Human Antimicrobial Peptides in Bodily Fluids: Current Knowledge and Therapeutic Perspectives in the Postantibiotic Era

ABSTRACT

Antimicrobial peptides (AMPs) are an integral part of the innate immune defense mechanism of many organisms. Due to the alarming increase of resistance to antimicrobial therapeutics, a growing interest in alternative antimicrobial agents has led to the exploitation of AMPs, both synthetic and isolated from natural sources. Thus, many peptide‐based drugs have been the focus of increasing attention by many researchers not only in identifying novel AMPs, but in defining mechanisms of antimicrobial peptide activity as well. Herein, we review the available strategies for the identification of AMPs in human body fluids and their mechanism(s) of action. In addition, an overview of the distribution of AMPs across different human body fluids is provided, as well as its relation with microorganisms and infectious conditions.

1. INTRODUCTION

Human antimicrobial peptides (AMPs) represent approximately 10% of all curated AMPs catalogued to date.1 Human host defense peptides are an intrinsic part of the innate immune system and exhibit a broad activity spectrum against bacteria, fungi, viruses, and parasites (Figs. 1, 2, 3), but also toward mammalian malignant cells, and represent a relatively large group of biomolecules.

Figure 1.

Phylogenetic tree representing all bacterial, viral, and fungal species targeted by human antimicrobial peptides found in body fluids. Created using phyloT.

Figure 2.

Global network depicting the distribution of human antimicrobial peptides across biological fluids and their microbial targets. Rectangular nodes correspond to biofluids, circular nodes to target species, and each edge correspond to a given gene encoding one or more antimicrobial peptides. The thicker the edge, the stronger is the association of a biofluid to a pathogen, representing increased number of antimicrobial peptides defending against such pathogen in such biofluid. Also, pathogens represented by bigger nodes (e.g., Escherichia coli, Candida albicans, Staphylococcus aureus) represent those that are targeted by more antimicrobial peptides across different biofluids.

Figure 3.

Distribution of species targeted by human antimicrobial peptides throughout human body fluids. Only most represented species are included.

While AMPs can be antibacterial (ABPs), antifungal, antiprotist, antiviral, anticancer, antiparasitic, insecticidal, spermicidal, chemotactic, antioxidant, protease inhibitors, or even exhibit wound healing properties (Supporting Information Table S1), their scope of action overlaps considerably and some peptides show activity at several levels (Fig. 2).1, 2, 3, 4, 5 For instance, on the one hand, human plasma adrenomedullin‐derived peptides are active against multiple bacteria species, human urinary and gingival fluid calcitonin gene‐related peptide displays antimicrobial activity against several bacteria and fungi species, and human neutrophil peptides (HNPs)/defensins isolated from most biological fluids display activity against several bacterial, fungal, and viral species (Fig. 2). On the other hand, dozens of human AMPs known to date display antimicrobial activity against the same human colonizers or pathogens, including Escherichia coli, Pseudomonas aeruginosa, and Staphylococcus aureus (Figs. 2 and 3). Also, it is intriguing to observe that even though some AMPs display broad activity spectra, others seem to be anywhere from species‐ to kingdom‐specific (Fig. 2). While likely to be confounded by observational bias (whereby some species, e.g., E. coli, tend to be tested more frequently than others), this observation also reflects that the activity spectra of AMPs depend not only on their physicochemical properties (which group them according to unique families) but also on target properties and on the environment/biological fluid in which they are found. As the large majority of AMPs identified to date are ABPs (∼83%, Figs. 1, 2, 3), most concepts and examples provided in this review will be based on results from studies focusing on ABPs, closely followed by those derived from studies on antiviral and antifungal peptides. However, the same principles tend to apply across all AMPs.

AMPs are very potent cationic molecules displaying minimum inhibitory concentrations (the minimum concentration that prevents bacterial growth) as low as 1–4 μg/mL,6 which highlights their promise as broad‐spectrum antimicrobial agents. Furthermore, AMPs act more rapidly than conventional antibiotics and are not affected by the typical resistance mechanisms involving conventional drugs, making them very attractive for therapeutic purposes. In addition, AMPs are not mere microbicide agents, as their scope of action may encompass other functions, such as cellular development and immune system modulation. For instance, fetal keratinocytes express significantly more AMPs compared to neonatal and adult keratinocytes, despite the lower degree of exposition to the environment or to pathogenic agents. These elevated levels of AMPs during fetal stages of development are correlated with increased expression of the histone demethylase JMJD3,7 which in turn also plays a key role in several other fundamental processes, including chromatin organization, cell fate commitment, mesodermal, cardiac muscle, and endothelial cell differentiation, and hippocampus development.8, 9 Also, tear fluid human AMPs such as cationic antimicrobial protein 37 and thymosin‐β4 are known to stimulate corneal epithelial cell migration and proliferation in response to injury,10, 11 while lacritin is a glycoprotein that stimulates tear secretion and acts as a mitogen but which, once cleaved, also displays antibacterial activity.12 Thus, AMPs appear to be key mediators involved in several biological processes and to be intrinsically regulated in tandem with other fundamental processes, working together to maintain organic homeostasis. Accordingly, AMPs detain innumerous immune‐modulatory properties, namely cytokine modulation, chemotaxis, angiogenesis stimulation, as well as mast cells’ degranulation and consequent vasodilatation, fibrinolysis inhibition, and tissue injury prevention by inhibiting proteases and by reinforcing wound repair. Further information regarding the interplay of AMPs in the innate immune system as well as their alternative roles is given elsewhere,13, 14, 15 and is beyond the scope of this review. Herein, we will revisit the antimicrobial effects of human AMPs.

2. PRODUCTION AND ELIMINATION OF AMPs

Considering the wide variety of amino acid sequences, structures, mechanisms of action, and performed functions, it is perhaps surprising to discover a high degree of conservation and analogy between mechanisms controlling AMPs formation, regulation, and elimination. In fact, these processes are very dynamic, involving many substrates, precursors, proteases/peptidases, activators, and inhibitors. The majority of AMPs identified in humans appears to result from the cleavage of precursor proteins with other unrelated functions, rather than from the direct transcription and translation into biologically active peptides. Nevertheless, AMPs are an integral element of the innate immune system and not just a product of protein cleavage, representing a defense strategy against invading microorganisms.16 Moreover, as later exemplified, several AMPs resulting from fragmentation of precursor proteins act synergistically with these, in order to fight foreign microorganisms at several defense barriers (Fig. 3).

Bioactive peptides can result from the cleavage of proteins targeted for exportation to the extracellular milieu, where both peptides and precursor protein may play their primary functions. This process requires prior commitment of the precursor proteins to the secretory pathway, which is ensued by the presence of a signal peptide in its sequence.17 Once committed to the secretory pathway, peptides are modified at several locations and time points, beginning in the endoplasmic reticulum, then moving to the Golgi apparatus, traveling inside secretory vesicles, and ending up secreted upon fusion of these vesicles with the cytoplasmic membranes.17, 18

While prokaryotic microorganisms are known to produce AMPs by both gene‐encoded (e.g., microcins and bacteriocins) and ribosome‐independent (e.g., vancomycin and daptomycin) pathways,19, 20, 21 there is no knowledge of ribosome‐independent AMPs formation in humans. Even when an AMP is constitutively produced and secreted, its activity level is not constant, as it may be further modulated by the action of proteolytic enzymes. In contrast with some circular prokaryotic peptides, which have a peptide bond connecting the N‐ and C‐termini, which confers them an enhanced degree of stability, human AMPs are linear in nature and thus considerably more susceptible to proteolytic cleavage.22 Such susceptibility is exploited by a whole set of proteolytic endopeptidases and exopeptidases capable of mediating the formation and elimination of AMPs, including kallikreins, kininases, matrix metalloproteinases, cathepsins, and others.23, 24, 25, 26, 27

Interestingly, human AMPs are subject to the action of both human and bacterial enzymes. Examples of bacterial enzymes capable of modulating human AMPs formation and degradation include metalloproteinase aureolysin and glutamylendopeptidase V8 from S. aureus, elastase from P. aeruginosa, gelatinase from Enterococcus faecalis, and metalloprotease from Proteus mirabilis.28 Moreover, the regulation of human AMPs formation and elimination is made even more complex when one considers the presence of naturally occurring protease inhibitors, which may also present a tissue‐specific expression pattern and modulate the antimicrobial activity of naturally occurring AMPs.29 Lastly, not only peptides are inactivated and eliminated by hydrolysis, they are also intracellularly recycled and cleared through renal filtration or excretion.30, 31

3. SEQUENCES, STRUCTURES, AND PHYSICAL PROPERTIES

AMPs tend to be found as small molecules, generally composed of fewer than 50 amino acid residues and most frequently in the L configuration (left‐handed enantiomer, resulting from the distribution of functional groups around a central carbon atom). Despite mostly being cationic and short in sequence, human AMPs can vary anywhere between 10 and 150 amino acids (human neurokinin A consists of ten, while human RegIIIα consists of 149 amino acids). Most present a hydrophobic content below 60% and their net charge varies between −3 and +20 (β‐amyloid peptide and CXCL9, respectively).32

Furthermore, predicting and rationalizing how any particular AMP accomplishes its biological activity can be done only if its detailed structure–function information is available. Given this information, AMPs tend to be organized into groups, facilitating such rationale, and several criteria can be used to classify and sort AMPs, including origin, size, sequence, biological class, mechanism of action, and secondary structure. Of these, amino acid sequence and secondary structure are the most important factors to take into consideration,33 perhaps because these allow us to predict how a peptide will interact with other molecules.

A rough estimate of all AMPs catalogued to date (Supporting Information Table S2) indicates 14.1% to be α‐helices, followed by β‐sheets (3.7%), and peptides presenting with both helices and β‐sheets (3.7%). However, approximately 60% of all AMPs currently identified do not have a known 3D structure, while only 0.3% have a known structure that is neither a helix nor a β‐sheet. Excluding those that have unknown structures, 99.2% of all AMPs are either α‐helices and/or β‐sheets, suggesting a highly conserved structure for a highly primordial biological function.

Secondary structures of human AMPs may include α‐helices (e.g., LL‐37),34 β‐sheets (e.g., defensin 1),35 and extended peptides (e.g., indolicidin).36 Considering that β‐sheets and α‐helices in polypeptide chains typically contain three to ten amino acids per strand and 3.6 residues per turn, respectively, and that amino acid sequences can be virtually unlimited, one can easily envision that a given peptide can display a multitude of secondary structures and, thus, many diverse conformations (see 37 for protein secondary structures predictive approaches).

