobservationthattheseproteogly-
structure and thereby presentation of the charged residues can molecules have minor influences on HIV attachment to
might be crucial (121). Detailed studies on the influence of hostcells.However,ithasbeendemonstratedthattheyareof
secondary structure domains on anti-HSV activity illustrated major importance for HIV entry and replication (8). In con-
that the (cid:2)-helicity of a peptide could not explain its antiviral trast,onlyhighlysulfatedheparanparticipatesintheentryof
activity (122), thus implying that the presentation of the hepatitisCvirus(14).Inhibitorsofheparinsulfatebiosynthe-
charged residues is of greatest importance with respect to sis,suchasheparin,heparinaseItreatment,andsodiumchlo-
anti-HSVactivity(119).Thisisinaccordancewithresultsfrom rate,alldemonstratetheabilitytoinhibithumancytomegalo-
D
Giansantietal.(77)inastudyonpeptidesderivedfrombovine virus(HCMV)infectioninadose-dependentmanner(231).It o
lactoferrin and hen ovotransferrin. They concluded that the has even been indicated that naked coxsackievirus B3 makes w
n
presence of hydrophobic and positively charged residues is use of a specific modified heparan sulfate molecule for viral lo
critical but not sufficient for antiviral activity, and this may entry(274). a
d
relatetodifferentconformationsadoptedbythesepeptidesin One might hypothesize that antimicrobial peptides that in-
e
thecontextofthenativeprotein(77). teract with heparan sulfate should be able to block many dif- d

(cid:7)-DefensinsareratherrigidcyclicpeptidesfromOldWorld ferentviralinfections.Thelargenumberofpositivelycharged fr
o
monkeys. Analogues have been designed with a focus on Ile- residues in antimicrobial peptides enables them to interact m
to-TyrorArg-to-Tyrsubstitutions,inadditiontotheextensive electrostaticallywithnegativelychargedcellsurfacemolecules,
h
use of D-amino acids, and have demonstrated the importance includingheparansulfate.Human(cid:2)-defensin,LL-37,andma- tt
p
of the charge and spatial conformation of the peptides (270). gainin have all been shown to interact with different glycos- :
/
Similarly,definedstructurescanprovideeffectiveantiviralsin aminoglycan molecules (116, 217, 218). Specific glycosamino- /c
m
the case of other types of peptides. Lactoferricin and glycanbindingdomainshavealsobeenidentifiedinbovineand
r
polyphemusin have (cid:3) structures stabilized by one and two humanlactoferricin.Thesedomainsinvolvethesequenceele- .a
internaldisulfidebridges,respectively.Thesedisulfidebridges ments G RRRRS and R KVR in human lactoferricin s
1 6 28 31 m
have been shown to be crucial for the antiviral activity of the (157) and the entire sequence of bovine lactoferricin (223). .
o
peptides(6,120,241)(Fig.1;Tables1and2). Thesequenceandstructuraldiversityintheseglycosaminogly- r
g
Despitetheirdiversestructures,manypeptidespossessanal- can binding peptides suggests that the critical factor driving /

o
ogous antiviral modes of action (119, 120), indicating that interaction is how charged residues are presented in the sec-
n
these peptides are able to interact with their targets, despite ondarystructure.Severallactoferricinanaloguesandsynthetic N
largestructuraldifferences.Apossibleexplanationwouldliein (cid:2)-helical peptides have been made in an attempt to better o
v
the observation that antimicrobial host defense peptides are understand these interactions. The results illustrate that the e
known to adopt amphipathic conformations that are intrinsic affinityforheparansulfateisonlypartlycorrelatedwiththenet m
b
to antibacterial activity and, we propose here, to antiviral ac- chargeofthepeptides(119,120).Forexample,analogueswith
e
tivity.Interestingly,althoughtheviraltargetofthesepeptides Arg residues appear to promote a higher glycosaminoglycan r
2
appears to vary, the demonstrated antiviral effects are quite affinity than comparable analogues substituted with Lys (67, 5
similar. 99,119,237). ,
2
Ithasbeendemonstratedthatlactoferricinandasetofshort 0
ModeofActionofAntiviralPeptides (cid:2)-helicalpeptidesareabletoblockHSVinfectionbybinding 18
to heparan sulfate in a way similar to that demonstrated by b
Blockingofviralentrybyheparansulfateinteraction.Pro- lactoferrin (5, 119, 120). This is supported by the fact that y
g
teoglycans are found in all types of tissue, in intracellular mixing the peptides with HSV prior to interaction did not u
granulesecretions(130),inextracellularmatrix(112),andon increaseantiviralactivity(5).Thepeptidesexhibiteddifferent e
s
the cell surface (19). Proteoglycans consist of a core protein antiviral effects for HSV type 1 (HSV-1) and HSV-2, an ob- t
and one or more covalently attached glycosaminoglycan servationattributedtothecombinedeffectsoftheaminoacid
chains. The degree of sulfation in the glycosaminoglycan content and the structures of the peptides (4, 119, 120, 122).
