MINIREVIEW Multifunctional host defense peptides: intracellular-targeting antimicrobial peptides Pierre Nicolas Biogene ` se des Signaux Peptidiques, ER3-UPMC, Universite ´ Pierre et Marie Curie, Paris, France Introduction There has been increasing interest in recent years in describing the complex, multifunctional role that anti- microbial peptides play in directly killing microbes, boosting specific inate immune responses, and exerting selective immunomodulatory effects on the host [1–4]. Furthermore, many antimicrobial peptides are quite inactive on normal eukaryotic cells. The basis for this discrimination appears to be related to the lipid com- Keywords antimicrobial peptides; cell-penetrating peptides; dermaseptin; intracellular target; membrane translocation Correspondence P. Nicolas, Biogene ` se des Signaux Peptidiques (BIOSIPE), ER3-UPMC, Universite ´ Pierre et Marie Curie, Ba ˆ timent A –5e ` me e ´ tage, Case courrier 29, 7 Quai Saint-Bernard, 75005 Paris, France Fax: +1 44 27 59 94 Tel: +1 44 27 95 36 E-mail: pierre.nicolas@upmc.fr (Received 1 May 2009, revised 25 July 2009, accepted 29 July 2009) doi:10.1111/j.1742-4658.2009.07359.x There is widespread acceptance that cationic antimicrobial peptides, apart from their membrane-permeabilizing ⁄ disrupting properties, also operate through interactions with intracellular targets, or disruption of key cellu- lar processes. Examples of intracellular activity include inhibition of DNA and protein synthesis, inhibition of chaperone-assisted protein folding and enzymatic activity, and inhibition of cytoplasmic membrane septum formation and cell wall synthesis. The purpose of this minireview is to question some widely held views about intracellular-targeting anti- microbial peptides. In particular, I focus on the relative contributions of intracellular targeting and membrane disruption to the overall killing strategy of antimicrobial peptides, as well as on mechanisms whereby some peptides are able to translocate spontaneously across the plasma membrane. Currently, there are no more than three peptides that have been convincingly demonstrated to enter microbial cells without the involvement of stereospecific interactions with a receptor ⁄ docking mole- cule and, once in the cell, to interfere with cellular functions. From the limited data currently available, it seems unlikely that this property, which is isolated in particular peptide families, is also shared by the hun- dreds of naturally occurring antimicrobial peptides that differ in length, amino acid composition, sequence, hydrophobicity, amphipathicity, and membrane-bound conformation. Microbial cell entry and ⁄ or membrane damage associated with membrane phase ⁄ transient pore or long-lived transitions could be a feature common to intracellular-targeting antimi- crobial peptides and mammalian cell-penetrating peptides that have an overrepresentation of one or two amino acids, i.e. Trp and Pro, His, or Arg. Differences in membrane lipid composition, as well as differential lipid recruitment by peptides, may provide a basis for microbial cell kill- ing on one hand, and mammalian cell passage on the other. Abbreviations MIC, minimal inhibitory concentration; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol. FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS 6483 position of the target membrane (i.e. fluidity, negative charge density, and the absence ⁄ presence of choles- terol), and the possession, by the microbial organism, of a large, negative transmembrane electrical potential. There is now a widespread acceptance that antimicro- bial peptides, apart from their membrane-permeabiliz- ing ⁄ disrupting properties, may also affect microbial viability by interactions with intracellular targets or disruption of key intracellular processes. Much of the focus in this area has been on the identification of tar- gets in the interior of the microbial cell and the mecha- nism by which antimicrobial peptides can enter the microbial cell in a nondisruptive way [5–7]. The prevailing dogma that the microbicidal effects of cationic antimicrobial peptides solely involve cyto- plasmic membrane permeabilization ⁄ disruption of target cells has been increasingly challenged by the realization that: (a) information on the membrane interactions and activity of antimicrobial peptides obtained in vitro using simple artificial membrane bilayers, or in vivo using intact microbial cells, is not clearly correlated with the observation of microbial death; (b) several antimicrobial peptides may recognize and inactivate cellular targets in vitro, such as nucleic acids, proteins, enzymes, and organelles, their mecha- nism of action being postulated to involve transloca- tion across the plasma membrane in a nonlethal manner; (c) regardless of which model of antimicrobial peptide-induced membrane permeabilization ⁄ disruption is correct, they all offer the peptide the possibility of rapidly crossing the cytoplasmic membrane and reach- ing macromolecular targets in the cell interior; and (d) most antimicrobial peptides show strong similarities in charge, structure and membrane interactions with cell- penetrating peptides, which are thought to enter mam- malian cells by passive transport [8]. The purpose of this minireview is to describe and critically analyze some widely held views about intracellular-targeting antimicrobial peptides. In particular, I focus on the proposed mechanisms by which antimicrobial peptides might translocate across microbial membranes to attack cellular targets. Microbial membrane permeabilization versus intracellular killing There has been increasing speculation in the last dec- ade that antimicrobial peptide-mediated permeabiliza- tion ⁄ disruption of the microbial cytoplasmic membrane is not the only mechanism of cell killing, and that antimicrobial peptide might also operate by entering the cells and interfering with their metabolic function. Antimicrobial peptides with varying antimicrobial potencies exhibit disparate extents of membrane permeabilization and cell killing Even though all cationic antimicrobial peptides are able to interact with microbial cytoplasmic mem- branes, and some strongly perturb bilayers, the num- ber of studies documentating a clear dissociation between cell death and the ability of some peptides to permeabilize the membrane, either in vitro or in vivo, has increased significantly during the last decade. For example, TWF, an analog of the cathelicidin-derived antimicrobial peptide