Among secondary structures, α‐helices (in addition to being the most frequently encountered in nature) are the most regular and predictable, and, consequently, the most extensively studied. AMPs with linear α‐helices are notably positively charged and amphipathic, but only adopt such secondary conformation upon binding to bacterial membranes.38 Therefore, their antimicrobial activity is directly dependent on the secondary structure.39 In contrast, β‐sheet AMPs possess well‐defined conformations due to the presence of disulfide bridges formed between thiol groups in cysteines. In these peptides, such bridges are often required for antimicrobial activity.40 Contrariwise, bovine lactoferricin was shown to be effective against E. coli 0111 even when such bridges were blocked by pyridylethylation.41

Other human AMPs, such as human β‐defensin 3 (hBD‐3), present both α‐helices and β‐sheets. In the specific case of hBD‐3, disrupting the three stabilizing disulfide bonds does not compromise its antimicrobial activity, but it does diminish its chemotactic properties. 42 Furthermore, the correlation of the conformation with peptides’ antimicrobial activity is not always as straightforward as one could imagine. For example, disruption of hBD‐1's disulfide bonds results in peptides with free cysteines at the carboxy terminal that show increased antimicrobial activity against the pathogenic Candida albicans and the Gram‐positive commensals Bifidobacterium and Lactobacillus.43

Moreover, linear extended peptides do not keep a fixed conformation in solution, and are enriched in one or more amino acids, particularly arginine, proline, and tryptophan. For instance, the AMP histatin found in human saliva is enriched in histidine residues,44 while prophenin is very rich in proline and phenylalanine residues.45

When it comes to loop rich peptides, the presence of many prolines makes an AMP much less likely to form amphipathic structures, adopting what is frequently referred to as polyproline helical type‐II (PPII) structures. These structures are specifically bound by SH3 domains, which are usually found in proteins that interact with other proteins and mediate assembly of specific protein complexes (upon proline‐rich peptides binding).46

Human antibacterial peptides (ABPs), which compose the largest group of human AMPs, exhibit a weighted average net charge of +5.8 per peptide, which is considerably above the average net charge of all AMPs identified to date. Also, despite being intuitive that the more positively charged an AMP is, the more potency it will present,47 human AMPs do not show greater potency or efficacy when compared to those of other species, suggesting that the environment and the presence of different amino acids (D and rare amino acids) also play a significant role.

Notwithstanding the large diversity of AMPs across human biologic fluids (Figs. 2 and 3, Table I), it should be noted that AMPs are much more diverse across all species, than those identified in humans, which present rather conserved properties. In fact, some AMPs from other species are characterized by more intriguing sequences, structures, and physicochemical properties. Case in point, Bacillus subtilis is known to produce several extremely potent lipopeptides (active at concentrations as low as 0.01–0.06 μM) with antimicrobial activity against several species. One of these, gageotetrin A is a very unique anionic AMP that consists of only leucine and glutamic acid residues conjugated with a unique fatty acid, 3‐hydroxy‐11‐methyltridecanoic acid.48 Another example is that of sonorensin from Bacillus sonorensis, a broad‐spectrum AMP from the heterocycloanthracin subfamily, active against both Gram‐negative and Gram‐positive bacteria, including Listeria monocytogenes and S. aureus. Sonorensin displays an amino acid sequence with multiple copies of the same motif, making this peptide rather unique.49 Baceridin is a circular peptide from a plant‐associated Bacillus that is synthesized by ribosome‐independent machinery and bears only six amino acids, 50% in the D configuration.50 Moreover, baceridin is a proteasome inhibitor, compromising cell cycle progression and inducing apoptosis in tumor cells by a p53‐independent pathway.50 Copsin from Coprinopsis cinerea is bactericidal against Enterococcus faecium and L. monocytogenes by interacting with the peptidoglycan precursor lipid II and interfering with the cell wall biosynthesis.51 Curiously, it is modified with a pyroglutamate, which confers a higher degree of thermal stability and resistance toward protease digestion.51

Table I.

Potential Sources of Human Antimicrobial Peptides found in Human Bodily fluidsa

^aHighlighted cells correspond to precursor genes that may also originate peptides with >100 amino acids displaying antimicrobial activity.

This article is being made freely available through PubMed Central as part of the COVID-19 public health emergency response. It can be used for unrestricted research re-use and analysis in any form or by any means with acknowledgement of the original source, for the duration of the public health emergency.

4. MECHANISMS OF ACTION

Even though some display fungicidal and virucidal properties, the chief activity of human AMPs is against Gram‐positive and Gram‐negative bacteria (Figs. 1, 2, 3). Furthermore, while AMPs represent an extremely diverse group of biological active molecules, most share common properties that contribute for their membranolytic or otherwise antimicrobial activity: positive charge, hydrophobicity, and amphiphilicity. Therefore, the mechanism of action of most AMPs involves membrane disruption (Fig. 4), a process that is also required when their translocation is warranted and the respective targets are localized intracellularly.39

Figure 4.

Mechanisms of action of antimicrobial peptides. Antimicrobial peptides may either (1) cluster at the cell surface and cause membrane disruption by several different mechanisms (e.g., barrel staves, carpets, toroidal pores), (2) translocate into cells and impair intracellular organelle machineries, (3) impair protein–protein interactions, enzymatic cascades, and cytosolic signaling pathways, (4) interact with nucleic (not in the case of bacteria) acids, trap replication forks, and compromise nucleic acids as well as protein synthesis, (5) preclude several steps of viral replication, (6) inhibit genetic material trafficking, reverse transcriptase, and viral proteases, (7) block the interaction between virus and host cells (e.g., viral envelope glycoproteins gp120 and gp41 or co‐receptor CXCR4), compromising virus binding and entry, and (8) cause membrane lysis on enveloped viruses. See text for detailed description. Designed using Servier medical Art.

The interaction of AMPs with bacterial membranes depends on the establishment of attractive electrostatic forces between these. Bacterial species are known to carry negatively charged outer membranes due to phosphate groups of the lipopolysaccharides in Gram‐negative bacteria and to the lipoteichoic acids in Gram‐positive bacteria.52 The fact that all Gram‐negative and Gram‐positive bacteria display these type of negatively charged lipids accounts for the lack of specificity of most ABPs, and promotes the attraction between AMPs and bacterial membranes while preventing their binding to most host cells membranes. Moreover, the inherent hydrophobicity and amphiphilicity allow ABPs to form clusters at the membrane's surface once attached.53, 54 At low concentrations (low peptide‐to‐lipid ratio), ABPs tend to adsorb onto the surface, adopting an orientation parallel to the membrane bilayer. However, clustering increases the peptide‐to‐lipid ratio, which subsequently promotes additional clustering. Once a certain peptide‐to‐lipid ratio threshold is reached, ABPs adopt a perpendicular orientation relative to the bacterial membrane, which allows them to insert themselves into the membrane,55 as will be further discussed (Fig. 4).

Peptides can adopt many different secondary structures and conformations, which depend on their amino acids sequence, as well as on the environment. Consequently, different peptides and different environments favor the formation of unique structures, such as barrel staves, carpets, and toroidal pores.40, 56, 57 When the peptides attach and are inserted into the membrane bilayer so that the hydrophobic regions align with the phospholipids’ acyl groups and the hydrophilic regions create the central region of a pore, a barrel stave is formed. In addition, these structures act as a primer for the aggregation of new peptides, thus expanding the pore's diameter.58 Due to ABPs’ cationic sites, membrane crossover would be very energetically unfavorable in their original parallel orientation. However, this orientation changes to a perpendicular one upon ABPs’ clustering, and the peptides are reoriented with their hydrophobic portions facing the membrane acyl groups and their cationic sites facing the interior of the pores. Thus, clustering and pore formation provides a means for membrane disruption and the leakage of intracellular molecules from bacterial cells.59

In contrast, if peptides remain parallel to the membrane surface with their hydrophobic residues facing the membrane while their hydrophilic residues face the surrounding milieu, a “carpet”‐like structure is achieved. With such organization, membrane's rigidity weakens progressively as new peptides are added, culminating in its dissolution and breakup. Mechanistically, this mode of action is less stringent because peptides do not need to adopt a specific structure. However, the necessary concentration of ABPs appears be much higher than that required for the formation of barrel staves.60, 61, 62

For toroidal pores to form, peptides need to aggregate onto the membrane's surface and insert themselves in a perpendicular fashion.63, 64 Subsequently, ABPs are reoriented parallel to the membrane so that polar residues interact with the polar heads of the membrane lipids, thus causing membranes to bend. This induced bending of the bilayer causes the upper and lower leaflets to meet, and allows a mixture of peptides and lipid head groups to line with the interior of the pore, forming toroidal or wormhole structures.60, 63, 65

ABPs may also inhibit bacterial growth and have a bactericidal effect by promoting the clustering of anionic lipids, such as phosphatidylglycerols (PGs) and cardiolipin. For instance, for LL‐37, initially believed to act according to the “carpet” model, there is evidence that its fragments, namely KR‐12, act as a magnet that competes for the negatively charged PGs, promoting their redistribution into “PG‐rich domains.” Such phospholipidic reorganization may disturb cell signaling and compromise the activity of membrane‐bound proteins, namely, the voltage‐dependent potassium channels, which can seriously jeopardize bacteria survival.34 Likewise, N‐acylated peptides derived from lactoferricin were shown to induce defects in E. coli cell division by rearranging the distribution of inner membrane phospholipids, essentially due to the clustering of cardiolipin and other anionic lipids.66

The bias toward positive charges confers AMPs the ability to easily interact not only with biological membranes, but also with other negatively charged molecules (Fig. 4) such as nucleic acids and those bearing phosphate groups.39, 67 In addition, because this type of interaction is largely unspecific, AMPs can exhibit not only a broad‐spectrum activity, but also multiple mechanisms of action (Fig. 3).68, 69 For instance, lactoferricin is known to inhibit ATP‐dependent multidrug efflux pumps as well as DNA, RNA, and protein synthesis.70 Ten human AMPs derived from lactoferricin and lysozyme are thought (by sequence homology) to act by trapping the replication fork and preventing base recombination and DNA repair.71 Furthermore, the presence of many prolines in some peptides (also known as PPII structures as aforementioned) hampers the organization of amphipathic structures.72 These structures favor interactions between peptidic structures and between these and nucleic acids, thus playing a major role in signal transduction and complexes assembly.73 However, PPII structures do not necessarily contain prolines,74 and ABPs rich in arginine residues are also able to translocate across membranes and to interact with nucleic acids and proteins. Hence, both proline and/or arginine‐rich ABPs may be capable of inhibiting prosynthetic signaling pathways and enzymatic activity (Fig. 4), shifting the conformation of cellular structures and activating autolysins. However, this is a speculation that warrants further studying.

