Antimicrobial Peptides vs Antibiotics: The Evidence

Antimicrobial Peptides

5,099 AMPs Cataloged

The APD3 database now catalogs 5,099 antimicrobial peptides from bacteria, plants, fungi, and animals. These molecules kill pathogens through mechanisms fundamentally different from conventional antibiotics.

APD3 Database, University of Nebraska Medical Center, 2025

APD3 Database, University of Nebraska Medical Center, 2025

View as image

Antimicrobial resistance killed an estimated 1.27 million people directly in 2019 and was associated with 4.95 million deaths globally, according to the Global Burden of Disease study published in The Lancet. Conventional antibiotics target specific bacterial proteins or enzymes, and bacteria evolve resistance to these single targets with alarming speed. Antimicrobial peptides (AMPs) attack bacterial membranes through physical disruption, a mechanism that has persisted in nature for hundreds of millions of years with limited resistance evolution.^[1] Whether AMPs can translate from ancient biological defense into modern therapeutics is the central question of this field. The evidence is substantial but complicated.

The Scale of the Antibiotic Resistance Crisis

The urgency behind AMP research is driven by numbers. The 2019 Global Burden of Antimicrobial Resistance study, published in The Lancet, attributed 1.27 million deaths directly to bacterial antimicrobial resistance and found it associated with 4.95 million deaths globally in a single year. Six pathogens (E. coli, S. aureus, K. pneumoniae, S. pneumoniae, A. baumannii, and P. aeruginosa) were responsible for the majority of resistance-associated deaths. The WHO's 2024 updated priority pathogen list classified several of these as critical-priority threats requiring urgent new therapeutic strategies.^[10]

The pipeline of new conventional antibiotics is thin. Most large pharmaceutical companies exited antibiotic development over the past two decades because short-course treatments generate less revenue than chronic disease drugs, and resistance can render a new antibiotic obsolete within years. This economic reality has created what infectious disease researchers call the "discovery void": a widening gap between the pace of resistance evolution and the pace of new drug development. AMPs represent one of the most active attempts to fill this void, alongside bacteriophage therapy, anti-virulence strategies, and microbiome-based interventions.

What Are Antimicrobial Peptides?

Antimicrobial peptides are short chains of amino acids, typically 12 to 50 residues long, that organisms produce as part of their innate immune defense. They are found in virtually every kingdom of life: bacteria produce bacteriocins, plants produce defensins and NCR peptides, insects produce cecropins, amphibians produce magainins, and mammals produce cathelicidins and defensins.^[2] As of January 2025, the APD3 database catalogs 5,099 peptides, including 3,306 natural AMPs from six kingdoms: 410 bacterial bacteriocins, 5 from archaea, 8 from protists, 29 from fungi, 268 from plants, and 2,580 from animals, alongside 1,299 synthetic and 231 computationally predicted peptides.

Most AMPs share three properties that enable their antimicrobial function. They are cationic (positively charged), which attracts them to negatively charged bacterial membranes. They are amphipathic, meaning they have both water-loving and fat-loving regions, allowing them to insert into lipid bilayers. And they are small enough to reach the bacterial membrane before proteases can degrade them. The selectivity of AMPs for bacterial over mammalian membranes arises from a fundamental difference in membrane composition: bacterial membranes are rich in negatively charged phospholipids (phosphatidylglycerol, cardiolipin), while mammalian cell membranes have a neutral outer leaflet dominated by phosphatidylcholine and cholesterol.^[1]

The human body produces its own arsenal. Alpha-defensins are stored in neutrophil granules at concentrations high enough to kill bacteria on contact. The cathelicidin LL-37 is expressed in epithelial cells, macrophages, and neutrophils across the skin, lungs, and gastrointestinal tract. Beta-defensins line epithelial surfaces throughout the body. These peptides do not merely supplement the immune system; in many contexts, they are the immune system's first line of killing.

AMPs are classified structurally into four main groups: alpha-helical peptides (like LL-37 and magainins, which form helical structures upon membrane contact), beta-sheet peptides (like defensins, stabilized by disulfide bonds), extended peptides (rich in specific amino acids like proline, tryptophan, or histidine), and loop peptides (containing a single disulfide bond creating a loop structure).^[2] Each structural class interacts with bacterial membranes differently, contributing to the diversity of killing mechanisms.