tritrpticin, in which Trp is replaced with Phe, is much more effective than TPA, in which the two Pro residues of tritrpticin are replaced with with Ala, against both Staphylococ- cus aureus and Escherichia coli [9]. However, TWF shows very little membrane-disrupting activity and no ability to depolarize the membrane potential of micro- bial cell targets, whereas TPA rapidly depolarizes the membrane and causes rapid leakage of negatively charged phospholipid vesicles. Dermaseptin B2 – GLWSKIKEVGKEAAKAAAKAAGKAALGAVSE- AVa – from frog skin and its C-terminally truncated analog [1–23]-dermaseptin B2 are both highly effective in permeabilizing calcein-loaded phosphatidylcholine (PC) ⁄ phosphatidylglycerol (PG) and phosphatidyletha- nolamine (PE) ⁄ PG vesicles [10]. Whereas dermaseptin B2 rapidly kills bacteria [11], [1–23]-dermaseptin B2 is devoid of antimicrobial activity and is inefficient in permeating intact bacterial cells. The bacterium- derived antimicrobial peptides polymyxin B and poly- mixin E1 failed to cause significant depolarization of the Pseudomonas aeruginosa cytoplasmic membrane but rapidly killed the test organism. In contrast, grami- cidin S caused rapid depolarization of the bacterial cytoplasmic membrane at concentrations at which no killing was observed [12]. These observations support the concept that, for some antimicrobial peptides, membrane perturbation and cell killing may be inde- pendent events that occur individually or complemen- tary to other mechanisms of action [13]. Antimicrobial peptides exhibit temporal dissociation between microbial membrane permeabilization and cell death Although there is a wealth of evidence that many anti- microbial peptides interact and increase the permeabil- ity of microbial membranes as part of their killing mechanism, it is not clear whether this is a lethal step. Intracellular-targeting antimicrobial peptides P. Nicolas 6484 FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS In addition, several antimicrobial peptides kill micro- bial cells in the absence of significant permeabiliza- tion ⁄ disruption of membrane structure and functions [14]. For some antimicrobial peptides, permeabilization of the microbial cytoplasmic membrane and cell killing begin concomitantly as quickly as a few minutes after exposure [15–17]. For others, there is a considerable lag period between these two events. For instance, although TWF and TPA are equipotent in inhibiting the growth of S. aureus and E. coli, TWF requires a lag period of about 3–6 h for bactericidal activity, whereas TPA kills bacteria after only after 30 min of exposure [9]. Experiments based on confocal micros- copy on living cells using the fluorescence of fluores- cein isothiocyanate, 4¢,6-diamidino-2-phenylindole and 5-cyano-2,3-ditolyl tetrazolium chloride revealed that sublethal concentrations of temporin L permeabilize the inner membrane of E. coli to small compounds, but do not allow the killing of bacteria [18]. At higher peptide concentrations, the bacterial membrane becomes permeable to large cytoplasmic components, and this is concomitant with death of bacteria. This shows that membrane permeabilization of bacteria by temporin L and TWF is not a lethal step per se in the absence of a catastrophic collapse of the membrane integrity, and that peptide-mediated killing required other additional events. The choice of a membrane model can influence the outcome of an in vitro study of lipid–peptide interaction Most models accounting for antimicrobial peptide- induced membrane permeabilization are inferred from data obtained with very simple, artificial membrane models that mimic microbial cell membranes, whether in the form of lipid monolayers, oriented bilayers, or vesicles (reviewed by Bhattacharjya and Ramamoorthy [19]; this issue). Extrapolating these in vitro data to an in vivo model is not straightforward, and the choice of the model system may profoundly influence the out- come of a study of lipid–peptide interaction. An ele- gant study of the interaction of the human cathelicidin antimicrobial peptide LL-37 with single phospholipid monolayers, bilayers and bilayers composed of binary mixtures of phospholipid species predominantly used in model membrane experiments, i.e. PC, PE, PG, and phosphatidylserine, showed the following [20]: (a) the effects on single lipid monolayers are not comparable to those on the corresponding bilayers; (b) there are four different modes of interaction of LL-37 on bilay- ers with the four different lipids used; and (c) there are significant differences in the mode of peptide–lipid interaction between the binary lipid mixtures PC ⁄ PG, PE ⁄ PG, and PC ⁄ phosphatidylserine, which all carry the same net charge. A similar disagreement was observed for the interaction of dermaseptin B2 with cardiolipin ⁄ PC and PG ⁄ PC vesicles. Peptide concentration dependence of antimicrobial action Research on the mode of action of antimicrobial pep- tides in vitro has usually been conducted at high multi- ples of the minimal inhibitory concentration (MIC) of peptides and ⁄ or high peptide ⁄ lipid ratios. Owing to technical limitations, these high peptide concentrations are necessary to determine the three-dimensional struc- ture of membrane-bound antimicrobial peptides and to observe perturbation of the thermodynamic parameters of the gel-to-crystalline phase transition of lipid mem- brane models, lipid flip-flop, calcein release on model liposomes, etc. However, there is no evidence that such peptide concentrations, which provide almost full bac- terial membrane coverage by the peptides, are really present at the surface of bacteria during bacterial kill- ing in vivo [21]. In addition, electron transport chains and ion and complex nutrient transport systems require the coordination over time and space of a net- work of interacting proteins, coenzymes, and sub- strates. That microbial cell death may result from nonspecific interference of cationic amphipathic peptides with the dynamic organization of membrane- bound pathways rather than just from membrane permeabilization has seldom been evaluated, and it is hardly possible to do so in vitro through the use of lipid membrane models [22]. The above-mentioned data collectively suggest that, at least near the MIC, the killing actions of some antimicrobial peptides are complex and may involve targets in the interior of the microbial cell. How antimicrobial peptides may enter microbial cells Two general mechanisms are proposed to describe the process by