In addition to targeting bacterial membranes, human defensins also bind to the peptidoglycan precursor lipid II and inhibit cell wall biosynthetic enzymes in Staphylococci species.75 Moreover, when stabilizing disulfide bridges between conserved cysteine residues in human AMPs with β‐hairpin or β‐sheet conformations are disrupted, the resulting linear peptides still maintain their antimicrobial properties despite losing membranolytic activity.76, 77

The antifungal histatins act at multiple levels by mechanisms of action conserved across the histatin family of AMPs, with histatin 5 the most potent AMP in this family. This AMP binds the yeast transmembrane receptor heat‐shock protein Ssa1/2p, is internalized, and then targets primarily yeast mitochondria.78, 79 By doing so, it leads to the formation of reactive oxygen species (ROS), causes ATP efflux, and inhibits oxidative phosphorylation.80, 81 Simultaneously, histatin 5 also interacts with the potassium transporter TRK1p, causing release of K⁺ ions and further ATP efflux.82 When the functional domain of histatin 5 was synthetically multiplied, its antifungal activities were significantly potentiated and it became significantly faster‐acting compared to the normal histatin 5.82 Such an observation might prove itself very useful, because it indicates that modulating the number of functional domains per AMP may enhance its antimicrobial activity without requiring the administration of increased quantities of AMP. Histatins may also exert their antimicrobial activities by inhibiting host and microorganismal proteolytic enzymes. Histatin 5 is a strong inhibitor of human matrix metalloproteases 2/9 and metalloproteases from Porphyromonas gingivalis.83 Thus, its inhibitory activity over proteolytic enzymes may attenuate tissue damage and microbial propagation during the onset of periodontal inflammatory disease. Also, by inhibiting the cysteine protease clostripain from Clostridium histolyticum, histatin 5 mitigates the virulence factors produced by this Clostridium species and may therefore alleviate the complications associated with gas gangrene syndrome.84

So far, the above‐mentioned mechanisms of action have been discussed in the context of ABPs or antifungal peptides, and sometimes the same peptide is active against both bacteria and fungi. In turn, the action of antiviral peptides can diverge from these mechanisms and should thus be independently commented (Fig. 4), although some overlap does exist.

Extracellularly, most native antiviral peptides often present as α‐helical structures, have primarily lytic activity on enveloped viruses, and can directly inactivate cell‐free virions (Fig. 4). Human lysozyme fragments are this group's prototype, capable of aggregating on the viral surface and causing disruption of membrane envelopes.85, 86 Similarly, CAP37 peptide appears to exert its antiviral activity by destructing the virus envelope of type I herpes simplex virus (HSV) or even by disrupting the capsids of adenoviruses.87 However, because membrane lipids of enveloped viruses derive from host cell membranes, special attention is required when considering the exploitation of this feature, as antiviral peptides may disrupt host cell membranes indiscriminately.

Alternatively, antiviral peptides may block the interaction between the virus and the host cell (Fig. 4). That is believed to be, at least in part, the mechanism of action of human defensins, θ‐defensin, human α1‐antitrypsin VIRIP, cyclic lactoferricin, and serine protease inhibitor elafin. For instance, it has been demonstrated that all α‐defensins [HNP‐1, HNP‐2, HNP‐3, and HNP‐4 and human α‐defensin (HD)‐5 and HD‐6] and hBD‐3 prevented HSV infection by blocking virus binding and entry. Such antiviral activity was attributed to HNP‐1–3, HD‐5, and hBD‐3's high affinity to HSV's glycoprotein B and HNP‐4, HD‐6, and hBD‐3's avidity toward endogenous heparan sulfate.88 Similarly, protection against HIV is conferred by peptides with lectin properties that can bind viral envelope glycoproteins gp120 and gp41, crucial for viral attachment, fusion of envelopes with host membranes, and subsequent entry into host cells.89 HNP‐4 that binds and blocks the action of gp120, and θ‐defensins (e.g., the pseudogene retrocyclin) together with the human α1‐antitrypsin‐derived 20‐residue viral‐inhibitory peptide both targeting gp41 represent examples of such peptides.90, 91, 92 Likewise, HSV‐2 infection can also be blocked by retrocyclin, which binds tightly to glycoprotein B2.93 The cyclic peptide lactoferricin seems to compete for the ligation to extracellular heparan sulfate or chondroitin sulfate, thus preventing the very first event in cytomegalovirus, HSV, and papillomavirus infection.94, 95, 96 Finally, elafin is thought to act indirectly in host cells by hampering viral attachment and probably by modulating host's antiviral and inflammatory responses toward HSV‐2 infection.97 Synergistically with this mechanism, antiviral peptides can also bind and inhibit the activation of viral co‐receptors, leading to further inhibition of glycoproteins‐mediated viral attachment. Not only that, HBD‐2 and ‐3 inhibit viral infection by promoting the internalization of its co‐receptor CXCR4.98, 99 Virucidal peptides may also preclude several steps of viral replication by interacting with intracellular targets (Fig. 4). This is the case of HNP‐1 and HD‐5, which may prevent HSV replication by binding directly to its DNA.88 Also, HNP‐1 arrests influenza virus replication by inhibiting protein kinase C phosphorylation, an essential step for nuclear trafficking events.100 Human cathelicidin (LL‐37) derived peptides, in turn, inhibit reverse transcriptase and viral proteases, thus preventing integration of the viral genetic load into human DNA.101, 102

5. RESISTANCE MECHANISMS AGAINST ABPs

The few resistance mechanisms against AMPs demonstrated so far among microorganisms have been observed almost exclusively in bacteria. With very few exceptions, bacteria do not seem to develop complete resistance against cationic ABPs, though some strategies do exist among bacterial species for diminishing their potency and efficacy. The difficulty in acquiring resistance may stem mainly from the fact that ABPs target an essential cellular component for bacteria survival, which cannot be easily modulated by the cell without causing significant adverse effects. Therefore, resistance mechanisms are not likely to be a limitation in the therapeutic utilization of ABPs. In light of this scarce resistance mechanisms de novo acquisition, what mechanisms do bacteria naturally develop/possess against ABPs?

First, one should consider that the primary target of ABPs consists of bacterial cell membranes and bear in mind how ABPs depend on the electrostatic interactions between membrane negative charges and positive peptidic residues. Accordingly, bacteria have learned how to modulate cell membrane charges and composition. Gram‐positive bacteria can partially modify the peptidoglycan's teichoic acid and Gram‐negative bacteria can partly change its lipopolysaccharide lipid A moiety, thus compromising this interaction. For instance, D‐alanine‐activating enzyme and D‐alanine transfer protein are known to transport positively charged D‐alanines to the surface anionic teichoic acids in Gram‐positive S. aureus, reducing the negative net charge of its outer membrane.103 Also in S. aureus, the five‐component system GraXSR‐VraFG increases the expression of mprF and dltABCD genes upon exposure to cationic AMPs. When S. aureus strains are mutated so that the regulation of these genes is disturbed, these strains become more susceptible to daptomycin, polymyxin B, HNP, and platelet factor 4‐derived peptide, and less infectious in vivo in a nonclinical endocarditis model.104, 105, 106 The gene product MprF attaches lysine residues, which bear a positively charged ε‐amino group, to the anionic phophatidylglycerols, making these less negatively charged and thus less susceptible to interactions with cationic AMPs.105, 106

The Gram‐negative Vibrio cholerae can increase its resistance to polymyxins more than 100‐fold by modifying its lipopolysaccharides in a process mediated by AlmF.107 The lptA gene product found in gonococcal species, such as the Gram‐negative bacteria Neisseria gonorrhoeae, protects lipopolysaccharide lipid A moieties by attaching phosphoethanolamine, making these bacteria less susceptible to AMPs and increasing their capacity to modulate host's immune response by evading complement‐mediated cellular death.108, 109 Moreover, in N. gonorrhoeae and Neisseria meningitides, the presence of phosphoryl and phosphoethanolamine in lipid A moieties enhances the activation of the NF‐κB pathway and the production of proinflammatory cytokines in human cells, which can amplify their virulence.110 Despite this modification‐dependent virulence, the absence of such modifications in commensal Neisseriae species allow these to colonize and coexist with the human host without inducing bactericidal host immune responses.111 In addition, the adaptation of the composition and, hence, the electrostatic properties of the membrane in order to circumvent the host immune system is not an exclusive phenomenon of pathogenic bacteria. Commensal bacteria of the intestinal tract can remove phosphate groups from their membrane's lipopolysaccharides, decreasing their negative charge and allowing greater survival and proliferation in the gut.112 Similarly to the above‐mentioned paradoxical modification‐dependent virulence, mutated commensal gut microbiota Bacteroides that fail to remove phosphate groups from their lipopolysaccharides are killed during inflammation while normal strains keep their resilience.112 Therefore, there appears to be an inverse relationship between acquired resistance and bacterial virulence, so that induced/acquired modifications render cells less virulent while simultaneously allowing these to more easily colonize and coexist within the human host environment.

Second, bacteria are also able to adapt the fluidity of the membrane by increasing the hydrophobic interactions within their outer membrane, thus hindering pore formation.113

Third, bacteria can express proteases against ABPs or transporters and efflux pumps like ATP‐binding cassette (ABC) transporters to circumvent such innate defense barriers.114, 115 Glu‐C protease produced by S. aureus induces the loss of antibacterial activity of neuroendocrine peptides, and chromofungin, procatestatin, and human catestatin are enzymatically degraded when treated with bacterial supernatants.116 In E. faecalis, bacitracin binds the ABC transporter EF2050‐2049, which interacts with the regulatory domain of the two‐component system EF0926‐27, increasing its expression and conferring resistance against bacitracin.117 Two‐component systems regulate the expression of ABC transporters and are constituted by a transport permease and a sensor kinase. However, at least in B. subtilis, the ABC transporter itself is required for both sensing of and resistance to bacitracin, suggesting these transporters to also function as environmental sensors.115 Lastly, bacteria may also synthesize molecules capable of directly neutralizing ABPs,118 which can provide a way for bacteria to bypass AMPs. However, such resistance mechanisms are very limited and resistance to AMPs does not take place to a large extent.

6. FAMILIES OF AMPs IN HUMAN BODILY FLUIDS

In humans, endogenous AMPs have been found in several fluids throughout the body, as depicted in Figures 2 and 3 as well as in Table I. As noted, multiple AMPs have been found in tears, saliva, milk, blood, urine, sweat (and others not‐so‐well‐explored biofluids), reflecting not only a remarkable versatility but also a huge conservation that reinforces its essential role. The robust knowledge of AMPs on some fluids (e.g., tears, urine, saliva) over others (e.g., amniotic fluid, cerebrospinal fluid) most likely reflects their availability, but different exposure to pathogens from the environment (Figs. 2 and 3) should also account for these discrepancies to a large extent. Additionally, saliva and urine screening is of special interest, since both the oral cavity and the urinary tract represent the main doorways for microbial invasion and colonization, and are thus enriched with host defense peptides. Likewise, milk screening is of high relevance due to the presence of particular proteins and AMPs that will boost neonate's immunity. Despite underrepresented, AMPs from other fluids are equally relevant (Figs. 2 and 3). For instance, AMPs collected from airway secretions, bronchoalveloar lavages, and nasopharyngeal aspirates can be valuable sources of information regarding the physiological response to pathogens of the respiratory tract and may, themselves, constitute raw models for new therapeutic drugs. Also, other than circulating immune cells derived, blood has not been regarded as a rich source of AMPs, but this view can be questioned (Figs. 2 and 3), most likely due to tissue‐derived AMPs’ passage to the blood circulation. However, it should be noted that human fluid reservoirs are not independent/isolated, and, consequently, peptides may be present throughout and cross between, possibly undergoing modifications between such reservoirs. AMPs may be present in fluids as a result of many different processes, including exocytosis, in situ proteolysis, tissue necrosis, and cellular lysis. Therefore, in this section, we will characterize naturally occurring AMPs based on structural and functional similarities rather than their source of collection.