For a broader look at how nature's organisms produce antimicrobial compounds, see Amphibian Skin Peptides: The Pharmacy on a Frog's Back.

How AMPs Kill Bacteria: Three Membrane Models

The mechanisms by which AMPs destroy bacteria have been studied for decades. Kim Brogden's 2005 review in Nature Reviews Microbiology established the framework that the field still uses, identifying three primary models of membrane disruption plus several intracellular mechanisms.^[1]

Barrel-Stave Model

AMP molecules insert perpendicularly into the lipid bilayer and oligomerize to form a barrel-like channel. The hydrophobic faces of the peptides align with the lipid core; the hydrophilic faces line the interior of the pore. The result is a transmembrane channel that allows uncontrolled ion flux and cytoplasmic leakage. Alamethicin is the best-characterized peptide operating through this mechanism.

Carpet Model

Rather than inserting into the membrane, peptides accumulate on the bacterial surface in a carpet-like fashion, lying parallel to the membrane. When a threshold concentration is reached, the accumulated peptides disrupt the membrane in a detergent-like manner, breaking it into micelles. This model does not require specific peptide-peptide interactions or pore formation, which is why bacteria find it difficult to develop resistance: there is no single molecular target to mutate.

Toroidal-Pore Model

Peptides insert into the membrane and cause the lipid monolayers to bend continuously, forming a pore lined by both peptide molecules and lipid headgroups. Unlike barrel-stave pores where only peptides line the channel, toroidal pores involve the membrane itself curving inward. Magainin-2, one of the first AMPs discovered in frog skin, and the human cathelicidin LL-37 operate through variations of this model.

Beyond Membrane Disruption

Brogden's review also noted that membrane pore formation may not be the only killing mechanism. Subsequent research confirmed that many AMPs, after crossing the membrane, target intracellular processes: inhibiting DNA replication, RNA transcription, protein synthesis, and enzymatic activity.^[1] A 2026 study demonstrated this dual functionality directly by designing FPON, the first antimicrobial peptide engineered to both disrupt bacterial membranes and inhibit protein translation simultaneously.^[3] FPON specifically targeted gram-negative bacteria, exhibited low toxicity to mammalian cells, and showed significant activity against drug-resistant bacteria both in vitro and in mouse infection models. For more on how AMPs interact with specific drug-resistant organisms, see AMPs Against MRSA: Peptide Approaches to Drug-Resistant Staph.

Why AMPs Resist Resistance (Mostly)

The central argument for AMPs as antibiotic alternatives is that bacteria struggle to evolve resistance against them. The reasoning is straightforward: conventional antibiotics typically target a single protein (a ribosomal subunit, a cell wall enzyme, a DNA gyrase), and a single point mutation in that protein can confer resistance. AMPs target the bacterial membrane itself, a fundamental physical structure that bacteria cannot easily redesign without losing viability.^[2]

Hancock and Sahl's 2006 analysis in Nature Biotechnology noted that AMPs have coexisted with bacteria for hundreds of millions of years, yet widespread high-level resistance remains rare.^[4] This contrasts sharply with conventional antibiotics, where clinically significant resistance often emerges within years of a drug's introduction.

But "rare" is not "absent." Cheung et al. demonstrated in 2018 that Staphylococcus aureus uses the Pmt ABC transporter to actively pump out human antimicrobial peptides.^[5] Their study provided the first direct in vivo evidence that AMP resistance contributes to actual infection: S. aureus strains with functional Pmt caused more severe skin infections in mice, and this virulence advantage disappeared when AMP activity was experimentally removed from the system. The implication is clear: AMP resistance mechanisms already exist in clinically important pathogens and already matter during human infections.

Other bacterial resistance strategies include modifying membrane lipid composition to reduce negative charge, producing proteases that degrade AMPs before they reach the membrane, and secreting molecules that sequester AMPs in the extracellular space. These mechanisms tend to provide partial, low-level resistance rather than the complete resistance often seen with conventional antibiotics, but they are real and should inform therapeutic AMP design. For a deeper analysis of how bacteria develop and deploy these defenses, see Can Bacteria Become Resistant to Antimicrobial Peptides?.