which antimicrobial peptides enter the microbial cells, spontaneous lipid-assisted translocation and stereospecific receptor-mediated membrane trans- location. The precise mechanisms whereby some antimicrobial peptides are able to translocate sponta- neously across the plasma membrane remain largely unknown, and may vary from peptide to peptide. However, membrane translocation seems to be a corol- lary of transient membrane permeabilization. There are currently several models accounting for antimicrobial P. Nicolas Intracellular-targeting antimicrobial peptides FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS 6485 peptide-induced membrane permeabilization ⁄ disruption of microbial cells. The Shai–Matzusaki–Huang unifying model proposes that a-helical antimicrobial peptides initially bind parallel to the membrane plane and car- pet the surface of the bilayer [23–26], with the apolar amino acids penetrating partly into the bilayer hydro- carbon core, and the cationic residues interacting with the negatively charged phosphate moieties of the lipid head groups, hence causing membrane thinning and positive curvature strain. To release the strain, a frac- tion of the peptides change their orientation from par- allel to transversal, forming transient mixed phospholipid–peptide toroidal pores. Upon disintegra- tion of the pores, some peptides become translocated to the inner leaflet of the membrane [27], suggesting that stochastic pore disassembly may be a mechanism by which antimicrobial peptides can reach the cell inte- rior. Note that only a few pores exist after the redistri- bution of the peptides between the two leaflets, because the pore formation is a cooperative process. Therefore, the integrity of the membrane is only tran- siently breached, and pores are hardly detectable in the equilibrium state by usual biophysical approaches. Once a threshold level of membrane-bound peptide is reached, this may lead to disruption ⁄ solubilization of the membrane in a detergent-like manner. The thresh- old between the toroidal pore and the detergent-like mechanisms of action may be related to two facets of the cell killing mechanism relying on the peptide con- centration: the membrane composition, and the final peptide ⁄ lipid ratio. Because the threshold peptide con- centrations required for membrane disruption are always close to full bacterial membrane saturation, doubts have arisen regarding the relevance of these thresholds and their importance in vivo [21]. However, rigorous calculations have demonstrated that antimi- crobial peptides with MIC values in the micromolar range can easily reach millimolar concentrations in a bacterial membrane, owing to high partition constants [28]. At this concentration level, there is a strong link between cell death and membrane disruptive events. On the other hand, at low peptide ⁄ lipid ratios, antimi- crobial peptides may translocate across the plasma membrane, perturbing its structure in a transient, non- lethal manner, and reach the cell interior. Another mechanism for breaching membrane perme- ability, the lipid phase boundary defects model, pro- posed that some b -sheeted peptides, such as cateslytin, a 15 residue Arg-rich antimicrobial peptide resulting from the cleavage of chromogranin A, form mainly flat aggregates at the surface of negatively charged bacte- rial membranes as patches of antiparallel amphipathic b-sheets forming rigid and thicker lipid domains enriched in negatively charged lipids [29,30]. These domains become ordered, mainly owing to the inser- tion of aromatic residues into the hydrophobic bilayer core. Zones of different rigidity and thickness bring about phase boundary defects that lead to permeability induction and peptides crossing through bacterial membranes. Thus, the peptides could pass through the membrane and interact with intracellular targets, as do other Arg-rich peptides (see below). The disordered toroidal pore model proposed that a nanometer-sized, toroidal-shaped pore is formed by a single a-helical or b-sheeted peptide that is able to insert into the membrane, because of the difference in mechanical stress between the two faces of the mem- brane, and ⁄ or because of the different electric field, i.e. the electroporation-like mechanism [31–34]. Above a threshold number of membrane-bound peptides, one peptide molecule becomes deeply embedded in the membrane interface. The membrane–water interface becomes unstable, and solvent molecules from the pep- tide-free interface are able to interact with hydrophylic groups of the embedded peptide, resulting in the devel- opment of a continuous pore. In contrast to the Shai– Matzusaki–Huang model of the toroidal pore, only one peptide is found near the center of the pore, and the remaining peptides lay close to the edge of the pore, maintaining a parallel orientation with respect to the membrane plane. The resulting pore is sufficient to allow the passage of the peptide from one side of the membrane to the other. A similar mechanism of tran- sient pore formation was proposed for the transloca- tion of the HIV-1 Tat cell-penetrating peptide across mammalian cell membranes [35]. Intracellular-targeting antimicrobial peptides Although there is no doubt that most cationic antimi- crobial peptides act at high concentrations by permea- bilizing ⁄ disrupting the microbial membrane, recent studies and reviews have reported an ever-growing list of peptides that are presumed to affect microbial via- bility at low to moderate concentrations through inter- action with one or more intracellular targets (Table 1). Examples of intracellular activity include inhibition of DNA and protein synthesis, inhibition of chaper- one-assisted protein folding, inhibition of enzymatic activity, and inhibition of cytoplasmic membrane sep- tum formation and cell wall synthesis. Very different amounts of data, acquired with different experimental protocols, have been presented for individual peptides in order to support this assumption, so that, in most cases, straightforward interpretation of these observa- Intracellular-targeting antimicrobial peptides P. Nicolas 6486 FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS Table 1. Amino acid sequence, membrane-bound structure and suggested internalization mechanism and effect on microbial functions of intracellular-targeting antimicrobial peptides. Hydrophobic residues are in bold. a, carboxamitaded. Name Sequence Membrane-bound Structure Uptake mechanism Intracellular targets Pyrrhocoricin