A. Defensins

Defensins are cationic peptides rich in cysteine residues, presenting as β‐sheet structures stabilized by disulfide bridges.4 According to the alignment of these stabilizing disulfide bounds, mammalian defensins are classified into three subclasses, α‐defensins, β‐defensins, and θ‐defensins.

A.1. α‐Defensins

Disulfide bounds of α‐defensins occur between cysteines 2–6, 1–4, and 3–5. These peptides consist of 29–35 amino acids and adopt a triple‐stranded β‐sheet conformation. Human neutrophils express four distinct α‐defensins, known as HNP‐1 to HNP‐4. Despite having been initially isolated from peripheral blood leukocytes, HNP‐1 to HNP‐3 can also be found in bone marrow, spleen, and thymus.119 The remaining two already identified human α‐defensins, HD‐5 and HD‐6, are found only in Paneth cells of the small intestine and in epithelial cells, thus being tissue specific.120 Expressed in the aforementioned cell types, AMPs can also be found in bodily fluids in direct contact with these cells. Among the most recent discoveries, HD‐6 has been found to be active against Bifidobacterium adolescentis and other anaerobic gut commensals, but not (directly) bacteriostatic or bactericidal against most pathogenic bacteria, and may therefore play an important role in maintaining a balance among the microbiota of the gut.121 Similarly, human HD‐5 is particularly potent against the Gram‐positive bacillus Clostridium difficile, one of the most common pathogens responsible for nosocomial infections, whose highly virulent strains often attack small intestine mucosa.122 In addition, α‐defensins may show both cellular and regional specificity throughout the gastrointestinal tract, either secreted or active inside intracellular granules and constitutively expressed or amenable to induction.15, 123, 124 It should also be emphasized that the activity spectrum of α‐defensins is not limited to bacteria and fungi, exhibiting activity against several virus, including HIV, HSV, Adenovirus, and human papillomavirus (HPV), and displaying several unique mechanisms of action (please refer to Section 4.).4 It should be noted that the antimicrobial activity of α‐defensins is influenced even by the difference in one amino acid, such as for HNP‐1 to HNP‐3. In fact, an additional acidic residue in HNP‐3 lowers its effectiveness (most accurately, its potency) against S. aureus, P. aeruginosa, and E. coli. This is explained by the easily donated proton, which makes this defensin less positively charged (at least when considering interaction domains only), thus reinforcing the importance of a positive net charge for the antimicrobial activity of AMPs,125, 126 as evidenced in Section 4..

In terms of disease, α‐defensins are known to have a paramount role, particularly when the demands for defense mechanisms are increased. Accordingly, a remarkable 15‐ to 25‐fold increase in the concentration of HNP‐1 and HNP‐2 (and HNP‐3 to a smaller extent) in human tears as been reported after ocular surgery, an increase that allowed these peptides to reach the required concentrations for antimicrobial activity but that is followed by a decrease once healing has been completed and the demands for immune defenses have returned to baseline.127 Nonetheless, the goal of achieving this increased α‐defensin concentration in tears is not restricted to the enhancement of the immune defenses, as it is also known to promote local cellular proliferation and tissue repair.128 In parallel, salivary α‐defensins are active against Streptococcus mutans, but their secretion has been found deregulated in caries‐positive human subjects.129 In contrast, prolonged physical exercise has been shown to significantly increase the levels of salivary α‐defensins,130 showing that lifestyles may have a considerable impact on our AMP‐based defenses against microorganisms and it is possible to modulated the activity thereof.

A.2. β‐Defensins

Disulfide bounds of β‐defensins occur between cysteines 1–5, 2–4, and 3–6. These defensins are longer than α‐defensins, spanning from 36 to 50 amino acids, but also share a triple‐stranded β‐sheet conformation. The first hBD was isolated from plasma samples and kidneys constitute the major source of hBD‐1.131 Since then, peptides derived from human hBD‐1 have been identified in tears, urine, vaginal lavage, sweat, and blood plasma, and these are effective primarily against E. coli.131, 132, 133, 134 In turn, hBD‐2 was isolated from psoriatic skin samples,135 displaying bactericidal activity toward Gram‐negative E. coli, P. aeruginosa, and C. albicans, and bacteriostatic activity toward Gram‐positive S. aureus.

Despite very effective in vitro, both hBD‐1 and hBD‐2 present attenuated activities in vivo as a result of their sensitivity to physiological salinity. In contrast, hBD‐3 is active against S. aureus and vancomycin‐resistant E. faecium even at physiological salt concentrations.136

The hBD‐119 defensin has been found in seminal plasma, and it has been reported as potential efficacious against E. coli, S. aureus, and C. albicans.137, 138 Furthermore, hBD‐114 has been identified in epididymal fluid and has been shown to regulate lipopolysaccharide‐mediated inflammation toward E. coli, S. aureus, and C. albicans, protecting sperm from motility loss.139 In disease conditions, increased concentrations of hBDs in plasma and bronchoalveolar lavage fluid were reported in patients with diffuse panbronchiolitis. In particular, β‐defensin 4A was suggested to play a role against P. aeruginosa.140 Also, while β‐defensins are expressed throughout the gastrointestinal tract, these seem to have evolved together with changes in human gut bacteria, displaying characteristic expression patterns in response to certain microorganisms. Their expression may be significantly altered in human diseases, most notably in inflammatory bowel diseases (for insightful reviews on this topic, please refer to16, 141 and enhanced in inflammatory conditions of the ocular surface134).

With α‐defensins, the role of the pH for AMPs’ activity is considered. Here, β‐defensins serve to illustrate the role of salt concentration. In fact, tears contain a remarkable concentration of sodium chloride, which inhibits the activity of human AMPs (by ∼40% in the case of hBD‐2, an inhibition that increases up to 90% in the presence whole tear fluid, suggesting the presence of other inhibiting factors).142 Therefore, whenever there is an increase in the demand of AMPs activity, (e.g., upon infection or tissue damage), their concentrations must surge to reach the required protective levels. However, an excessive increase in the concentration of any AMP may lead to tissue damage and local functions may be compromised. For this reason, a complex interplay of AMPs is required at surfaces like the ocular epithelia, which is achieved by a qualitative enrichment, in addition to the expected quantitative increase (for a depiction of the multitude of AMPs found in human tears, please refer to 143).

Similarly to α‐defensins, the activity spectrum of β‐defensins is not limited to bacteria and fungi, and they are active against several virus, namely, HIV, influenza A virus (IAV), respiratory syncytial virus, and vaccinia virus, and display several unique mechanisms of action, depending on their target (see above section on mechanisms of action).4

Lastly, genomic studies have predicted additional β‐defensins to be expressed in human cells, but their presence in human bodily fluids and their antimicrobial activity in vivo is yet to be demonstrated.144

A.3. θ‐Defensins

While α‐ and β‐defensins adopt triple‐stranded β‐sheet conformations, θ‐defensins have shorter sequences and adopt a complete circular conformation, with stabilizing disulfide bounds between cysteines 1–6, 2–5, and 3–4. The peptide rhesus θ‐defensin 1 is the prototype of this more recently discovered family of AMPs. Like α‐ and β‐defensins, θ‐defensins possess broad‐spectrum microbial activity against bacteria and fungi, while also being capable of protecting mononuclear cells from infection by HIV‐1.145, 146 Moreover, these peptides display antiviral activity against HSV, HIV, IAV, and severe acute respiratory syndrome (SARS) coronavirus, acting by several mechanisms, depending on their target.4 Structurally, θ‐defensin peptides are very interesting and very odd, representing the first cyclic peptide family discovered in animals, albeit not in humans. Moreover, their circular structure appears to be required for antimicrobial activity.145 These peptides were first isolated as antimicrobial octadecapeptides expressed in leukocytes of rhesus monkeys, and their formation results from the head‐to‐tail joining of two independent nine‐amino acid peptides derived from truncated pro‐α‐defensins.146

In contrast to the previously described classes of defensins, human θ‐defensins are thought to be completely inactivated, despite the existence of mRNAs encoding at least two θ‐defensins expressed in the bone marrow.145 While genes encoding θ‐defensins in humans also encode premature stop codons (thus hampering θ‐defensin expression), synthetic human θ‐defensins (called retrocyclins) have been shown as promising antiviral agents.147 In fact, retrocyclin‐1 inhibits the entry of HIV, HSV, and IAV into host cells. In other mammals, retrocyclin‐1 also protects from infection by Bacillus anthracis spores and the rhesus θ‐defensin 1 protects mice from SARS coronavirus infection.147

B. Cathelicidins

Cathelicidins are a very diverse group of cationic α‐helical and amphipathic AMPs that exhibit a broad‐spectrum activity against bacteria, fungi, and virus.54 The term “cathelicidin” originally referred only to the entire precursor protein, though currently is commonly used for the whole family, including the resulting peptides with antimicrobial activity.

In contrast to the mammalian trend in which cathelicidins and defensins are the main AMPs, there is only one human gene encoding multiple cathelicidin‐related peptides,144, 148 located on chromosome 3 and expressed in the airways, mouth, tongue, esophagus, epididymis, and small intestine.149, 150 Despite this tissue expression pattern, LL‐37 is constitutively formed in spleen, liver, stomach, intestine, and bone marrow. LL‐37 is highly positively charged (+6 charge at physiological pH 7.4), due to the high content of arginine and lysine amino acids, and adopts an α‐helical structure in solutions with ionic composition similar to that of human plasma.151

While defensins share common structural features, cathelicidin‐related peptides are highly heterogeneous despite deriving from the same precursor. Cathelicidins are characterized by a highly conserved N‐terminal signal peptide (the so‐called “cathelin domain”) and a highly variable C‐terminal antimicrobial domain that can be released after cleavage by proteinases.148, 152 Cathelicidin is cleaved into the antimicrobial peptide LL‐37 by both kallikrein 5 and kallikrein 7 serine proteases, it is upregulated by vitamin D, and has been shown to significantly reduce the risk of death from infection dialysis patients.153 LL‐37 was shown to disrupt bacterial membranes through the formation of toroidal pores and carpet structures,63 and to exhibit chemotactic properties, attracting leukocytes and activating secretion of chemokines. In addition, LL‐37 is internalized by cells, acidified in endosomes, and activates the signaling pathway downstream to Toll‐like receptor 3 by interacting with double‐stranded RNA.154

Profiling of airway fluids collected from premature infants’ tracheal aspirates during infection evidenced the production of LL‐37 even at an early stage of development.155 In these samples, LL‐37 was found in significantly increased concentrations during pulmonary or systemic infections, suggesting this peptide to be as important to avoid infectious processes as to fight already established infections.155