chains makes them among the most anionic compounds Interestingly,differencesinantiviraleffectsagainstHSV-1and
present on mammalian cell surfaces (249). This strong net HSV-2 have also been reported for other polycationic and
negative charge permits glycosaminoglycans to bind to small evenpolyanioniccompounds(109,138).Thesedifferencesmay
cations (192), proteins (110), enzymes (198), growth factors reflecttheviralspecificityofparticularreceptormoleculesand
(53,127,155),cytokines(34),chemokines(101),andlipopro- thedifferentialabilityofpeptidestointeractwiththedifferent
teins(151,182),inadditiontoanumberofpathogens,includ- viralreceptors.
ingviruses(166,234). 3-O-Heparan sulfate is a specific HSV entry receptor with
Heparan sulfate is the most important glycosaminoglycan structural similarities to the usual heparan sulfate, probably
molecule with respect to viral attachment (166, 234); conse- with increased affinity potential for cationic peptides. Peptide
quently, blocking of heparan sulfate can reduce the viral in- interactionwith3-O-heparansulfatemayresultininterference
fection(222,260).Theimportanceofheparansulfatefordif- or blocking of HSV glycoprotein D binding, resulting in spe-
VOL.19,2006 ANTIMICROBIAL HOST DEFENSE PEPTIDES 497
cific inhibition of HSV entry. This might explain why several inhibit fusion between the HIV envelope and the host cell
cationic peptides have demonstrated higher antiviral activity membrane(177)throughspecificbindingoftheviralenvelope
againstHSV-1thanagainstHSV-2(119,120),since3-O-hepa- proteingp120andtheT-cellsurfaceproteinCD4(242).
ran sulfate can serve as an entry receptor for HSV-1 and not Membrane or viral envelope interaction. (i) Viral envelope
forHSV-2(261). interaction.Antimicrobialpeptidesareknownfortheirability
Bovine lactoferricin demonstrated an antiviral activity to interact with lipid membranes, resulting in destabilization,
against human cytomegalovirus and human papillomavirus at translocation,poreformation,orlysis(46,226).Thismakesthe
thevirus-cellinterface(6,55).Bovinelactoferricinalsoexhib- viralenvelopeapotentialtargetfordirectinteraction.Indolici-
ited anti-HIV activity, and this might be related to heparan din causes a direct inactivation of the HIV-1 particle in a
sulfatebinding.BindingofHIVtotheCD4surfacereceptoris temperature-sensitive fashion, indicating a membrane-medi-
knowntoinduceconformationalchangesingp120intheviral ated antiviral mechanism (208). Dermaseptin also exerts an
envelope, resulting in increased affinity for heparan sulfate. anti-HIVactivitypriortoviralentrybydirectinteractionwith
Thisfindingimpliesthatheparansulfateisimportantatalater theviralparticle,disturbingitsorganizationanddisruptingthe
D
stageofthevirus-cellattachmentprocess(254). viral membrane (153). In contrast, dermaseptin has no such o
(i) Blocking of cell-to-cell spread. The effect of antiviral direct effect on the HSV envelope. The anti-HSV activity of w
n
peptidesisalsorelatedtotheirabilitytoinhibitthespreadof dermaseptinisproposedtoresultfromblockingofviralentry lo
virusfromonecelltoaneighboringcellacrosstightjunctions byinteractionwithviralorcellularsurfacemoleculesinvolved a
d
(cell-to-cellspread)orinhibitionofgiantcell(syncytium)for- in the attachment/adsorption/fusion phase of HSV (16). The
e
mation. This is a property of the (cid:2)-helical alpha and gamma antibacterialselectivityofdermaseptindependsinpartonthe d

interferons,whichreducethecell-to-cellspreadofHSV(167). lipidcompositionofthemicrobialmembranerelativetothatof fr
o
Rabbit(cid:2)-defensinNP-1hasalsobeenreportedtoinhibitboth thehostcell(52).Similarrequirementscouldalsobehypoth- m
primary entry and cell-to-cell spread of HSV (225). Similar esizedfortheabilityofantiviralpeptidestointeractwithand
h
anti-HSV activity has been indicated for bovine lactoferricin destabilizeviralmembranes. tt
p
(5,119),whilethepolyphemusinanalogueT22andtachyplesin Humanneutrophilpeptide-1(HNP-1)isan(cid:2)-defensinthat :
/
I have been demonstrated to inhibit syncytium formation in neutralizes HSV-1 in a time-, temperature-, and pH-depen- /c
m
cocultures of persistently HIV type 1 (HIV-1)-infected cells dent manner. Neutralization of HSV-1 is also antagonized by
r
(171,177). serum or serum albumin (45). Preincubation of HSV-2 and .a
Blocking of viral entry by interaction with specific cellular HNP-1 prior to infection has been shown to reduce infection s
m
receptors. Antimicrobial peptides might interact directly with by(cid:8)98%.Incontrast,pretreatmentofhostcellswithHNP-1 .