VDKGSYLPRPTPPRPIYNRN Reverse turns at the termini bridged by an extended segment Receptor ⁄ docking component on inner membrane Dnak; DNA and protein synthesis Apidaecin GNNRPVYIPQPRPPHPRI Receptor ⁄ docking component Dnak; DNA and protein synthesis Drosocin GKPRPYSPRPTSHPRPIRV Receptor ⁄ docking component Dnak; DNA and protein synthesis Bactenecin-7 RRIRPRPPRLPRPRPRPLPFPR PGPRPIPRPLPFPRPGPRPIP RPLPFPRPGPRPIPRP Poly-proline-II helix? MIC: receptor ⁄ docking component > MIC: membrane permeabilization ⁄ disruption DNA? Histatin-5 DSHAKRHHGYKRKFHEKHHSHRGY Amphipathic a-helix < MIC: receptor-mediated endocytosis (heat shock protein 70, permease); MIC: transient membrane leakage (membrane potential-dependent) Vacuole (nonlethal) Mitochondrial F 1 F 0 -ATPase Buforin-2 TRSSRAGLQFPVGRVHRLLRK Amphipathic a-helix Transient toroidal pores DNA? Indolicidin [K6,8,9]-Indolicin ILPWKWPWWPWRRa ILPWKKPKKPWRRa Extended boat-shaped amphipathic structure No uptake Unknown DNA synthesis? Magainin-2 GIGKFLHSAKKWGKAFVGQIMNS Amphipathic a-helix Transient toroidal pores ? Polyphemusin I RRWCFRVCYRGFCYRKCRa Amphipathic b-hairpin with two disulfide bonds Transient pores? ? Tachyplesin I KWCFRVCYRGICYRRCR b-Hairpin with two disulfide bonds Transient pores DNA? Pleurocidin (P-Der) ALWKTMLKKAAHVGKHV GKAALTHYLa Amphipathic a-helix MIC: disordered transient pores > MIC: membrane permeabilization Macromolecular synthesis Cryptdin-4 GLLCYCRKGHCKRGERVR GTCGIRFLYCCPRR Triple-stranded b-sheet with three disulfide bonds Transient pores or defects ? Tritrpticin TWF TPA VRRFPWWWPFLRR VRRFPFFFPFLRR VRRFAFFFAFLRR Amphipathic turn structure Uptake not shown Uptake not shown ? ? P. Nicolas Intracellular-targeting antimicrobial peptides FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS 6487 tions is difficult or, at best, arbitrary. Representative examples will be used to elaborate this issue, starting with the more documented examples and moving towards those that are less documented. Pro ⁄ Arg-rich antimicrobial peptides Pro ⁄ Arg-rich antimicrobial peptides form a heterolo- gous group of linear peptides isolated from mammals and invertebrates that are predominantly active against Gram-negative bacteria. Members of this group include pyrrhocoricin, drosocin and apidaecin from insects, and the cathelicidin-derived peptides bactenecin, PR-39 and prophenin from mammals [36]. The mechanism of action by which these peptides kill bacteria involve a stereospecific interaction with a receptor ⁄ docking mole- cule that may be a component of a permease-type transporter system on the inner membrane, followed by translocation of the peptide into the interior of the cell. Once inside the cell, the peptides interact with the tar- get, which, for pyrrhocoricin and drosocin, has been clearly defined as the chaperone Dnak, or interfere with DNA and protein synthesis through binding to nucleic acids [37–39]. Interestingly, Bac-7 has recently been shown to inactivate bacteria via two different modes of action, depending on its concentration: (a) at near-MIC concentrations via stereospecific-dependent uptake that is followed by its binding to an unknown intracellular target, which may be DNA; and (b) at concentrations higher than the MIC via a nonstereospecific membran- olytic mechanism [40]. Histatin Histatin-5 is a 24 residue, His-rich and weakly aphi- pathic a-helical antimicrobial peptide found in human salivary secretions that displays high candidacidal and leishmanicidal activities at micromolar concentrations. Previous research has indicated that histatin-5 binds heat shock protein 70 (Ssa1 ⁄ 2), located on the cell wall, and is subsequently transferred to a membrane permease that transports the peptide across to the cytoplasm in a nonlytic manner [41]. Ensuing studies demonstrated that the uptake of histatin-5 is actually a dichotomous event [42]. Below the MIC, the peptide translocates into the cytoplasm of the parasite through receptor-mediated endocytosis (see above) and is inter- nalized into the vacuole without harmful effects on the parasite. Under physiological concentrations, histatin-5 induces a concentration-dependent perturbation at a spatially restricted site on the cell surface of Candida, leading to rapid translocation of the peptide into the cytoplasm in a nonstereospecific, receptor-independent manner, causing only a fast but temporary depolariza- tion and limited damage to the plasma membrane, as shown by membrane depolarization, entrance of the vital dye SITOX green, electron microscopy, and time- lapse confocal microscopy on live cells. Once inside the cell, the peptide accumulates in the mitochondrion, inducing bioenergetic collapse of the parasite, caused by the decrease of mitochondrial ATP synthesis through inhibition of F 1 F 0 -ATPase. Concurrent with the internalization and accumulation, rapid expansion of the vacuole with a parallel loss of cell volume is observed, leading to cell death. Histatin-5 shows poor translocation capacity in anionic liposomes. The dependence of histatin-5 internalization on the membrane potential may provide an explanation for a single rupture per cell, rather than multiple breaches, as once there is one site of leakage, the membrane potential is lost, and this prevents a second rupture. Buforin Buforin II is a 21 residue truncated analog of buforin I, the histone H2A-derived antimicrobial peptide, which adopts a helix–hinge–helix structure in apolar media [43]. Buforin II kills bacteria without lysing the cell membrane, even at five-fold the MIC. It binds selectively to negatively charged liposomes, and trans- locates even below the MIC across artificial bilayers efficiently via the transient formation of toroidal pores, without inducing significant permeabilization or lipid flip-flop. The induction of a positive curvature strain by the peptide on the membrane is related to the trans- location process [44,45]. Pro11 in the hinge region of the peptide plays a key role in the cell uptake mecha- nism by distorting the helix and concentrating basic residues in a limited amphipathic region, thus destabi- lizing the pore by electrostatic repulsion, enabling effi- cient translocation [46] Confocal laser fluorescence microscopy on living bacterial cells shows that, even below the MIC, the peptide penetrates the cell mem- brane and accumulates in the cytoplasm [47]. Although buforin II was shown to bind DNA in vitro, the con- nection between nucleic acid binding and antimicrobial activity has not been demonstrated. Indolicidin Indolicidin