The concentration of LL‐37 has been measured in seminal plasma samples from healthy donors and found to be up to 70‐fold higher than in blood plasma.149 Furthermore, it was found to be attached to spermatozoa, eliciting a possible role in human fertilization or its precursor to be extensively cleaved by the high concentration of serine proteases present in seminal plasma.149

LL‐38, an alternative form of human cathelicidin peptides, contains one more alanine at the N‐terminus than LL‐37 and was initially found to be produced in female vaginal secretions due to the action of gastricsin on sperm precursor cathelicidin.156 Both LL‐37 and LL‐38 are active against bacteria such as E. coli, S. aureus, P. aeruginosa, and Bacillus megaterium, and synergistically with peptides from seminal plasma play an essential defense role protecting the human reproductive system.156, 157 Interestingly, both citrullination and ADP (Adenosine diphosphate)‐ribosylation of arginines can compromise the ability of LL‐37 to prevent endotoxin‐induced sepsis, perhaps because the attachment of these negatively charged molecules reduces the peptide's cationicity or by raising steric hindrance.158, 159

C. Histatins

Histatins are a normal component of human saliva and part of the innate immune system, protecting against a broad array of infectious agents, especially against Candida species, Saccharomyces cerevisiae, Cryptococcus neoformans, and Neurospora crassa.160, 161, 162 There are only two genes enconding histatins (both located on chromosome 4), and they are exclusively expressed in human salivary glands.163 However, other active AMPs resulting from the cleavage of histatins 1 and 3 also exist.164 Histatins 1 and 3 are encoded by genes htn1 and htn3, respectively.165 Histatins 2, 4, and 5–12 are the products of posttranslational proteolytic cleavage of histatins 1 and 3, though histatin 5 can also result from posttranscriptional modification of histatin 3 mRNA.165, 166

Peptides derived from histatin 1 bear distinct domains associated with specific functions, namely an N‐terminal domain with antimicrobial activity and a C‐terminal domain with wound‐healing properties. 167, 168 Histatin 5 has two important metal‐binding motifs, consisting of an amino‐terminal copper (II)/nickel (II)‐binding motif and a zinc (II)‐binding motif, although they can also bind to cobalt (II) ions.169, 170 The ability to coordinate metal ions appears to be important for the stabilization of the secondary structure, particularly in the presence of negatively charged membranes.170 Nevertheless, histatins lack any defined secondary structure and are unordered when in aqueous solutions, assuming α‐helical structures only in organic solutions.171, 172 In addition, histatin 5 binding to copper (II) and nickel (II) ions induces the formation of ROS, which contributes for its fungicidal activity.80, 169

Histatins can bind to a receptor on fungal cell membranes and induce cell death by disrupting the cell cycle and causing nonlytic loss of ATP, as discussed above.173 While histatins are effective, potent, and fast‐acting AMPs against several antimicrobial species, these peptides are nontoxic to human cells and human microflora, which may be due to the fact that their translocation across the membrane is site and species specific.78, 174

Both parotid, submandibular, and sublingual glands secret histatins, and their concentration and secretion have been shown to decrease with age, even when accounted for total protein concentration.175 Moreover, oral candidal infections increase with age, suggesting a link between age‐associated deficiency of human histatins and an increased risk of oral infections.175

D. Dermcidins

Dermcidins are broad‐spectrum anionic AMPs with no apparent homology to other known AMPs. Peptides in this group comprise 110 amino acid residues, which are subsequently processed into smaller but still active ones. Dermcidins are constitutively expressed in human eccrine sweat glands and secreted in sweat, exhibiting antimicrobial activity against common human pathogenic microorganisms such as S. aureus, Staphylococcus epidermidis, E. coli, E. faecalis, and C. albicans.176, 177, 178 This constitutive production contrasts with human defensins and cathelicidins, which are induced during inflammatory and stressful conditions.

A small percentage of human breast cancer cells displays RNA encoding dermcidin, and after oxidative stress induction different types of tumor cells leads to the production of diverse and biologically active proteolytically processed dermcidin peptides.179, 180 Therefore, the upregulation of dermicidins may confer human breast cancer cells a selective advantage compared to nonmalignant cells. In contrast, reduced expression of dermcidins in sweat from atopic dermatitis patients contributes to skin infections and altered skin colonization.181 Hence, the downregulation of dermicidins in atopic dermatitis patients may represents a suppression of an innate and fundamental defense mechanism, which, in turn, could confer an advantage to infectious microorganisms.

Unlike histatins, which are nontoxic to human cells and human microflora (see above), dermicidins can have serious deleterious consequences to the human host. Dermcidin isoform 2 is known as a stress‐induced oxidative protein capable of inhibiting the synthesis of nitric oxide in endothelial cells, insulin in pancreatic islet β‐cells, and hepatic hormone synthesis.182 Also, hypertensive patients show significantly higher levels of plasma dermcidin, but salicylic acid has been shown to reduce dermcidin levels while simultaneously attenuating dermcidin‐mediated insulin inhibition in patients with acute myocardial infarction.183 Therefore, the exploitation of dermcidin isoform 2 for antimicrobial therapeutic purposes is most likely not possible due to severe side effects. Despite this, it may be a promising therapeutic target for hypertension and diabetes management.

In animal models, dermcidin has been shown to induce platelet aggregation and coronary artery disease by activating platelet cyclooxygenase and inhibiting the constitutive form of nitric oxide synthase, respectively, with an efficiency 40 times higher than that of ADP at activating cyclooxygenase.182 Notably, dermcidin was able to stimulate the development of coronary artery disease in one animal model, even at suboptimal concentrations of ADP within 30 min.183

Together, these observations suggest that the ability of AMPs to fight infections depends on their induction by infectious microorganisms, the particular peptides generated, and the local environment/tissue in which they act. Also, biological roles performed by human AMPs are more diverse than immediately expected, playing essential functions other than fighting infectious microorganism(s). In fact, dermcidin‐mediated inhibition of insulin is unique in the sense that no other protein is known to arrest pancreatic insulin synthesis in such way, and similar observations might result in novel and unexpected applications of human AMPs.

E. Hepcidins

AMPs expressed by human liver (LEAP 1 and 2) were initially discovered in human blood, and subsequently found in urine. These peptides are known as hepcidins and are characterized by a high content in cysteine residues (approximately 32% of all residues, with eight disulfide bounds).184

LEAP‐1 or hepcidin‐25 is essential for maintaining iron homeostasis, and, when mutated, leads to severe iron overload and juvenile hemochromatosis.185 Hepcidins are the main regulators of plasma iron concentration and its secretion is promoted during inflammatory responses, particularly by the action of interleukin 6, which is one of the main mediators of the acute phase response.186 As iron bioavailability is a limiting factor for bacterial growth, hepcidins represent an organic mechanism for sequestering iron from invasive pathogens. However, when several iron sources are available, as it happens within the human host, eliminating a single reservoir may not be sufficient to attenuate microorganisms’ virulence. Moreover, bacteria are known to exploit almost every host iron‐binding protein and reservoir and, consequently, human LAEPs/hepcidins are a rather limited source of antimicrobial activity.187 For these reasons, the antimicrobial role of iron sequestration depends not only on hepcidins, but also on several other mediators, such as transferrin, lactoferrin, ceruloplasmin, haptoglobin, and hemopexin.

Mutations responsible for decreasing the levels of hepcidins or compromising the function of other regulators of iron homeostasis, such as the major histocompatibility complex class I like hereditary hemochromatosis protein (HFE), transferrin receptor 2, and ferroportin, lead to increased iron blood levels. When levels are high enough to produce hemochromatosis, individuals are at an increased risk of infectious complications, including bacteremia and meningitis caused by E. coli,188 bacteremic cellulitis, and hemorrhagic bullous skin lesions caused by V. cholera,189 multiple pyogenic abscesses in the liver parenchyma as a result of Yersinia enterocolitica infection,190 meningitis, endocarditis, and pericarditis resulting from infection with L. monocytogenes,191 and fatal septicemia during infection with Vibrio vulnificus.192

In contrast, hemochromatosis can also be associated with decreased macrophage iron storages in the case of HFE‐associated hemochromatosis. This depletion makes these cells more efficient in fighting Salmonella enterica subsp. enterica, serovar Typhimurium, and Mycobacterium tuberculosis infections.193, 194 However, these protective effects of macrophages iron storage depletion have only been demonstrated in animal models and appear to be mediated not by hepcidins, but rather by lipocalin‐2, which reduces the availability of iron for Salmonella, and transferrin and lactoferrin, both of which compromise iron acquisition by M. tuberculosis.193, 194

F. Neuroendocrine AMPs

Neuropeptides and neuroendocrine peptides may exert mitogenic actions through which innate barriers are reinforced and may display neurotransmitter, immunoregulatory, and direct antimicrobial activity.

Vasostatin‐1 was the first natural antifungal N‐terminal chromogranin A derived fragment peptide, having been first isolated from chromaffin secretory granules from bovine adrenal medulla.195 However, its sequence is highly conserved in humans (97% identity homology) and its microbial activity also overlaps, to some extent, with that of bovine origin.196 Irrespectively, human vasostatin‐1 displays antimicrobial activities against Gram‐positive bacteria (Micrococcus luteus, B. megaterium) and a large variety of filamentous fungi (N. crassa, Aspergillus fumigatus, Alternaria brassicola, Nectria haematococca, Fusarium culmorum, Fusarium oxysporum, Trichophyton mentagrophytes) and yeast cells (S. cerevisiae, C. albicans).195 Moreover, many autocrine and paracrine biological activities of chromogranin A, its precursor protein, have been attributed to peptides located along its sequence, which are co‐released with catecholamines by chromaffin cells upon stimulation and can be found in human bodily fluids.197, 198 Therefore, the antimicrobial activity spectrum of vasostatin‐1 may be extended by that of co‐released peptides also formed by chromogranin A proteolytic cleavage, and some of the coproduced peptides are even more potent than vasostatin‐1 itself.195

In addition to the structural requirements for the antimicrobial activity of chromogranin A derived peptides, posttranslational modifications may also modulate their activity. For instance, S‐pyridylethylation (an alkylating modification consisting of the addition of pyridylethyl moieties to the –SH functional groups of cysteine residues that disrupts disulfide bridges) renders it selectively active against B. megaterium but not M. luteus, while oxidation seems to render it inactive against bacteria but still active against fungi. In addition, chromogranin A derived peptides also undergo phosphorylation and glycosylation, which can modulate their physicochemical properties. 195

Currently, vasostatin‐1 is known to be stored in endocrine, neuroendocrine, and neuronal cells, and to be released from stimulated chromaffin and immune cells upon stress. Moreover, the stimulation of polymorphonuclear immune cells induces the processing of chromogranin A and the secretion of vasostatin‐1 and other chromogranin A derived peptides,195 which may account for local inflammation. Lastly, recombinant and synthetic human vasostatin‐derived fragments have been shown to inhibit vascular contraction induced by endothelin‐1, parathyroid hormonal secretion, neuronal survival, expression of neurofilaments, and neuronal GABA uptake.199, 200