o
specific viral receptors on the host cell (42, 240), influencing hadnoobviouseffectontheanti-HSVactivity.Althoughthe r
g
viral attachment, entry, or intracellular shuttling. The most peptide did not compete with viral envelope proteins for /

o
obvious example of this is the known ability of the binding to cellular heparan sulfate, it still prevented viral
n
polyphemusinanalogueT22tobindtothechemokinereceptor entry (225). N
CXCR4,whichservesasacoreceptorforHIV-1entryintoT Aconcentration-,time-,andtemperature-dependentinacti- o
v
cells (173, 243). Thus, it antagonizes that subgroup of HIV vationofvesicularstomatitisvirus(VSV)isobservedwhenthe e
strains which use this chemokine receptor but not those that virusisincubatedwithtachyplesin-1oritsisopeptidespriorto m
b
useCCR5(263). infection. Electron micrographs of the tachyplesin-1-treated
e
RecentlyanewHSVentryreceptorwasdescribed(196);it VSV particle showed naked and damaged virions, confirming r
is effectively blocked by binding of an (cid:2)-helical peptide in a thedirecteffectofpeptidesontheviralenvelope.Similarbut 25
coiled-coil formation (197). Whether this receptor is specific weaker inactivation has been observed for influenza A virus ,
2
forHSV-1oralsoallowsHSV-2entryisunknown;therefore, (IAV) (type H1N1), while HSV-1, HSV-2, adenovirus 1, reo- 0
1
this cannot definitively explain differences in the antiviral ef- virus2,andpoliovirus1wereresistant(174). 8
fectsofpeptidesonthetwoviruses.However,thereiscertain Neither cecropin, magainin, nor bovine or human lactofer- b
y
evidencethatthisreceptormaybeapotentialtargetforseveral ricin possesses the ability to directly inactivate HSV when
(cid:2)-helical cationic peptides (119). Similar coiled-coil domains mixedtogetherpriortoinfection(3,5).Usingelectronmicros- gu
(196)arefoundinfusionproteinssuchasthehemagglutininof copy, it has been demonstrated that bovine lactoferricin did e
s
influenzavirusandgp41ofHIV.Studieshaveillustratedthat not interact directly with the HSV particle, indicating that t
peptides mimicking these heptad repeat domains specifically interactionswithHSVglycoproteinsdonotoccur(H.Jenssen,
interferewithmembranefusionandviralinfection(115,228). unpublishedresults).