Two Therapeutic Classes: Killers and Modulators

Hancock and Sahl's framework divided AMPs into two distinct therapeutic categories with different clinical trajectories.^[4]

Direct Antimicrobial Peptides

These kill bacteria through the membrane disruption and intracellular mechanisms described above. They are rapid-acting (killing within minutes rather than hours), broad-spectrum, and potent. Clinical trials have tested several direct-acting AMPs including pexiganan (a magainin derivative), omiganan (a cathelicidin-derived synthetic), and various defensin analogs. To date, clinical success has been limited primarily to topical applications.^[4]

A 2022 study advanced the topical delivery approach by encapsulating omiganan in liposomes for treating atopic dermatitis and psoriasis.^[6] The liposomal formulation achieved 72% encapsulation efficiency with vesicles of 120 nm diameter, demonstrated controlled release, and showed better skin permeation than conventional gel formulations. In mouse models of atopic dermatitis, omiganan liposomal gel significantly reduced inflammatory markers.

Immunomodulatory Host-Defense Peptides

These peptides boost the host's own infection-fighting capacity rather than killing bacteria directly. They modulate innate immune responses: recruiting immune cells to infection sites, enhancing phagocytosis, promoting wound healing, and dampening excessive inflammatory responses that cause tissue damage during sepsis. The theta-defensin RTD-1 illustrates this dual capacity: a 2025 study showed it simultaneously activates antiviral interferon pathways through JAK-STAT1 signaling while inhibiting pro-inflammatory NF-kB signaling.^[7] RTD-1 also inhibited SARS-CoV-2 infection in human cell models.

This immunomodulatory approach may ultimately prove more clinically viable than direct killing because it works with the immune system rather than trying to replicate what antibiotics do.

AMPs Against Biofilms

Biofilms represent one of the most intractable problems in infectious disease. Bacteria embedded in biofilm communities can tolerate antibiotic concentrations 100 to 1,000 times higher than planktonic (free-floating) bacteria. AMPs show particular promise in this area because their membrane-targeting mechanism can penetrate biofilm matrices that block conventional antibiotics.

A 2026 study characterized NCR169C17-38, a plant-derived AMP fragment, and found it effectively inhibited biofilm formation and eradicated pre-formed biofilms of Acinetobacter baumannii at sub-MBC concentration (3.2 micromolar), a result not observed for the conventional antimicrobial peptide polymyxin B.^[8] Transcriptomic analysis revealed a global shutdown of bacterial metabolic functions, with 450 of 503 differentially expressed genes being downregulated. This multi-target assault on bacterial function makes resistance evolution substantially harder than against drugs with single targets.

A complementary strategy emerged from biofilm disruption research. Kurbatfinski et al. (2025) showed that bacteria newly released from biofilms by a targeted monoclonal antibody become transiently hypersensitive to killing by respiratory tract AMPs including beta-defensin 1, beta-defensin 3, and LL-37.^[9] In three animal models of biofilm infections, antibody-mediated disruption alone (without co-delivered antibiotics) achieved rapid bacterial clearance and disease resolution. This two-step approach, breaking the biofilm shield and letting innate AMPs clean up, could transform management of chronic infections like recurrent sinusitis, otitis media, and device-related infections. For the full landscape of AMP-biofilm interactions, see AMPs and Biofilm Infections: Breaking Through Bacterial Fortresses.

Synergy: AMPs Plus Conventional Antibiotics

Using AMPs alongside existing antibiotics rather than as replacements may be the most practical near-term clinical application. A 2026 comprehensive review in Pharmacological Reviews cataloged dozens of successful AMP-antibiotic synergistic combinations against WHO priority pathogens.^[10]

The mechanistic basis for synergy is logical: AMPs disrupt bacterial membranes, creating entry points for antibiotics that would otherwise be excluded. Several specific mechanisms drive synergistic effects:

• Membrane permeabilization: AMPs create pores or weaken membrane integrity, allowing antibiotics to reach intracellular targets they could not otherwise access.
• Efflux pump inhibition: Some AMPs interfere with bacterial efflux pumps that normally expel antibiotics, restoring antibiotic sensitivity.
• Biofilm penetration: AMPs penetrate biofilm matrices, delivering antibiotics into communities that would otherwise be protected.
• Intracellular delivery: AMPs that translocate across membranes can carry co-administered antibiotics into the bacterial cytoplasm.