Adrenomedullin was first isolated from a human adrenal pheochromocytoma and its plasma concentration is raised under specific physiological conditions, namely, renal failure,201 hypertension,202 and sepsis.203 This peptide has antibacterial activity against several Gram‐positive and Gram‐negative bacteria, including Propionibacterium acnes, S. aureus, M. luteus, P. gingivalis, Actinomyces naeslundii, S. mutans, C. albicans, Eikenella corrodens, Actinobacillus actinomycetemcomitans, Streptococcus pneumoniae, Haemophilus influenzae, Streptococcus pyogenes, Bacteroides fragilis, E. coli, and Helicobacter pylori.204

Human neuropeptide Y displays antimicrobial activity against bacteria and fungi, but truncated fragments are tenfold more potent than the intact precursor neuropeptide Y, perhaps because the net charge of the fragments is more positive than that of the intact neuropeptide.205

Encephalins and their derived peptides may either enhance or inhibit the immune response and the corresponding circulating immune cells/mediators in both humans and animal models.206, 207 as well as their subsequent antimicrobial effects, may thus take place indirectly in immune system modulation. Simultaneously, antibacterial assays have revealed that peptide B/enkelytin (a C‐terminal fragment of proenkephalin A) specifically targets Gram‐positive bacteria, including M. luteus, B. megaterium, M. luteus, and S. aureus, while having no effect over Gram‐negative bacteria.208, 209 Classically, proenkephalin A has been ascribed to the secretory granules from adrenal medullary chromaffin cells and various brain regions, such as the striatum and the pituitary, according to some animal studies.210, 211 Despite this, it has also been demonstrated to be expressed in and secreted by polymorphonuclear neutrophils (accounting for its particular enrichment in wound fluids),208 T and B lymphocytes, macrophages, and mast cells.212, 213 However, it should be noted that the antimicrobial effects of encephalins and their derived peptides result mostly from animal studies and have not been adequately studied in human secretions, despite the high conservation of their sequences across species, which most likely contribute for the similar activity spectrum. Even so, phosphorylation and glycosylation are characteristic of these peptides,208, 214 and it is possible to envision that differences at the level of posttranslational modifications or as a result of gene divergence may result in distinct activity spectra. Interestingly, peptide B/enkelytin is metabolized in vivo to opioid peptides, thus revealing an intricate relationship between innate immunity and pain modulation, and its plasma concentration is increased after coronary artery bypass grafting in humans.215 Because it is expressed in both immune and neuronal cells, it is possible that the same stimuli may act on both cells populations, thereby accounting for immune system modulation at the periphery and alterations in central nervous system signaling. However, the exact roles and mechanisms of this interplay are not clear.

In addition, other AMPs found in epithelial secretions from the gut and the skin, for example, include peptides derived from alpha‐melanocyte‐stimulating hormone, which is active against S. aureus and C. albicans.216 Accordingly, evidences suggest that AMPs like these act by inhibiting mRNA and protein synthesis rather than inducing direct membrane disruption, as is the case for most AMPs.217

G. Nonantimicrobial Proteins Give Rise to Bioactive AMPs

Several cationic proteins display nonenzymatic antimicrobial activity, which is sustained by the resulting cleavage peptides after protein fragmentation. For example, lactoferrin is a glycoprotein enriched in milk and neutrophilic granules, which has the ability to sequester iron and thus prevent bacterial overgrowth.218 When digested by pepsin, lactoferrin yields lactoferricin H (human), an AMP able to neutralize bacterial toxins, regulate gene transcription, inhibit complement activation, viral infection, and tumor overgrowth, thus further extending the antimicrobial properties of its precursor.76

Human chemokines are secreted by macrophages and polymorphonuclear immune cells and, therefore, found in the blood circulation.219 Moreover, during allergic inflammatory responses to bacterial infections of the airways, the eosinophil‐recruiting chemokines eotaxin‐1/CCL11, eotaxin‐2/CCL24, and eotaxin‐3/CCL26 are generated by mast cell proteases, active against several airway pathogens, including S. pneumoniae, S. aureus, H. aeruginosa.220

Peptides resulting from the cleavage of pepsinogen A and C prosequences have also been shown to exert antimicrobial activity. Although these were initially identified from the stomach of bullfrogs, synthesized human pepsinogen C prosequences with similar structural characteristics (amphipathic and helical structures) to these cleavage peptides have also exhibited antimicrobial activity.221

Calcitermin is an AMP that results from the cleavage of a 15 amino acids long sequence from the C‐terminal region of protein calgranulin C, isolated from human airway secretions.222 By adopting an alpha‐helical conformation in bacterial membranes, calcitermin is able to target Gram‐negative bacteria (e.g., E. coli and P. aeruginosa) and fungi (e.g., C. albicans).222 Furthermore, the precursor protein and fragment peptide may act synergistically to protect the human host against foreign microorganisms present in the airways. On the one hand, the cleavage peptide calcitermin inserts into membranes of microorganisms. On the other hand, the precursor protein calgranulin C has a proinflammatory activity. By acting as a danger‐associated molecular pattern molecule and binding to the receptor for advanced glycation end products in innate immune cells, it activates the MAP‐kinase and NF‐κB signaling pathways, leading to the production of proinflammatory cytokines and histamine, upregulation of cell adhesion molecules, recruitment of leukocytes, and degranulation of granulocytes.223, 224

Dermcidin‐derived peptides such as DCDs, SSLs, and LEKs (the letters correspond to the first three amino acids of the corresponding precursor dermcidin protein) are also formed in eccrine sweat glands (but not in apocrine sweat glands) at these body sites with high probability for contact with pathogens (e.g., palms, face, arms) and exhibit antimicrobial activity against a large number of microorganisms, including S. aureus, E. faecalis, E. coli, S. epidermidis, Pseudomonas putida, methicillin‐resistant S. aureus, and rifampicin/isoniazid‐resistant M. tuberculosis.176, 177, 178 Such dermcidin‐derived AMPs are regulated by proteolytic processing at several levels (see Section 2.) and are differentially expressed during inflammatory conditions. 225 These peptides are resistant to proteolytic degradation in sweat up to at least 40 hr and, interestingly, most dermcidin‐derived peptides are anionic in nature. If such distinct properties may lead to different antimicrobial activity spectrum is unclear, though not surprising, as skin represents a harsh and particularly unique environment. As of specific interest, it should be noted that the consensual average skin surface pH has been decreasing in the last few years and is now generally accepted to be below 5. While showering and cosmetic products decrease skin's surface pH, plain tap water can increase the skin pH up to 6 hr.226 Noticeably, negative charges seem to be important for dermcidin‐derived peptides’ antimicrobial activity, in contrast with what would be expected for AMPs. In fact, most dermcidin‐derived peptides are naturally anionic, but low pH (high H⁺ concentration) leads to an increase in their net charge (toward positivity). Therefore, while an acidic skin actually keeps the resident bacterial flora attached to it, a more alkaline skin compromises such attachment, suggesting that the more negatively charged such skin peptides are (alkaline skin), the more efficient these become.226

H. Large Antimicrobial Human Peptides

Despite the average length of 32.76 residues and net charge of +3.21 for the close to 3000 AMPs already identified, human AMPs tend to be longer and more positively charged. AMPs with more than 100 amino acids have been identified across the six kingdoms of life. Curiously, most (approximately half) appear to be of human origin, presenting as amino acid sequences of 127 residues on average and suggesting large AMPs to be a human feature. Moreover, compared to all the remaining AMPs, large human AMPs tend to be biased toward leucine (8.2%) and lysine (10.5%) residues, displaying nonpolar aliphatic and positively charged side groups, respectively (Table I). Therefore, large human AMPs are enriched with amino acids that promote distinguishing features of AMPs: high overall positive charge, hydrophobicity domains, and amphiphilicity. Nevertheless, this bias toward lysine residues is neither unique to human AMPs nor more pronounced among these. For instance, the amphibian cathelicidin RC‐1 is composed by 32% lysine residues, and its homologous snake crotalicidin contains 38% lysine residues, showing high potency against S. pyogenes, Acinetobacter baumannii, E. faecalis, S. aureus, E. coli, Klebsiella pneumoniae, and P. aeruginosa.227

As evidenced in Table I, large human AMPs have been found in a variety of bodily fluids, including tears, saliva, urine, airways secretions, breast milk, blood, sweat, and epididymal fluids. In addition, other large AMPs are also expected to exist despite not constitutively found in biologic fluids, due to their inducible nature. For instance, Paneth cells in the small intestine release AMPs‐rich granules upon stimulation by cholinergic or bacterial stimuli,228 and granulocytes are known to contain AMPs in granules destined for extracellular secretion.229, 230 Therefore, as the synthesis and secretion of large AMPs may not be constitutive but amenable to induction by microbial macromolecules and inflammatory cytokine stimuli, several other large AMPs may yet to be catalogued. Hence, exploration of novel large antimicrobial human peptides is an ongoing work that may yield novel therapeutic agents.

7. PEPTIDOMICS OF NATIVE PEPTIDES IN HUMAN BODILY FLUIDS

Peptidomics is the complete study or global analysis of the peptides present in a biological sample, at a given time, under a particular condition. 231, 232 The identification of naturally occurring AMPs in human biological fluids is a laborious, daunting, and sometimes unsuccessful task. Such naturally occurring peptides can directly result from gene encoding, from passage to bodily fluids as a result of cellular damage and renewal, or from extracellular proteolytic processing of precursor proteins. The latter, considered to be the most prevalent, hampers the identification and interpretation of biologically active AMPs because precursor proteins may themselves display antimicrobial activity, but may also be cleaved into multiple functional and nonfunctional peptides.233, 234 Moreover, owing to the large dependency of AMPs on their sequence and 3D structure, these peptides should not be subject to mass spectrometry (MS) based approaches employing enzymatic or chemical digestion, neither to any in vitro step that may cause their fragmentation. Instead, top‐down proteomic strategies are preferred. Classically, different peptides fractions have been separated and tested for their biological activity, often without ever addressing the peptides’ biological/antimicrobial activity, their sequence, target, or hypothetical mechanism(s) of action. However, current strategies are more focused, precise, and informative, capable of addressing all these parameters, sometimes in a single study (see below).129, 235, 236