Blocking of viral entry by interaction with viral glycopro- (ii) Cellular membrane interaction. Host cell membranes
teins.Antimicrobialpeptideinteractionswithglycoproteinsin are involved in several stages of viral interaction, and due to
the viral envelope have been proposed to influence the viral theabilityofpeptidestointeractwithandpermeabilizemem-
entry process. (cid:7)-Defensin (retrocyclin 2) interacts with the branes, this must be considered as a potential target. Similar
HSV-2 glycoprotein B with high affinity, thus protecting the permeabilization of the host cell membrane appears to occur
cellsfromHSV-2infection(270).Thecloselyrelatedretrocy- (139), and the resulting alteration of host membranes could
clin-1bindsHIVgp120withhighaffinity,aslongastheenve- affect the efficiency of viral entry. An eight-residue cyclic DL-
lopeproteinisglycosylated,probablyresultinginananti-HIV (cid:2)-peptidehasbeenshowntospecificallypreventthelowering
activity.Thismakesthe(cid:7)-defensinthefirstantimicrobialpep- ofpHinendocyticvesicles,thusarrestingtheentryofadeno-
tideisolatedfromvertebrateswithalectin-likecharacter(256). virus particles by this pathway and abrogating the infection
The polyphemusin analogue T22 has been demonstrated to withouthavinganapparenteffectonhostcellviability(102).
498 JENSSEN ET AL. CLIN.MICROBIOL.REV.
TABLE 3. Selectedexamplesofnaturalantibacterialpeptides
Peptide Structure Source(s) Proposedantibacterialmechanism Reference(s)
Magainin (cid:2)-Helix Frog Permeabilizesbacterialmembrane 161,273
CecropinA (cid:2)-Helix Silkmoth Membranedestabilizing 73,106
Mellitin (cid:2)-Helix Bee Membranedestabilizing 32,63
LL-37 (cid:2)-Helix Human Membranepermeabilization;stronglysaltantagonized 12,172,176
BuforinII (cid:2)-Helix/extended Toad Bindingofnucleicacid 187,188
(cid:2)/(cid:3)-Defensins (cid:3)-Sheet Mammals,analoguesin Manyarestronglysaltantagonized;cellmembraneand 108,147,262
insectsandfungi intracellulartargets,inhibitsmacromolecularsynthesis
Protegrin (cid:3)-Sheet Human,porcine Verypotent,membranepermeabilization 95
Polyphemusin (cid:3)-Sheet Horseshoecrab Verypotent,translocatesintocells 202,280
Indolicidin Extended Bovine Inhibitsmacromolecularsynthesis,Ca2(cid:1)-calmodulininteraction 61,104,227
PR-39 Extended Porcine InhibitsDNA/RNA/proteinsynthesis,noporeformation 22
D
o
w
n
This effect has been hypothesized to be a result of the hasbeenshowntoinhibitJuninvirusmultiplicationbyreduc- lo
peptide’s membrane permeabilization properties. Similar tion of its protein synthesis under conditions where the syn- a
d
effects could be anticipated for other membrane-acting an- thesisofhostcellproteinsremainsunaffected(3).Melittinand
e
timicrobial peptides. cecropinAarealsoabletoinhibitcell-associatedproductionof d

Intracellulartargetsandhostcellstimulation.Itisknownthat HIV-1 by suppressing HIV-1 gene expression (255). Early fr
o
antimicrobialhostdefensepeptidessuchasPR39andLL-37are steps in the HIV-1 replication cycle are inhibited by prote- m
abletocrosslipidmembranes,includingtheplasmaandnuclear grin-1 (236), while HIV-1 integrase is effectively inhibited by
h
membranesofhostcells,whileothersareconstitutivelylocatedas indolicidin(134).Retrocyclinhasbeendemonstratedtoinhibit tt
p
precursorsinsidehostcellvacuoles(5,93,139).Cellularinternal- proviral DNA formation and protect immortalized and pri- :
/
izationoftheseantimicrobialpeptidescanresultingene/protein maryhumanCD4(cid:1)lymphocytesfrominvitroHIV-1infection /c
m
stimulation, influencing host cell antiviral mechanisms (26), or (42). Transport of HSV-2 tegument protein VP16 to the cell
r
mightblockviralgene/proteinexpression(255). nucleus and expression of ICP4 are effectively blocked in the .a
The effect of antimicrobial peptides has also been demon- presence of rabbit (cid:2)-defensin (NP-1) (225). In addition, the s
m
strated to be crucially dependent on the experimental condi- human(cid:2)-defensin-1hasdemonstratedadirectinactivationof .