is a Trp-rich, 13 residue antimicrobial peptide isolated from bovine neutrophils that adopts an extended wedge-type conformation when bound to biological membranes. Owing to the presence of Trp residues interspersed with Pro residues throughout the sequence, it probably assumes a structure distinct from Intracellular-targeting antimicrobial peptides P. Nicolas 6488 FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS the well-described helical and b-structured peptides. Indolicidin is active against a wide range of microorgan- isms, including bacteria, fungi, and protozoa, and lyses erythrocytes. Close to the MIC, indolicidin causes sig- nificant membrane depolarization of the bacterial cyto- plasmic membrane by forming transient pores, but does not enter the cell and does not lead to cell wall lysis, suggesting that there is more than one mechanism of antimicrobial action [48]. Earlier investigations have shown that indolicidin mainly reduces the synthesis of DNA, rather than RNA and protein, and that inhibi- tion of DNA synthesis causes E. coli filamentation and contributes to the antimicrobial activity of indolicidin [49]. Unlike indolicidin, [K6,8,9]-indolicidin and [K6,8,9,11]-indolicidin do not depolarize the membrane and accumulate in the cytoplasm, as shown by confocal laser microscopy on living E. coli cells [50]. Gel-retarda- tion assays showed that [K6,8,9]-indolicidin and [K6,8,9,11]-indolicidin bind strongly to DNA in vitro, suggesting inhibition of intracellular functions via inter- ference with DNA ⁄ RNA synthesis. Whether indolicidin uses its membrane-binding properties to permeabilize the cytoplasmic membrane, activate extracellular targets or enter the cytoplasm and exert its antimicrobial activ- ity by attacking intracellular targets is presently unclear. Magainin Magainin-2, an a-helical peptide isolated from the Afri- can clawed frog Xenopus laevis, forms toroidal transient pores in the lipid bilayer of liposomes near the MIC, inducing lipid flip-flop and the translocation of peptides into the inner leaflet of the bilayer coupled to mem- brane permeabilization. Interaction of F5W-magainin- 2, an equipotent analog of magainin-2, with unfixed Bacillus megaterium was investigated by confocal laser microscopy [51]. At four times the MIC, magainin-2 binds to bacteria, permeabilizes the cytoplasmic mem- brane within seconds, and internalizes simultaneously. The influx of fluorescent markers of various size into the cytosol revealed that magainin-2 permeabilizes the bacterial membrane by forming toroidal pores with a diameter of  2.8 nm. However, there is no informa- tion available from which to evaluate whether magai- nin-2 disrupts key intracellular processes, and, if so, to what extent this may contribute to its killing action. Polyphemusin The horseshoe crab antimicrobial peptide polyphe- musin I is a 18 amino acid peptide that is stabilized into an amphipathic, antiparallel b-hairpin by two disulfide bridges [52]. It has excellent antimicrobial activity against bacteria, demonstrating rapid killing within 5 min of treatment. At two times the MIC, polyphemusin I is only able to depolarize the E. coli cytoplasmic membrane by 50% [53]. At the MIC, polyphemusin I is able to translocate through mem- brane bilayers of negatively charged model vesicles, inducing flip-flop between membrane leaflets. Biotin- labeled polyphemusin I accumulates in the cytoplasm of E. coli within 30 min after addition, with only modest cytoplasmic membrane disruption, and causes disorganization of cytoplasmic structures [54]. In these studies, permeabilization of E. coli with Triton X-100 was performed after fixation with glutaraldehyde, so as to allow streptavidin fluorescent conjugate to access intracellular biotin-labeled polyphemusin I. Moreover, the mechanism of translocation and the nature of the intracellular targets are as yet undefined. Tachyplesin Tachyplesin I is a cyclic b-sheet antimicrobial peptide of 17 amino acids isolated from the hemocytes of the horseshoe crab [55]. The peptide forms transient pores in membranes containing acidic phospholipids, and induces lipid flip-flop coupled to calcein leakage, the latter being coupled to the translocation of the peptide across lipid bilayers upon pore disintegration. The pep- tide induced rapid inner membrane permeabilization of E. coli at MIC, concomitant with a rapid decrease of cell viability [56,57]. Gel-retardation assays and foot- printing-like techniques using DNase I protection, dimethyl sulfate protection and bleomycin-induced DNA cleavage revealed that tachyplesin I interacts with the minor groove of the DNA duplex in vitro [58]. It is not known yet whether tachyplesin I is able to enter living cells, and whether its antibiotic activity is due to its capacity to bind DNA or to depolarize the cytoplasmic membrane. Pleurocidin Pleurocidin and dermaseptins are a-helical antimicro- bial peptides isolated from winterflounder and frog skin, respectively. When used at its MIC, the hybrid of pleurocidin and dermaseptin, P-der, inhibits E. coli growth, but does not cause bacterial death within 30 min, and demonstrates a weak ability to permeabi- lize the bacterial membrane [59]. When used at 10 times the MIC, the peptide causes rapid depolarization of the cytoplasmic membrane and cell death, indicating that the cell membrane is a lethal target for the peptide applied at high concentrations. Both sublethal and lethal concentrations of P-der inhibit macromolecular P. Nicolas Intracellular-targeting antimicrobial peptides FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS 6489 synthesis within 5 min. P-der is able to translocate across the lipid bilayers of liposomes without causing calcein leakage or flip-flop. It has been proposed that pleurocidin translocates in vitro from one side of the membrane to another through disordered transient pores, allowing the peptide to reach the cell interior [60]. As discussed above, membrane crossing remains to be shown in living bacterial cells. In addition, the relative contributions of intracellular targeting and membrane disruption to the overall killing strategy of pleurocidin, as well as the precise mechanism by which the peptide inhibits macromolecular synthesis in vivo, remain to be defined. Cryptdin Cryptdin-4 is a 32 amino acid amphipathic antimicro- bial peptide that adopts a triple-stranded antiparallel b-sheet structure constrained by three disulfide bridges. Near the MIC, cryptdin-4 induces E. coli cell permea- bilization