Human biological fluids consist of complex samples, containing lipids, salts, proteins, carbohydrates, and other molecules that make the study of endogenous peptides extremely difficult. Consequently, enrichment and separation steps are almost always required. Of these biomolecules, proteins and larger peptides as well as salts are the most complicated to separate from the target peptides. Luckily, salts can sometimes be removed in the same procedures applied for protein removal. First and foremost, proteases/peptidases have to be inactivated, which can be done by denaturing procedures (e.g., microwaving, boiling) or by adding protease/peptidase inhibitors.237, 238 However, both of these can modify the peptide structure and, thus, these steps can be avoided when samples are resistant enough to further protease/peptidase activity. This is the case for urine samples, for example, as, when collected, have long been subject to proteolytic processing in the bladder. Proteins are frequently removed by selective precipitation, most commonly by organic solvent (e.g., acetone) or acidic (e.g., trichloroacetic acid) precipitation, which also removes salts (left in the supernatant).236, 239 However, precipitation‐based proteins removal is not complete and may lead to the formation of peptide aggregates, which are consequently lost in the precipitate.240

Since peptides can present a great variety of sizes, sequences, structures, hydrophobicity properties, net charges, and other characteristics, their isolation is complex. Still, such complexity allows for isolation steps to take place at multiple dimensions, increasing peptides isolation efficiency and accuracy. For instance, peptides can first be isolated based on their size and charge,241, 242 bias toward specific amino acids (e.g., cysteines, tryptophans, methionines),243, 244, 245 or on the presence of post‐translational modifications (PTMs) (e.g., phosphopeptides, glycopeptides).246, 247 Still, because of the aforementioned limitations, the search for endogenous AMPs in human bodily fluids has mostly relied on the identification of candidate AMPs by chromatographic techniques (e.g., high‐performance liquid chromatography, HPLC) and sequencing (e.g., Edman sequencing), followed by their de novo chemical synthesis and in vitro antimicrobial activity testing. For these reasons, the identification and in vivo testing of novel AMPs is often time consuming and requires multiple studies/phases and resources. In order to mitigate this problem, candidate human AMPs have come to be first predicted in multiple ways, including mining of the entire human genome for particular motifs known to confer antimicrobial activity, and screening of human peptidomes (e.g., salivary, urinary) in order to find homologous peptides to other known AMPs. Accordingly, several libraries and databases are currently available for data comparison, and some search programs can even combine both previously obtained data and de novo information.248

Even so, the identification of AMPs benefits the most from combinations of chromatographic and immunological techniques, genomics and proteomics. While genomics and proteomics both allow the prediction of novel AMPs and have the potential for contributing with a larger number of putative AMPs, peptides with proven in vivo antimicrobial activity have to be more accurately identified by combinations of chromatographic and immunological techniques, chemical sequencing and antimicrobial assays. Though immunoassays can detect the intended target in a variety of complex biological samples, these techniques do not provide an integrative view of the peptidome, can suffer from cross‐reactivity or lack of specificity issues and require previous knowledge of the target structure.249, 250 Moreover, this issue becomes a particular problem when considering that shorter or longer peptides resulting from the same precursor protein may all contain the same epitopes and thus be recognized as the same peptide fragment despite exhibiting distinct biological activities.

Electrospray ionization (ESI)236 and matrix‐assisted laser desorption ionization (MALDI)251 have been the most frequently and successfully applied ionization techniques in the field of bodily fluids peptidomics. In line with these, peptides are most frequently previously separated by LC‐based251 and capillary electrophoresis (CE) based approaches, allowing different peptide fractions to be independently separated.252 From this point onwards, almost any mass analyzer can be used, although in practice quadrupole time‐of‐flight (Q‐TOF) and ion traps (IT) instruments are the most frequently utilized ones in peptidomics, especially in exploratory studies.253, 254, 255 In turn, Fourier transform‐ion cyclotron resonance‐MS is more suitable for targeted/confirmatory analysis when dealing with fewer peptides.256 Because these apparatuses allow tandem MS to be performed, peptides can be readily sequenced in addition to their identification, circumventing the above‐mentioned limitations concerning immunological assays and largely simplifying peptidomics workflows.

In summary, when aiming at the isolation of a particular peptide, with predicted or known charge and molecular weight, an MS‐based procedure can theoretically be very straightforward. First, fluid samples require (sometimes multiple) centrifugation steps to remove cells, cellular debris, wastes, and possible contaminants. Then, proteins and peptides in the supernatant are separated and fractionated by multidimensional chromatographic techniques. Within these procedures, proteins and peptides can be initially fractionated by gel filtration (size dimension), followed by cation‐exchange (charge dimension) fractionation. Cation‐exchange chromatography uses a negatively charged ion exchange resin with affinity for molecules with net positive charge. A salt gradient is used to separate the peptides of interest from other bound peptides, based on their predicted isoelectric point. Finally, peptides can be subject to MS analysis. They are typically ionized by ESI or MALDI and subsequently analyzed in a Q‐TOF, IT, or hybrid LTQ. As AMPs tend to be highly basic peptides due to a bias toward arginine and lysine residues, these have a propensity to be particularly enriched when applying cation exchange chromatography based techniques.257 However, because biological fluids are complex matrices, these sometimes require multiple fractionation, isolation, and extraction steps. These are laborious optimization techniques for the detection and quantification of AMPs and, despite their inherent complexity, they are often based in not always accurate functional and structural predictions.

A recent approach has proved itself capable for the isolation of peptides present in bodily fluids that can bind directly to certain bacterial membrane components, such as lipopolysaccharide, and induce an inflammatory response against these.129, 235 After obtaining a bacterial homogenate and extracting membrane components with an organic solvent such as n‐butanol, functionalized beads with N‐hydroxysuccinimidyl‐sepharose (an agarose used in affinity and protein chromatography) are conjugated with these membrane bacterial components and incubated with a human biological fluid (e.g., saliva, urine). Then, peptides bound to the conjugated beads are eluted and analyzed in a LTQ‐Orbitrap hybrid Fourier transform apparatus. Moreover, functionalized beads conjugated with bacterial membrane components can be tested in vivo for their capacity to induce the production of inflammatory mediators. Thus, this exploratory approach could result in the identification of potential human AMPs with selective properties capable of binding bacterial membranes and inducing an inflammatory response.

Recently, Dallas and colleagues have attained the most complete and prolific analysis of naturally occurring peptides with potential antimicrobial activity in human milk by nano‐LC quadrupole‐TOF tandem mass spectrometry (MS/MS), while searching against libraries of human milk peptides and known AMPs.236 In order to remove milk peptides produced by in vitro proteolysis, fresh milk derived peptides were compared to control samples of the same origin but immediately subject to boiling. As boiling denatures proteases, in vitro proteolysis does not take place and peptides present in these controls could then be confidently assigned in the original analysis. Moreover, masses fragmented in the first round of MS/MS were excluded from subsequent rounds of MS.236 The authors realized that more than 80% of the +300 unique naturally occurring peptides identified with 99% confidence have yet unknown functions, which serves to illustrate how much there is still to discover regarding the biological roles of human endogenous peptides in biological fluids. Nevertheless, the antimicrobial activity of the milk peptides (41 predicted AMPs) was confirmed in vivo by radial diffusion assays against the Gram‐negative E. coli and the Gram‐positive S. aureus.236 Furthermore, this study reinforced the observation that the background against which mass spectra are searched seems to be a more critical factor than the mass analyzer or the ionization technique used, when it comes to both the number of peptides identified and the confidence by which these can be assigned.

In the same study, authors have emphasized and demonstrated the importance of avoiding in vitro hydrolysis of AMPs in human bodily fluids, which can be attained by adding protease inhibitors after sample collection or by adequately accounting for this phenomenon, as the authors did by using an appropriate control. In contrast, not all bodily fluids seem to be susceptible to this degree of hydrolysis, at least not to the same extent. Urine, for instance, is thought to be more stable and not to require treatment with protease inhibitors.258 However, by the time urine is collected it has already remained stagnant for considerably longer periods of time, and, hence, probably already extensively exposed to the action of proteases/peptidases inherently present in urine. Therefore, how intact urinary AMPs actually are should always be questionable and these should ideally be interpreted considering the background of hydrolytic enzymes present in the same urine sample.

Also questionable is the inhibition of proteolysis in seminal plasma samples. In fact, the liquefaction process of human semen performed by seminin and other proteolytic enzymes is physiological,259 and, therefore, naturally occurring AMPs should be analyzed in samples that were allowed to liquefy without the addition of protease inhibitors. Curiously, seminin, which is the primary agent responsible for liquefaction of the seminal coagulum, is partially destroyed by freezing at −20°C,259 reinforcing the vitality of analyzing fresh samples. Similarly, the degradation of peptide fragments derived from semenogelins in seminal plasma samples is time dependent and responsible for the decline in HIV‐enhancing activity of sperm, once again underscoring the importance of timely performing analyses of peptides in bodily fluids.260

As previously stated, AMPs activity depends largely on their 3D structures, and their correct characterization becomes of paramount importance. The best experimental techniques have been and currently are X‐ray crystallography 261 and nuclear magnetic resonance,262 both of which allow for the study of structure–function relationships, though very expensive. Therefore, currently, these are commonly replaced or complemented by bioinformatics prediction tools whenever possible. Using such tools, the sequence of amino acids of a peptide allows for the accurate and immediate prediction of the presence of secondary structures (e.g., α‐helices, β‐sheets). Then, the overall 3D structure is devised by algorithms employing energy minimization principles and molecular dynamics, sometimes resulting in multiple and not mutually exclusive 3D conformations, which is in accordance with peptides dynamic structure in biological fluids.263, 264 The most common identification software referenced in the literature for peptide/precursor protein identification are SEQUEST and MASCOT.265, 266 Currently, however, there is a considerable need for algorithms and software capable of correctly identifying AMPs and predicting their structure–function relationship.

8. APPLICATION OF AMPs BEYOND CONVENTIONAL PHARMACOLOGICAL THERAPY

AMPs are a promising replacement for conventional antibiotics owing to their effectiveness against multiresistant bacteria and over multiple bacterial species simultaneously, invoking a diverse set of mechanisms that cannot be easily shortcut by bacteria, fungi, and/or virus. An ideal AMP would be (i) highly selective against its target while leaving human host cells unaffected, (ii) not prone to resistance mechanisms, (iii) relatively easy to produce at low costs, and (iv) stable during storage or upon administration. Drug design of such AMPs focuses on peptidic and nonpeptidic mimetic drugs, modulating the hydrophobicity, positivity, and amphiphilicity of endogenous AMPs to increase their antimicrobial potency. However, exploiting AMPs for therapeutic purposes is actually more complex than initially envisioned. While linear AMPs present a simpler design and a more predictable structure–function relationship, circular peptides may allow the design of longer acting and more resistant AMPs. Such peptides have been designed by inserting the intended AMP into other circular peptide (cyclotides) and have been proven efficacious against HIV and inflammatory diseases by circumventing their instability and poor bioavailability.22, 267 Similarly, while humans produce AMPs with L‐amino acids and their antimicrobial activity may depend on the action of endogenous enzymes, which can only act upon L‐type amino acids, producing AMPs with D‐amino acids may render these more resistant to hydrolysis.