o
tions.Saltmayinfluencethestructureofthepeptidesandtheir HIV-1 in the absence of serum, an effect that is abolished by r
g
association with anionic cell molecules, thus affecting their the presence of serum. Conversely, in the presence of serum /

o
antimicrobial activity (117); e.g., both the antibacterial and the peptide inhibits HIV-1 replication, partly by interfering
n
antifungalactivitiesofcathelicidinandtheantibacterialactiv- withhostcellproteinkinaseCsignaling(37). N
ity of (cid:3)-defensin are salt dependent (12, 230). Similarly, the o
antiviralmodeofactionof(cid:2)-defensinhasbeendemonstrated ANTIBACTERIALACTIVITY ve
to be serum dependent (37). This indicates that the peptide’s m
b
actioninvivomaybedependentonthephysiologicalmilieuat Byfarthebest-studiedclassofcationicantimicrobialpeptides
e
thesiteofinfection.Transmissionelectronmicroscopystudies arethosewithantibacterialactivity(60).Itiswellunderstoodthat r
2
haverevealedthathumanandbovinelactoferricinscantrans- regardlessoftheiractualtargetofaction,allantibacterialcationic 5
locateintracellularly(5).Bovinelactoferricinwasabletoenter peptidesmustinteractwiththebacterialcytoplasmicmembrane ,
2
bothchondroitinsulfate-andheparansulfate-deficientcellsin (92).Thedrivingphysicalforcesbehindantibacterialactivityhave 0
1
an energy-independent manner (5). This mechanism was de- been defined in detail (see references 46, 91, and 92 for over- 8
scribed earlier (66, 67, 239), and it appears that the arginine views)andincludenetpositivecharge(enhancinginteractionwith b
y
contentofthepeptidesisanimportantfactor,asitisaknown anionic lipids and other bacterial targets), hydrophobicity (re-
g
featureofnuclearlocalizationsignals.Ashumanlactoferricin quiredformembraneinsertionandoftendrivenbythisprocess), u
has multiple Arg residues (195), this probably contributes to andflexibility(permittingthepeptidetotransitionfromitssolu- e
s
theshuttlingofthepeptideintothenucleus,whereitcanbind tion conformation to its membrane-interacting conformation). t
DNA. Consistent with this, the antiviral peptides LL-37 and Eachofthesecharacteristicscanvarysubstantiallyoverapartic-
indolicidinbothcanactasnuclearlocalizationsignalstotrans- ular range but are essential for the function of the peptides as
locateantisensenucleicacids(215). antimicrobial agents and allow them to interact with bacterial
Because of the known ability of peptides to interact with membranes, which is critical to their exertion of antimicrobial
DNA (104, 123, 187, 232), one might speculate that they can effects.
directly influence viral nucleic acid synthesis, as shown for
polyphemusin T22 and lactoferrin. Conversely, peptides are
StructuralRequirementsforAntibacterialPeptides
knowntohaveimmunomodulatoryactivitieswhichincludethe
upregulation of interferons and chemokines (24, 27, 37, 219), As mentioned above, cationic antimicrobial peptides are
andthuspeptidesmightexerttheirantiviralactivitiesinpartby generallycategorizedintofourstructuralclasses,i.e.,(cid:2)-helical,
stimulatingtheantiviralimmunesystemofthehostcell. (cid:3)-sheet, loop, or extended structures (21, 87); however, there
Direct effects of peptides on the intracellular steps of viral are many peptides that do not fit into this simplified classifi-
infection have been demonstrated. For example, cecropin A cation scheme (Fig. 1; Table 3). Many bacterially produced
VOL.19,2006 ANTIMICROBIAL HOST DEFENSE PEPTIDES 499
peptides,forinstance,havetwodomains,oneofwhichis(cid:2)-he- membrane permeabilizing at their minimal effective concen-
licalinnaturewhiletheotherhasa(cid:3)structure(251).Formany trations or at concentrations well above or well below these
peptides these secondary structures are observed only when concentrations. Nonetheless, antibacterial peptides seem
thepeptidesinteractwithmembranes;e.g.,bovineneutrophil largely able to effect their antimicrobial activity because of
indolicidin is unstructured in an aqueous environment but theiramphipathicityoramphiphilicityandbecauseofthepres-
adopts a boat-like conformation when interacting with mem- ence of regions within the folded structure with high concen-
branesandmembranemimeticssuchassodiumdodecylsulfate trationsofpositivelychargedresidues(202).