coupled to rapid potassium efflux, a sensitive index of cell death. The lipid ⁄ polydiacrylate colorimet- ric assay and fluorescence resonance energy transfer from the Trp of the peptide to the dansyl chromo- phore in the membrane vesicles of various lipid com- positions suggested that cryptdin-4 inserts deep into the membrane of highly negatively charged PG-con- taining or cardiolipin-containing vesicles and then translocates via transient membrane defects to the inner membrane leaflet as a consequence of closure and disintegration of these short-lived formations [61]. Cardiolipin seems to be the key lipid constituent con- ferring sensitivity to cryptdin-4-induced vesicle permea- bilization. Because this lipid is able to form domains in E. coli cells, it was suggested that cardiolipin domains might serve as highly charged ‘gates’ to facili- tate movement of cryptdin-4 into and through lipid membranes. Although these studies provide evidence that the membrane disruptive action of cryptdin-4 is linked to peptide translocation through lipid defects, or pores, information on the internalization and the fate of the peptide within microbial cells, as well as the nature of the putative intracellular target, if any, needs to be provided to decipher whether the microbicidal activity of cryptdin-4 is due to its membrane permeabi- lization ⁄ disruption effect or to its ability to impede intracellular processes. Tritrpticin Tritrpticin consists of 13 residues and belongs to the cathelicidin family of antimicrobial peptides from the bone marrow of mammals. Tritrpticin has a broad spectrum of antimicrobial activity, and exhibits a high content of Trp (23%) and positively charged Arg ⁄ Lys residues (31%). It adopts a well-defined amphipathic turn–turn secondary structure in a membrane-mimetic environment (organic solvents or dodecylphosphocho- line micelles [62]. At high enough peptide concentra- tions, interaction of tritrpticin with membranes was postulated to cause positive curvature strain, which leads to toroidal pore formation membrane permeabili- zation and cell death in accordance with the Shai–Mat- zusaki–Huang model. In contrast, TWF, in which Trp is replaced with Phe, is highly potent against both S. aureus and E. coli, but shows very little membrane- disrupting activity and no ability to depolarize the membrane potential of the microbial cell targets [9]. Moreover, a lag period of about 3–6 h is required for bactericidal activity. It was thus suggested that TWF- mediated cell death occurs as a result of a nonmem- branolytic mechanism, but testing of this hypothesis awaits further investigation. A closer look shows that only a small number of the above-mentioned antimicrobial peptides have been convincingly demonstrated to fulfill the criteria to be considered as microbial cell-penetrating peptides that attack internal targets in vivo, and, of these, few spon- taneously cross the cytoplasmic membrane. For instance, in most cases: (a) the connection between intracellular target binding in vitro and antimicrobial activity has not been demonstrated, and ⁄ or the state of integrity of the membrane has not been checked – thus, it is not known whether the microbicidal activity of the peptides is due to their membrane permeability effect, their effects on intracellular targets, or a combi- nation of these effects; (b) although a substantial num- ber of these antimicrobial peptides have been shown to translocate through model membrane vesicles in vitro, detailed information on the internalization obtained with living cells, and quantification of peptide uptake and degradation, is still lacking – most of the confocal and electron microscopic studies reporting internaliza- tion of antimicrobial peptides have been conducted on fixed cells, and the possibility that the fixation changed the distribution of peptides cannot be ignored [63]; (c) if intracellular targeting exists, one would expect the peptide to evoke some degree of alteration of back- ground transcript profiles, even if the peptide is present at sublethal concentrations – this has seldom been evaluated [22,64,65]; (d) the possibility that antimicro- bial peptides interfere with the coordinated and highly dynamic functioning of membrane-bound multienzyme complexes, rather than killing through interaction with intracellular targets, has been largely ignored [22]; (e) several putative microbial cell-penetrating peptides are Intracellular-targeting antimicrobial peptides P. Nicolas 6490 FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS synthetic analogs of naturally occurring antimicrobial peptides that differ from the parent molecule by one or more amino acid substitutions – as it is well known that the microbicidal potency and selectivity of antimi- crobial peptides, as well as their membrane-bound structure and mode of action, are exquisitely sensitive even to single amino acid substitutions, the penetrating properties of the analog may not represent that of the parent peptide; and (f) most antimicrobial peptides that are proposed to attack internal targets exhibit an overrepresentation of one or two amino acids, i.e. Trp and Pro, His, or Arg, hence resembling cell-penetrating peptides (see below). Cell-penetrating peptides working as antimicrobial peptides, and antimicrobial peptides working as cell-penetrating peptides A substantial number of mammalian cell-penetrating peptides, including TP-10, pVEC, Tat, Pep-1, MAP, and penetratin, have the capacity to work as cell-pene- trating peptides or as antimicrobial peptides, the threshold between these two properties relying on the composition on the membrane and the peptide concen- tration (Table 2). Their microbicidal action is thought to be due to their ability to inhibit key intracellular functions by crossing the microbial membrane, rather than to create pores in the cell surface. Although this picture is accepted by most authors, because observa- tions of translocation in model membrane systems and in living bacteria for some cell-penetrating peptides might support the existence of uptake mechanisms governed by lipid-assisted pore formation, quantitative comparison of the uptake and antimicrobial effects of these peptides in bacteria and yeasts have demon- strated that their uptake route, intracellular concentra- tion, fate and microbicidal effects vary widely among peptides and microbial organisms. In several cases, the experimental protocols that have been used suffer from the same limitations as those mentioned above