The practical consequence is that sub-inhibitory concentrations of AMPs, doses too low to kill bacteria on their own, can reduce the effective antibiotic dose needed by several fold. This approach could extend the clinical lifespan of existing antibiotics while reducing dose-related toxicity. For detailed analysis of specific AMP-antibiotic pairings and their clinical potential, see Combination Therapy: AMPs Plus Antibiotics for Synergistic Effects.

LL-37 Derivatives: A Case Study in AMP Optimization

Human cathelicidin LL-37 is the most studied human AMP and illustrates both the potential and the challenges of therapeutic AMP development. It is 37 amino acids long, broad-spectrum against gram-positive and gram-negative bacteria, and has immunomodulatory properties. It is also expensive to synthesize, susceptible to proteolytic degradation, and toxic to mammalian cells at higher concentrations.

A 2025 study designed a rationally optimized LL-37 derivative that addressed several of these limitations.^[11] The derivative showed lower minimum inhibitory concentrations (MICs) and faster bactericidal kinetics than the parent LL-37 against clinical multidrug-resistant E. coli and ESKAPE pathogens (Enterococcus faecium, S. aureus, Klebsiella pneumoniae, A. baumannii, Pseudomonas aeruginosa, Enterobacter spp.). It maintained low cytotoxicity to human epithelial and immune cells at concentrations exceeding bactericidal levels. ELISA quantification showed immunomodulatory effects on IL-6 and TNF-alpha at bactericidal concentrations, confirming that optimized derivatives can retain the dual antimicrobial-immunomodulatory properties of the parent peptide.

Machine Learning Is Accelerating AMP Discovery

Traditional AMP discovery relied on isolating peptides from natural sources, a slow and labor-intensive process. Machine learning has fundamentally changed this. A 2024 review in Nature Reviews Bioengineering documented how ML models now predict antimicrobial activity, hemolytic toxicity, and proteolytic stability from peptide sequences alone.^[12]

The approach works because AMP activity depends on physicochemical properties (charge, hydrophobicity, amphipathicity, secondary structure) that are computationally predictable from amino acid sequences. ML models trained on databases of known AMPs can screen millions of candidate sequences in hours, identifying novel peptides with predicted activity profiles that would have taken years to discover experimentally.

Several recent advances illustrate the pace:

• De novo peptide generation models create entirely new sequences not found in nature, optimized for specific bacterial targets
• Multi-objective optimization simultaneously maximizes antimicrobial potency while minimizing human cell toxicity and maximizing stability
• Transfer learning models trained on broad antimicrobial data are fine-tuned for specific pathogens, enabling targeted drug design

The limitation is that computational prediction still requires experimental validation. Many computationally promising AMPs fail in the laboratory, and many that succeed in vitro fail in animal models. ML accelerates the discovery pipeline but does not eliminate the need for rigorous biological testing.

The Clinical Reality: Why Aren't AMPs in Every Hospital?

Despite decades of research and thousands of characterized peptides, only a handful of AMPs have reached clinical use. Nisin (a bacteriocin used as a food preservative), gramicidin (a topical antibiotic), polymyxins (last-resort intravenous antibiotics for multidrug-resistant gram-negative infections), and daptomycin (a lipopeptide antibiotic) are among the few.^[4]

Several factors explain this gap between laboratory promise and clinical reality:

Proteolytic degradation. AMPs are peptides, and the human body is full of proteases that degrade peptides. Serum half-lives are often measured in minutes. Strategies to address this include D-amino acid substitution (which makes peptides invisible to most human proteases), cyclization, peptidomimetics, and nanoparticle or liposomal delivery systems.^[6]

Mammalian cell toxicity. The same properties that allow AMPs to disrupt bacterial membranes, positive charge and amphipathicity, can also damage human cell membranes at higher concentrations. The therapeutic window between bactericidal and cytotoxic concentrations is often narrow. Rational design approaches, like the dual-function FPON peptide that selectively targets gram-negative membranes, are widening this window.^[3]

Manufacturing cost. Solid-phase peptide synthesis is expensive at pharmaceutical scale. Recombinant production in bacteria or yeast can reduce costs but introduces challenges with peptide folding, purification, and endotoxin contamination.