and dodecyl phosphocholine (212) (Fig. 1). The plasticity of
the secondary structure of indolicidin has been suggested to
ModeofActionofAntibacterialPeptides
permit different interactions with different molecules, includ-
ingDNAandmembranes(104). Itwasoriginallyproposedthatpermeabilizationofthebac-
One approach to increase the antibacterial activity of cat- terial cell membrane was the sole mode of action of antibac-
ionicpeptideshasbeentoaltertheirflexiblesecondarystruc- terialpeptides.Thereisanincreasingbodyofevidence,how-
D
tures.Bychangingthemembrane-associatedshapeofindolici- ever, that indicates that some antimicrobial peptides exert o
dinsothattheNandCterminiweredrawnclosertogether,the their effects through alternative modes of action or that they w
n
activity against gram-negative bacteria was increased. This mayinfactactuponmultiplebacterialcelltargets.Regardless lo
shape could also be stabilized by adding a cysteine residue to of their precise mode of action, the activities of antibacterial a
d
each end and creating a disulfide bridge (213). Conversely, a peptides are almost universally dependent upon interaction
e
synthetic indolicidin analogue was made by introducing a co- with the bacterial cell membrane (92). The first step in this d

valent cross-link between Trp6 and Trp9 (183). Both changes interactionistheinitialattractionbetweenthepeptideandthe fr
o
resulted in decreased protease sensitivity, a strong potential target cell, which is thought to occur through electrostatic m
assetinvivo,butdidnotinhibitantimicrobialactivity.Similar bonding between the cationic peptide and negatively charged
h
attempts to stabilize specific structural elements have been components present on the outer bacterial envelope, such as tt
p
made with a cecropin-melittin hybrid peptide, in which the phosphategroupswithinthelipopolysaccharidesofgram-neg- :
/
(cid:2)-helical structure in solution was stabilized by the introduc- ative bacteria or lipoteichoic acids present on the surfaces of /c
m
tion of a covalent lactam bond between two residues four gram-positive bacteria. In the case of gram-negative bacteria,
r
aminoacidsapart,resultinginimprovedactivityofsomecon- peptides are inserted into the outer membrane structure in a .a
structswhiledecreasingtheactivityofothers(103).Stabiliza- processdrivenbyhydrophobicinteractionsandpossiblyinvolv- s
m
tion of the helical structure in cecropin A has similarly dem- ing prefolding of the peptides into a membrane-associated .
o
onstrated the importance of this structural domain in structure; this leads to a disturbance of the outer membrane r
g
antibacterial activity against E. coli (68). Alternatively, intro- structure and permeabilizes this membrane to other peptide /

duction of a disulfide bond within the C-terminal (cid:2)-helix of moleculesinaprocesstermedself-promoteduptake.Thenet o
n
sakacinPtoincreasetheamountof(cid:2)-helicalstructureledto resultisthatthepeptidesarriveatthecytoplasmicmembrane, N
a broadening of the spectrum of activity (251). Thus, precon- where they enter the interfacial region of the membrane (the o
v
ditioning peptides to adopt structures related to their final interfacebetweenthehydrophilicandhydrophobicportionsof e
membrane-associated ones can occasionally be advantageous themembrane)inaprocessdrivenbyelectrostaticandhydro- m
b
whilegivingrisetootherassetssuchasproteasestability. phobic interactions. The higher proportion of negatively
e
Theantibacterialactivityofcationicpeptidescanalsobemod- chargedlipidsonthesurfacemonolayerofthebacterialcyto- r
2
ulatedthroughalterationofthepeptide’shydrophobicityornet plasmicmembraneplaysanimportantroleintheselectivityof 5
charge.Studieshavedemonstratedthathighlevelsofhydropho- antimicrobialpeptidesforbacterialcellsovereukaryoticcells, ,
2
bicity can decrease selectivity between the desired bacterial tar- inwhichunchargedlipidspredominateatthehostcellsurface. 0
1
getsandhostcells(136,275).Similarly,incorporationofcharged Theeventsthatoccuratthemembranesurfacearethesubject 8
residues above a certain maximum (varying with each peptide) ofconsiderabledebate,andseveralprominentmodels(called b
y
doesnotleadtoanincreaseinactivity(46).Thus,thisbalanceof variouslythebarrel-stave,carpet,detergent,toroidalpore,and
g
charge and hydrophobicity can be delicate and must be empiri- aggregate models) have been proposed. Each of these indi- u
callydeterminedforeachseriesofpeptides. cates a different type of intermediate that can lead to one of e
s
Consequently, the inclusion of a particular peptide into a threetypesofevents:formationofatransientchannel,micel- t
structuralgroupdoesnotgiveanindicationastoitsmodeof larization or dissolution of the membrane, or translocation