for antimicrobial peptides, preventing a clear conclusion to be drawn about the mechanism(s) by which these peptides exert their antimicrobial action. TP-10, a 21 amino acid deletion analog of the chi- meric cell-penetrating peptide transportan, causes rapid permeabilization of S. aureus cell membranes, followed by cell entry, dispersion throughout the cytoplasm, and subsequent death of the bacteria. pVEC, an 18 amino acid peptide derived from murine vascular endothelial-cadherin protein, MAP, and penetratin, has weak ability to depolarize the membrane potential of S. aureus cells and the calcein-entrapped negatively charged bacterial membrane-mimicking vesicles [66– 70]. The peptides internalize within these cell lines, but all were degraded to various extents inside the cells [68,69]. It was suggested that the microbial cell mem- brane permeabilization might not be the only mode of peptide uptake. For instance, the import route of pVEC by B. megaterium is consistent with two distinct uptake mechanisms: one operating via a transporter with high affinity and low capacity, which is sensitive to the chirality of the peptide and reminiscent of that of histatin-5; and another with low affinity and high capacity that could be caused by the membrane- permeabilizing activity of the peptide. Tat(47–58), an Arg-rich cell-penetrating peptide derived from the HIV-1 regulatory protein Tat, exhibits antimicrobial activity against Gram-positive and Gram-negative bacteria, and antifungal activity against Malassezia furfur, Saccharomyces cerevisiae and Tricho- sporon beigelii in the low micromolar range [71]. Tat showed no ability to depolarize the membrane potential of S. aureus cells and to leak calcein-entrapped nega- tively charged lipid vesicles. Tat peptide internalizes in the fungal cells and rapidly accumulates in the nucleus without causing visible damage to the cell membrane. The penetration pathway of Tat is independent of energy, time, and temperature. After penetration, the peptide blocks the cell cycle process of Candida albicans through arrest at G 1 phase. Pep-1 is a synthetic cell-penetrating peptide com- posed of an N-terminal Trp-rich domain and a C-ter- minal nuclear signal domain, KKKRKV [72], which kills E. coli and Bacillus subtilis in the low micromolar range, but has low activity against Salmonella, Pseudo- monas, and Staphylococcus. The peptide strongly inter- acts with negatively charged lipid bilayers, causing local perturbation and depolarization of the membrane potential, and crosses the membrane by a mechanism promoted by the transmembrane potential [73]. The mechanism of translocation is controversial. Deshayes et al. [74] proposed a transient transmembrane-pore- Table 2. Amino acid sequences of designed mammalian cell-pene- trating peptides with antimicrobial activity. Hydrophobic residues are in bold. a, carboxamitaded. Name Sequence Tat-[48–60] GRKKRRQRRRPQa pVEC LLILRRRIRKQAHAHSKa MAP KLALKLALKALKAALKLAa TP 10 AGYLLGKINLKALAALA Pep-1 KETWWETWWTEWSCPKKKFKVa Penetratin RQIKIWFQNRRMKWKKa P. Nicolas Intracellular-targeting antimicrobial peptides FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS 6491 like structure promoted by the a-helical conformation of the hydrophobic domain when it interacts with membranes. This was disputed by other groups, because no membrane leakage was observed. Conversely, the capacity to translocate across the mammalian cell membrane has been clearly demon- strated for some antimicrobial peptides. Confocal laser microscopy on fixed human cervical carcinoma HeLa and fibroblastic TM12 cells, and on live Chinese ham- ster ovary K1 cells, showed that magainin-2 permeabi- lized the cells, forming pores in the cell membrane that allowed the entry of a large molecule (diame- ter, > 23 nm) into the cytosol. Pore formation and subsequent cell entry are closely related to cell death. The peptide is internalized within a time scale of tens of minutes [44,51], and once it has entered the cell, accumulates in mitochondria and nuclei. The permea- bilization of Chinese hamster ovary cells was accompa- nied by extensive deformation, including membrane budding. Whether magainin-2 kills mammalian cells by dissipating membrane potential or damaging mito- chondria is presently unknown. Likewise, studies of buforin suggest a similar ability to translocate into mammalian cells, but by a temperature-independent, less concentration-dependent passive mechanism, and without showing any significant cytotoxicity [51]. These observations show that mammalian cell-pene- trating ability and microbial cell-permeabilizing ability can coexist within a single peptide, but the unifying rules that govern these two properties remain to be fully elucidated. Broadly speaking, evidence exists for two main, simultaneous mammalian cell-entering pathways, including direct penetration of peptides in parallel with different forms of endocytosis, the endocytosic path- way being a preferred form of entry of cell-penetrating peptides, at least when attached to bioactive cargo. The direct penetration mechanism remains elusive, and has long been thought not to involve membrane dam- age, because no indication of membrane disruption has been seen at relevant concentrations of peptide. How- ever, mammalian membrane disorganization associated with penetration is very difficult to observe, because the membrane repair response masks membrane distur- bance by mobilizing vesicles within seconds to patch any broken membranes [75]. Cell entry and ⁄ or membrane damage may be a common feature of some antimicrobial peptides and cell-penetrating peptides through very similar mecha- nisms. Cell entry may involve membrane phase ⁄ transient pores or long-lived transitions that can be dependent on peptide and membrane composition. Differences in membrane lipid composition, as well as differential lipid recruitment by peptides, may provide a basis for microbial cell killing on the one hand and mammalian cell passage on the other. For instance, the translocation properties of Arg-rich cell-penetrating peptides have been shown to be directly associated with the presence of Arg residues. Transmembrane crossing of these peptides is affected by their flexibility and am- phipathicity, and is critically dependent on the number and spacing of guanidinium groups [76]. In the case of Tat peptides, replacement of Arg with Lys, or with His or ornithine, strongly reduced the translocation ability [77]. Charge neutralization of the guanidinium groups through bidendate