Rather than administering exogenous AMPs, it is possible and perhaps more secure and controllable to pharmacologically stimulate the production of endogenous AMPs. For instance, 1,25‐dihydroxycalcypherol (active vitamin D3) has been successfully used to induce the expression of DEFB4 (defensin, beta 4) and CAMP (cationic antimicrobial peptide) genes, the production of both LL‐37 and human β‐defensin 2 in cell cultures of keratinocytes from diabetic foot ulcers and to promote wound healing.268 Accordingly, the persistence of functional milk human κ‐caseinoglycopeptides in plasma of human infants after human milk feeding serves the purpose of exogenously administering AMPs without the requirement for their de novo synthesis.269

AMPs can help preventing biofilm formation and microorganisms’ growth. That is the case of LL‐37 and its fragments, which were already shown to have antibiofilm formation properties against several methicillin‐resistant S. aureus strains and P. aeruginosa, common pathogens of hospital‐acquired infections, as well as against the melioidoisis’ etiological agent Burkholderia pseudomallei and the uropathogenic E. coli.270, 271, 272, 273 Another AMP with antibiofilm formation activity is hBD‐3, found to prevent the development of methicillin‐resistant S. epidermidis and S. aureus biofilms, largely responsible for orthopedic implants associated infections.274 Likewise, AMPs may also inhibit the formation of fungal biofilms. For instance, histatin 5 has been demonstrated to successfully inhibit the growth of C. albicans biofilm on denture acrylic.275 These examples may broaden the applications of AMPs in the clinical and biomedical fields by allowing functionalization of catheters. Alternatively, implants and other medical devices can also reduce bacterial colonization, biofilm formation, and infection rates, presenting long lasting functionality, broad‐spectrum activity, and minimal cytotoxicity against human cells.276, 277

Human AMPs could, theoretically, also be used as biosensors immobilized onto microchips for identification purposes, allowing the identification of microorganisms. However, such applications have only been exploited with animal or synthetic AMPs. These biosensors on electrical impedance alterations resulting from the presence of bacterial or fungal surface elements, or on the specificity of peptide nucleic acid probes against their targets, presenting higher versatility, lower costs, and lower sensitivity limits compared to other conventional methodologies.278, 279, 280 Alternatively, AMPs could be used as drug delivery systems in the form of conjugates in order to accurately target drugs and other agents to tumor sites or intended organs, as typified by monodisperse “endosomolytic” nanoparticles for in vivo gene delivery using intravenous injection, or by functionalized nanoparticles with higher membrane penetration targeted against gliomas.281, 282 Such functionalization may allow for deeper tissue penetration, increased antimicrobial potency, sustained release, and novel routes of administration to be achieved.

9. DISCLOSERS AND CHALLENGES TO OVERCOME

Ever since the isolation of the first peptide from the frog skin in 1983,283 several breakthroughs were made in the isolation, synthesis, and application of AMPs. Still, many challenges are yet to overcome in the field of AMPs peptidomics,284 reflecting mainly their huge diversity and how little is known about their behavior in vivo and their structure–function relationships.

In terms of costs, producing AMPs can be several hundred times more expensive than the production of conventional antibiotics.285 Moreover, AMPs’ design can be very complex and the quest for the perfect AMP a never‐ending search. In fact, considering the possibility of working with only 20 amino acids, a small peptide with ten amino acids could have up to 20¹⁰ = 1.02 × 10¹³ different sequences, a larger peptide with 20 amino acids could present with up to 20²⁰ = 1.05 × 10²⁶ different sequences, and one with 100 amino acids could adopt any one of 20¹⁰⁰ = 1.27 × 10¹³⁰ different sequences. Such possibilities do not even account for the presence of posttranslational modifications. Then, each one of these peptides could shape into several different secondary structures and fall into a wide spectrum of activity, depending on the milieu conditions. Furthermore, in a way to overcome the challenge of peptide stability, one has to consider the possibility to synthesize and to administer AMPs containing D‐amino acids, thus decreasing their susceptibility to hydrolysis by human enzymes.267 To deal with such large amount of amino acid combinations one should mind using peptide libraries and computational models (such as by quantitative structure–activity relationship analysis) in order to narrow large collections to few surrogate AMPs (for a review please see Blondelle and Lohner's paper.286

Another challenge to overcome is the lower AMP's activity in vivo when compared to that observed in vitro. This has largely hampered the development of AMPs as therapeutic agents, as this decrease in activity stems from differences in pH and salt concentrations, the presence of proteases and corresponding inhibitors, and other interacting molecules hindering antimicrobial activity. Accordingly, several AMPs (e.g., Pexiganan,287 Iseganan,288 Neuprex 289 have reached phase II clinical trials only to fail after approval for marketing because they did not evidence superior activity over already marketed conventional antibiotics.

In early 2016, clinicaltrials.gov (a registry and result database of publicly and privately supported clinical studies of human participants conducted around the world) listed few clinical trials involving AMPs. For instance, LL37 is currently being studied for its efficacy in melanoma patients due to its ability to stimulate the immune system, while C16G2 and chromogranin A derived peptides are under scrutiny for their therapeutic potential against dental diseases. Simultaneously, therapeutic strategies are also being studied to augment AMPs’ expression in AMP‐deficient patients. This deficiency is believed to be due to an increase in Th2 lymphocytes derived cytokines, IL‐4, IL‐13, and IL‐10, and a decrease in TNF‐α, IL‐6, IL‐1, and interferon‐γ. One example of such strategy is Pimecrolimus, a calcineurin inhibitor that binds with high affinity to macrophilin‐12, preventing the translocation of nuclear factor of activated T cells to the nucleus and the transcription and release of such inhibitory cytokines (for more up‐to‐date trials, readers can access clinicaltrials.gov).

Another possible caveat concerns the cytotoxicity to mammalian cells when in concentrations above naturally occurring values. When bacteria produce AMPs, they also develop means to avoid their targeting by such peptides, such as efflux pumps and the sequestering of enzymes. However, eukaryotic cells are equally prone to damage if the administration of exogenous AMPs is carried out in high enough concentrations. Therefore, administering AMPs not constitutively produced or at concentrations above those normally found in the human host environment can be toxic and very detrimental to human cells. Furthermore, rapid metabolism and low bioavailability are characteristic of AMPs because these peptides can suffer extensive proteolysis in vivo, thereby accounting for their inactivation and short half‐life.22, 28, 290 Similarly, for peptides to be available at sites distant from their origin, these would have to cross many cellular membranes. However, due to AMPs’ polarity, membrane hydrophobicity may prevent them from doing so.291 Moreover, in contrast to other highly polar mediators (e.g., ionic salts and a few metabolites), endogenous peptides tend not to have a transporter, further compromising their bioavailability in different compartments and the subsequent impossibility to be absorbed through the gastrointestinal tract.

Therefore, the number of possibilities rapidly escalades, and each AMP‐based therapy can be further optimized based on the modulation of chemical properties and engineering of delivery systems. Even acknowledging all these factors, the human host is most certainly an unpredictable environment and the same peptide may behave differently, displaying unique individual pharmacokinetics and pharmacodynamics. On a final note, due to AMPs’ interaction with and dependence on other immune mediators, variability will most likely result from immune status alterations as those frequently found in human diseases, including infectious ones.

Taken together, the above discussed results support AMPs as constituting a paramount group of defense molecules in human biological fluids, being present in most fluids, contributing for the sterility of some of these, but being most notably enriched in biological fluids bathing tissues with the largest load of microorganisms (Fig. 2), such as saliva and urine. Also, while some AMPs seem the display activity spectra against only a handful of microorganisms, others display broad‐spectrum effects. Thus, the expression and activity patterns of human AMPs enriched in biofluids may have suffered selective pressures resulting from exposers to microorganisms specifically present in each fluid (Fig. 2).

Supporting information

Supplementary Table S1: Distribution of Known Antimicrobial Peptides Listed in The Antimicrobial Peptides Database by Activity Spectrum (2016)

Supplementary Table S2: Distribution of Known Antimicrobial Peptides Listed in The Antimicrobial Peptides Database by Structural Information (2016)

Biographies

Paulo Bastos is a biomedicine graduate student currently enrolled in a Master's program at the Biomedical Research Foundation, Academy of Athens, working on mass spectrometry assays development. He is also still performing research at iBiMED, University of Aveiro, on human diseases and biomarker research. His research background includes the study of host‐microorganism interactions at the Institut Pasteur in Paris. PB is soon joining the Institute of Science and Technology, Klosterneuburg, Austria.

Fábio Trindade has graduated in Biochemistry (2012) and finished his master degree in Clinical Biochemistry in 2014 at the Department of Chemistry of the University of Aveiro. He has done research at the Mass Spectrometry Center in Aveiro mainly focusing in salivary peptidomics and salivary antimicrobial peptides both in health and in chronic periodontitis. Now he is a PhD student in Biomedicine and he is developing research in urine and pericardial fluid proteomics and myocardial reverse remodeling in both iBiMED (Department of Medical Sciences, University of Aveiro) and UnIC (Faculty of Medicine, University of Porto). His main interests are to understand the molecular basis of human diseases, biomarker research and identification of therapeutic targets.

João Pinto da Costa graduated in Biotechnological Engineering at the University of the Algarve (Portugal) in 2005. Soon thereafter, he successfully concluded his Master+s Degree in Medical Diagnostics, awarded by the Cranfield University (UK). During 2006, he was a visiting scientist at Exelixis, Inc. (USA), a development‐stage biotechnology company focusing on the development of novel small molecule therapeutics for the treatment of cancer. From 2007 to 2009, he worked as a research fellow at Cranfield and Aveiro Universities in different fields of research, including mass spectrometry, and applied mycology. In April 2013, he concluded his Ph.D. thesis in Environmental Chemistry. Since then, João has worked in multiple research projects as either a Post‐Doctoral or research fellow at the University of Porto and the University of Aveiro. As of May 2015, João is a member of the Analytical Chemistry Group at the Chemistry Department of the University of Aveiro, where he is currently developing efforts on the monitoring environmental contaminants in real time by autonomous biosensor platform (NFETs). Furthermore, he frequently cooperates with iBiMED research members on proteomics studies focusing on a wide array of topics, ranging from post‐translational protein modifications to fundamental aspects of human ageing.

Rita Ferreira has a degree and PhD in Biochemistry. She is an assistant Professor in the Chemistry Department of the University of Aveiro, where she teaches curricular units from the Bachelor's and Master's degrees in Biochemistry. Rita Ferreira has developed extensive research work in the field of muscle plasticity in pathophysiological conditions like cancer and works in the application of proteomics for the identification of the biological processes underlying muscle plasticity in pathophysiological conditions.

Rui Vitorino has a degree and PhD in Biochemistry. He is Assistant Professor in the Medical Sciences Department of the University of Aveiro. Rui Vitorino started his research work in 2000 with saliva proteomics research. Since then, he extended the application of proteomics to the characterization of other biofluids' proteome and peptidome in order to identify the biological pathways modulated by diseases such as diabetes mellitus and bladder cancer. Currently, he works to a large extent with bioinformatics tools for the identification of enzymes involved in the regulation of proteome remodeling.