action or its spectrum of activity. In fact, some peptides with acrossthemembrane.Asaresult,thepeptidecanpermeabil-
very similar secondary structures have quite opposite charac- izethemembraneand/ortranslocateacrossthemembraneand
teristics with respect to antibacterial activity (Table 3). The into the cytoplasm without causing major membrane disrup-
(cid:2)-helicalmelittinfrombees,whichisthoughttoperforatethe tion.Hence,themodesofactionofantibacterialpeptidescan
membranesofbothprokaryoticandeukaryoticorganisms,falls be broadly categorized as either membrane acting or non-
within the (cid:2)-helical structural class. Conversely, the (cid:2)-helical membrane acting. While most cationic antibacterial peptides
toad peptide buforin translocates into cells and acts on mac- studied so far have been characterized as membrane perme-
romolecular synthesis. Studies of (cid:2)-helical analogues of a abilizing, it should be noted that virtually any cationic am-
cecropin-melittinhybridpeptidehaverevealedthatevenpep- phiphilic peptide will cause membrane perturbation in model
tides that have similar secondary structures and minimal dif- systems if a high enough concentration is applied, and there
ferences in the primary sequence can possess quite different are few examples of studies with intact bacteria (193, 279).
antibacterialactivities(64).Indeed,differentpeptidesmaybe Thus, it is possible that such conclusions reflect the extensive
500 JENSSEN ET AL. CLIN.MICROBIOL.REV.
studiesofmodelmembranesystemsinwhichalternativemech- 267).Inthebarrel-stavemodel(57)(Fig.2C),peptidesreori-
anismsofactionwouldnotbedetectedand/ortheuseofmedia enttobecomethe“staves”ina“barrel”-shapedclusterwhich
whichdonotreflectphysiologicsaltorpHconditions,leading orients perpendicular to the plane of the membrane. The hy-
toanoverestimationofthepermeabilizingactivityofthepep- drophobicregionsofeachpeptideintheclusterareassociated
tide in question. Studies conducted with whole bacterial cells withthelipidcore,whilethehydrophilicregionsarefacingthe
have revealed that several antibacterial peptides translocate lumen of the newly formed transmembrane pore. This model
into cells and do not cause membrane permeabilization but predicts that there will be a consistent channel size (or sizes,
rathermediatebacterialcelldeathbytargetingessentialintra- also called substates), and this is not true for most peptides,
cellularprocesses.Belowisanoverviewofthecurrentmodels although experimental evidence supports this mechanism of
forantibacterialpeptide-mediatedkillingthrougheithermem- membrane permeabilization for the fungus-derived (nonca-
brane-permeabilizingornon-membrane-permeabilizingmech- tionic)antimicrobialpeptidealamethicin(94,233)andforthe
anisms. cyclicdecamericcationicpeptidegramicidinS(279).
Membrane-permeabilizingpeptides.Severaldifferentmodels Incontrasttothebarrel-staveandtoroidalporemodels,the
D
havebeenproposedtoexplainhow,followinginitialattachment, carpetmodelproposesthataggregatesofpeptidealignparallel o
antibacterial peptides insert into the bacterial membrane to tothelipidbilayer,coatinglocalareasinacarpet-likefashion w
n
formtransmembraneporeswhichresultinmembraneperme- (200) (Fig. 2D). At sufficiently high concentration, this is lo
abilization (Fig. 2). The amphipathic nature of antimicrobial thought to have a detergent-like activity (causing patches of a
d
peptidesisakeycharacteristicforthisprocess,ashydrophobic themembranetobreakupintomicelles),causinglocaldistur-
e
regions are necessary to interact directly with the lipid com- bancesinmembranestabilitywhichcanleadtotheformation d

ponents of the membrane, while hydrophilic regions either of holes in the membrane. Many cationic antimicrobial pep- fr
o
interactwiththephospholipidheadgroupsorfacethelumen tides will do this at high enough concentrations due to their m
of the pore. Generally, these models can explain the pore- amphipathiccharacter;however,thereisverylimitedevidence
h
formingabilityof(cid:2)-helicalantibacterialpeptides;however,the demonstratingthatmostpeptidescausemembranedissolution tt
p
mechanisms utilized by (cid:3)-sheet peptides such as defensins at the minimal effective concentrations in vivo (or, in many :
/
have not been as well studied. While (cid:3)-sheet antimicrobial studied cases, in vitro; for example, Zhang et al. [279] exam- /c
m
peptidescanadoptamphipathicfolds,thereislittleexperimen- inedninerepresentativepeptidesanddemonstratedthatmost
r
tal evidence to indicate which of the following models is ap- of them were able to cause membrane flip-flop at far lower .a
plicabletodefensins. concentrations than those at which they could cause calcein s
m
In all models peptides first interact preferentially with the release). Based on mechanistic studies, however, this mecha- .