hydrogen bonding with the phos- phate groups of the bilayer is thought to be necessary for effective internalization into mammalian cells, and the efficiency of the peptide uptake is directly associated with the existence of a transmembrane potential and an appropriate balance between hydrophobicity and hydrophylic surface groups. Interestingly, bidendate hydrogen bonding of the guanidinium groups of prote- grin, an Arg-rich antimicrobial peptide, with the phos- phate groups of the bilayer was demonstrated to be crucial for insertion and pore formation of the peptide within bacterial membranes [78]. Molecular dynamic simulations of the Tat peptide crossing zwitterionic membranes suggest a mechanism of translocation that involves thinning of the membrane bilayers with increasing concentrations of Tat, owing to strong inter- actions between the guanidinium groups of the peptide and the phosphate groups on both sides of the mem- brane bilayers [35]. This is followed by the insertion of charged side chains into the bilayer. As the charged side chains enter the acyl core of the membrane, water also penetrates and solvates the charged groups, favouring the formation of a transient pore. Once the pore is formed, the Tat peptide translocates across the mem- brane by diffusing on the pore walls. The fast, tran- sient nature of the pore may explain why mammalian cell death because of membrane leakage was not observed with Tat [35]. This mechanism is highly reminiscent of the disordered toroidal pore-electro- poration mechanism proposed for some antimicrobial peptides. This suggests that general mechanisms that involve fluctuations of the membrane surface, such as transient pores and the insertion of charged side chains, may be common and central to the functions of both cell-penetrating peptides and antimicrobial peptides. Final comments There is a widespread acceptance that antimicrobial peptides, apart from their membrane-permeabiliz- Intracellular-targeting antimicrobial peptides P. Nicolas 6492 FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS [...]... constraints Intracellular-targeting antimicrobial peptides References 1 Hancock RE & Sahl HG (2006) Antimicrobial and host- defense peptides as new anti-infective therapeutic strategies Nat Biotechnol 24, 1551–1557 2 Yeaman MR & Yount NY (2007) Unifying themes in host defence effector polypeptides Nat Rev Microbiol 9, 727–740 3 Holzl MA, Hofer J, Steinberger P, Pfistershammer K & ¨ Zlabinger GJ (2008) Host antimicrobial. .. how antimicrobial peptides have multiple roles in immune defense Trends Immunol 3, 131–141 5 Brogden KA (2005) Antimicrobial peptides: pore formers or metabolic inhibitors in bacteria? Nat Rev Microbiol 3, 238–250 6 Otvos L Jr (2005) Antibacterial peptides and proteins with multiple cellular targets J Pept Sci 11, 697–706 7 Hale JD & Hancock RE (2007) Alternative mechanisms of action of cationic antimicrobial. .. Bozzi A, Di Giulio A & Rinaldi AC (2004) Effects of the antimicrobial peptide temporin L on cell morphology, membrane permeability and viability of Escherichia coli Biochem J 380, 859–865 Bhattacharjya S & Ramamoorthy A (2009) Multifunctional host defense peptides: functional and mechanistic insights from NMR structures of potent antimicrobial peptides FEBS J 276, doi:10.1111/j.1742-4658.2009 07357.x... limited to particular peptide families When looking for different parameters that could promote the cellular penetration properties of antimicrobial peptides, it is noticeable that microbial cellpenetrating antimicrobial peptides and antimicrobial cell-penetrating peptides have very distinct sequences, but, nonetheless usually share several characteristics, such as their high positive net charge, clustered... cationic antimicrobial peptides on bacteria Expert Rev Anti Infect Ther 6, 951–959 8 Henriques ST, Melo MN & Castanho MA (2006) Cellpenetrating peptides and antimicrobial peptides: how different are they? Biochem J 399, 1–7 9 Yang ST, Shin SY, Hahm KS & Kim JI (2006) Different modes in antibiotic action of tritrpticin analogs, cathelicidin-derived Trp-rich and Pro ⁄ Arg-rich peptides Biochim Biophys... antimicrobial peptides in living microbial cells with the help of state-of-the-art cell biological methods, together with the implementation of in vitro and in vivo biochemical assays to characterize their intracellular targets, should increase the panel of the so-called intracellular-targeting antimicrobial peptides However, it is unlikely that the specific abilities of some antimicrobial peptides to... membrane-targeting antimicrobial peptides Infect Immun 69, 4916– 4922 15 Tossi A, Tarantino C & Romeo D (1997) Design of synthetic antimicrobial peptides based on sequence analogy and amphipathicity Eur J Biochem 250, 549–558 16 El Amri C, Bruston F, Joanne P, Lacombe C & Nicolas P (2007) Intrinsic flexibility and structural FEBS Journal 276 (2009) 6483–6496 ª 2009 The Author Journal compilation ª 2009 FEBS 6493 Intracellular-targeting. .. 1200 antimicrobial peptides with different origins have been isolated or predicted Currently, there are only a handful of antimicrobial peptides in the literature that have convincingly been demonstrated to spontaneously enter microbial cells and, once inside the cell, to interfere with cellular functions Without any doubt, a case-by-case systematic analysis of the uptake, fate and integrity of antimicrobial. .. cationic antimicrobial peptides with model membranes J Biol Chem 276, 35714–35722 Pag U, Oedenkoven M, Sass V, Shai Y, Shamova O, Antcheva N, Tossi A & Sahl HG (2008) Analysis of in vitro activities and modes of action of synthetic antimicrobial peptides derived from an alpha-helical ‘sequence template’ J Antimicrob Chemother 61, 341– 352 Strahilevitz J, Mor A, Nicolas P & Shai Y (1994) Spectrum of antimicrobial. .. Ferre R & Castanho MA (2009) Antimicrobial peptides: linking partition, activity and high membrane-bound concentrations Nat Rev Microbiol 7, 245–250 Jean-Francois F, Castano S, Desbat B, Odaert B, Roux ¸ M, Metz-Boutigue MH & Dufourc EJ (2008) Aggregation of cateslytin beta-sheets on negatively charged lipids promotes rigid membrane domains A new mode of action for antimicrobial peptides? Biochemistry 47, . MINIREVIEW Multifunctional host defense peptides: intracellular-targeting antimicrobial peptides Pierre Nicolas Biogene ` se. cell-penetrating peptides (see below). Cell-penetrating peptides working as antimicrobial peptides, and antimicrobial peptides working as cell-penetrating peptides A