Route of administration. Most clinical success has been topical. Systemic delivery faces the combined challenges of proteolytic degradation, renal clearance, and potential immunogenicity. This limits AMPs primarily to skin infections, wound care, and mucosal surfaces, though emerging delivery technologies are expanding the possibilities.

Regulatory pathway. AMPs do not fit neatly into existing regulatory categories. They are biologics (peptides) that act like small-molecule antibiotics. The FDA approval pathway can be uncertain and expensive.

These are genuine obstacles, not reasons for dismissal. Each has active research programs developing solutions, and the urgency of antimicrobial resistance continues to increase the incentive for regulatory and commercial investment.

The Bottom Line

Antimicrobial peptides represent a fundamentally different approach to fighting bacterial infections than conventional antibiotics. The evidence for their membrane-disrupting mechanisms, broad-spectrum activity, and capacity for synergy with existing drugs is strong and growing. The evidence for their clinical viability is more mixed: real obstacles in stability, toxicity, cost, and delivery have limited therapeutic translation despite decades of research. The most promising near-term applications are AMP-antibiotic combination therapies, topical formulations using advanced delivery systems, and immunomodulatory peptides that boost host defenses rather than replacing antibiotics. Machine learning is accelerating the pace of discovery and optimization. AMPs will not replace antibiotics wholesale, but they are increasingly likely to become part of the antimicrobial arsenal alongside them.

Frequently Asked Questions

What are antimicrobial peptides and how do they work?

Antimicrobial peptides are short proteins, typically 12 to 50 amino acids, produced by nearly all living organisms as part of innate immune defense. They kill bacteria primarily by disrupting cell membranes through physical mechanisms (barrel-stave pores, carpet-like disintegration, or toroidal pores) rather than targeting specific enzymes like conventional antibiotics. Many AMPs also inhibit intracellular processes including DNA replication and protein synthesis after crossing the membrane.^[1]

Can bacteria become resistant to antimicrobial peptides?

Bacteria can develop partial resistance to AMPs, but high-level resistance is rare compared to conventional antibiotics. Known resistance mechanisms include efflux pumps (like the Pmt transporter in S. aureus), membrane charge modifications, and protease secretion.^[5] The membrane-targeting mechanism of AMPs makes complete resistance difficult because bacteria cannot fundamentally redesign their cell membranes without losing viability.

Why aren't antimicrobial peptides used as drugs yet?

Several AMPs are in clinical use (nisin, polymyxins, daptomycin, gramicidin), and others are in clinical trials (pexiganan, omiganan). Broader adoption is limited by short half-lives due to proteolytic degradation, narrow therapeutic windows between bacterial killing and mammalian cell toxicity, high manufacturing costs, and challenges with systemic delivery.^[4] Most clinical success has been with topical applications.

Are antimicrobial peptides better than antibiotics?

AMPs and antibiotics have different strengths. AMPs kill faster (minutes vs hours), have broader spectrum activity, and face less resistance evolution. Antibiotics have better pharmacokinetics, established manufacturing, proven systemic delivery, and decades of clinical data. The most promising approach may be using them together: AMP-antibiotic combinations show synergistic effects, with AMPs permeabilizing membranes to enhance antibiotic entry.^[10]

How is AI being used to discover new antimicrobial peptides?

Machine learning models trained on databases of known AMPs can predict antimicrobial activity, toxicity, and stability from peptide sequences alone. These models screen millions of candidate sequences computationally, identifying novel peptides optimized for specific bacterial targets.^[12] De novo generation models can design entirely new peptides not found in nature, though all computational predictions still require experimental validation.

Do antimicrobial peptides work against biofilms?

AMPs show particular promise against biofilms because their membrane-targeting mechanism can penetrate biofilm matrices that block conventional antibiotics. A 2026 study showed a plant-derived AMP eradicated pre-formed A. baumannii biofilms at concentrations where polymyxin B failed.^[8] A complementary approach uses antibodies to disrupt biofilms, releasing bacteria into a vulnerable state where natural AMPs like defensins and LL-37 can kill them.^[9]