o
negativelychargedlipidheadgroupsatthemembranesurface, nism has been proposed for ovispirin, a rationally designed r
g
adopting an orientation parallel to (i.e., in the plane of) the antibacterial peptide based on the sheep cathelicidin Smap29 /

o
membrane at the membrane interface. A mechanism, known (264). It is important to recognize that each of these models
n
as the aggregate model, with some similarity to the toroidal mayhavevalidityunderdifferentcircumstances,andexamina- N
pore model, has been proposed by our laboratory (259) (Fig. tion of a broad range of peptides with different sizes and o
v
2A).Thismodelhasalessformalnatureandcanexplainboth structures has indicated that they leave quite different signa- e
membranepermeabilization,wherebyinformalchannelswitha tures of membrane interaction (280), while differential scan- m
b
variety of sizes and lifetimes form (259), and translocation ning calorimetry studies on LL-37 (98) and polyphemusin
e
acrossthebilayer,whichisknowntooccurforseveralpeptides (203)haveshownoppositeeffectsonmembranecurvature. r
2
(203).Inthismodelpeptidesreorienttospanthemembraneas Peptidesthatdonotactbymembranepermeabilization.All 5
anaggregatewithmicelle-likecomplexesofpeptidesandlipids peptidesmustinteractwiththecytoplasmicmembrane,ifonly ,
2
(as also suggested by Matsuzaki et al. [161, 162] and by the to get to their site of action. It is now well established that 0
1
toroidalporemodel),butinthismodelthepeptidesadoptno severalantimicrobialpeptidesdonotcausemembraneperme- 8
particular orientation. Predictions of this model are that the abilizationattheminimaleffectiveconcentrationyetstillresult b
y
lack of a formal channel structure will lead to channels that in bacterial death. A growing number of peptides have been
g
varydramaticallyincharacter,thatthepeptidehasthecapacity shown to translocate across the membrane and accumulate u
totranslocateacrossthebilayerastheaggregatescollapse,and intracellularly,wheretheytargetavarietyofessentialcellular e
s
thatthemembranewillundergonegativecurvaturestrain,all processes to mediate cell killing. Novel modes of action that t
of which have been shown for the horseshoe crab peptide haverecentlybeendemonstratedincludeinhibitionofnucleic
polyphemusin.Inthetoroidalporemodel(Fig.2B),aggregates acid synthesis, protein synthesis, enzymatic activity, and cell
of peptides insert themselves in an orientation perpendicular wallsynthesis(28)(Fig.2).
to the membrane to form a pore, with the membrane also ThefrogantimicrobialpeptidebuforinIItranslocatesacross
curving inward to form a hole with the head groups facing thebacterialmembranewithoutcausingpermeabilizationand
towardsthecenterofthepore,andthepeptideslinethishole. binds to both DNA and RNA within the cytoplasm of E. coli
Onepredictionofthismodelisthatthemembranewillexhibit (187).Similarly,(cid:2)-helicalpeptidessuchasderivativesofpleu-
positive curvature strain (due to the membrane bending rocidin,afish-derivedantimicrobialpeptide,anddermaseptin,
around to form the toroidal hole with the peptides within), isolated from frog skin, cause inhibition of DNA and RNA
although the formation of a formal toroidal channel has not synthesisattheirMICswithoutdestabilizingthemembraneof
actually been demonstrated. Examples of antimicrobial pep- E. coli cells (193, 238) (Fig. 2E). Inhibition of nucleic acid
tides that are proposed to form this type of transmembrane synthesis has also been demonstrated for antimicrobial pep-
pore include the magainins, melittin, and LL-37 (85, 98, 162, tidesfromdifferentstructuralclasses,suchasthe(cid:3)-sheethu-