Polymyxins (Colistin): Last-Resort Peptide Antibiotics

Peptide Antibiotics

1959 Year Discovered

Polymyxins were isolated from Paenibacillus polymyxa in 1947 and entered clinical use in the late 1950s. Abandoned for decades due to kidney toxicity, colistin (polymyxin E) has been forced back into service as the last line of defense against gram-negative superbugs.

Falagas & Kasiakou, Clinical Infectious Diseases, 2005

Falagas & Kasiakou, Clinical Infectious Diseases, 2005

View as image

In 2015, researchers in China identified a gene called mcr-1 on a mobile plasmid in E. coli isolated from pigs and hospital patients. That discovery, published by Liu et al. in The Lancet Infectious Diseases, confirmed what infectious disease specialists had feared: resistance to polymyxins, the antibiotics held in reserve for bacteria that nothing else could kill, had become transferable between bacterial species.^[1] Polymyxins are cyclic lipopeptide antibiotics, peptide molecules with a fatty acid tail that physically tear apart gram-negative bacterial membranes. Their story is a cautionary arc in modern medicine: discovered in the 1940s, abandoned in the 1980s for severe nephrotoxicity, and dragged back into clinical use decades later because multidrug-resistant superbugs left physicians with no alternatives.

For context on how antimicrobial peptides broadly compare to conventional antibiotics, see Antimicrobial Peptides as Alternatives to Antibiotics: Can They Solve Resistance?.

The Polymyxin Family: Structure and Members

Polymyxins are nonribosomal peptides produced by the soil bacterium Paenibacillus polymyxa. Five variants (A through E) were originally identified in the 1940s, but only two reached clinical use: polymyxin B and polymyxin E, better known as colistin. Both are cyclic lipopeptides consisting of a 10-amino-acid peptide ring linked to a fatty acid tail. Six of the ten amino acids are L-alpha,gamma-diaminobutyric acid (L-Dab), which gives polymyxins their strong positive charge at physiological pH.^[2]

The structural difference between polymyxin B and colistin is minimal: a single amino acid substitution at position 6 (D-phenylalanine in polymyxin B versus D-leucine in colistin). This small change produces measurably different pharmacokinetic profiles. Polymyxin B is administered directly as the active sulfate salt. Colistin is administered as colistimethate sodium (CMS), an inactive prodrug that undergoes hydrolysis in vivo to release active colistin. This prodrug conversion is slow and variable, which creates dosing challenges that polymyxin B avoids.

The fatty acid tail is essential to function. It anchors the molecule into the lipid membrane after the cationic peptide ring binds to lipopolysaccharide on the bacterial surface. Without the tail, polymyxins lose most of their antibacterial activity. Velkov et al.'s 2010 structure-activity analysis in the Journal of Medicinal Chemistry established that the amphipathic balance between the cationic ring and hydrophobic tail determines both antimicrobial potency and mammalian cell toxicity.^[2]

This amphipathic architecture is shared with other peptide antibiotics in clinical use. Daptomycin uses a similar lipopeptide structure to target gram-positive bacteria, while vancomycin takes a glycopeptide approach to cell wall disruption. Polymyxins are unique in their specificity for gram-negative organisms.

How Polymyxins Kill Bacteria

Polymyxins kill gram-negative bacteria through a multi-step process of membrane destruction. The mechanism has been studied since the 1960s, with structural and biophysical research progressively refining the model.

Step 1: Electrostatic Binding to LPS

The initial interaction is electrostatic. Gram-negative bacteria have an outer membrane containing lipopolysaccharide (LPS), a molecule with multiple negatively charged phosphate groups on its lipid A component. The positively charged L-Dab residues on the polymyxin ring bind these phosphate groups with high affinity, displacing the divalent cations (calcium and magnesium) that normally bridge and stabilize adjacent LPS molecules.^[2] This displacement destabilizes the outer membrane packing. A 2026 study on AMP resistance in Salmonella confirmed that LPS modifications reducing surface negative charge directly decrease polymyxin binding efficiency.^[3]

Step 2: Outer Membrane Disruption

After binding, the fatty acid tail of the polymyxin molecule inserts into the hydrophobic interior of the outer membrane. This insertion, combined with the displacement of stabilizing cations, creates packing defects in the lipid bilayer. The outer membrane becomes permeable. Small molecules and even other antibiotics can now pass through. This is the basis for polymyxin synergy with other drugs: by permeabilizing the outer membrane, polymyxins allow molecules that would otherwise be excluded to reach their intracellular targets.

Step 3: Inner Membrane Damage and Cell Death

Polymyxins that cross the destabilized outer membrane then interact with the inner (cytoplasmic) membrane. The same electrostatic and hydrophobic interactions repeat: binding to anionic phospholipids (phosphatidylglycerol, cardiolipin), insertion of the fatty acid tail, and membrane disruption. The resulting loss of membrane integrity collapses the proton motive force, causes uncontrolled ion flux, and leads to cell lysis.^[4]

For a detailed explanation of how peptides form pores in bacterial membranes, see How Antimicrobial Peptides Kill Bacteria: Pore Formation Explained.

Additional Mechanisms

Evidence from the past decade suggests that membrane disruption alone does not fully explain polymyxin lethality. Polymyxins also bind to lipid A in the inner membrane, potentially inhibiting essential respiratory chain enzymes. They generate reactive oxygen species (ROS) within bacterial cells, contributing to oxidative damage. And their initial binding to LPS neutralizes endotoxin activity, which has clinical relevance in gram-negative sepsis where free endotoxin drives the inflammatory cascade.

The Rise, Fall, and Return of Colistin

Discovery and Early Clinical Use (1947-1980s)

Polymyxins were first isolated in 1947. Colistin entered clinical use in the late 1950s and was widely prescribed throughout the 1960s and 1970s for gram-negative infections, particularly those caused by Pseudomonas aeruginosa. During this period, reports of nephrotoxicity and neurotoxicity accumulated. By the 1980s, the arrival of aminoglycosides, carbapenems, and fluoroquinolones offered less toxic alternatives, and polymyxin use was largely discontinued in Western medicine.^[5]

Colistin continued to be used extensively in veterinary medicine and agriculture, particularly in China, where it was added to animal feed as a growth promoter at enormous scale. This widespread agricultural use created a massive selective pressure for colistin resistance to evolve in environmental and animal-associated bacteria.

The Forced Revival (2000s-Present)

By the early 2000s, the emergence of carbapenem-resistant Enterobacteriaceae (CRE), multidrug-resistant Pseudomonas aeruginosa, and extensively drug-resistant Acinetobacter baumannii (XDR-Ab) left clinicians with a critical gap: infections that no approved antibiotic could treat. Falagas and Kasiakou's 2005 review in Clinical Infectious Diseases formally argued for reconsidering colistin, documenting that its historical toxicity had been overestimated by earlier studies that used imprecise dosing.^[5]

The return was reluctant. Colistin was not a good drug brought back for new purposes; it was a problematic drug brought back because nothing better existed. A 2019 review of cationic antimicrobial peptides described colistin as both "the paradigmatic example of a peptide antibiotic in clinical use" and "a cautionary tale about what happens when the antibiotic pipeline runs dry."^[6]

A 2026 case report documented the use of a triple combination of aztreonam, colistin, and tigecycline to treat a neonatal carbapenemase-producing Enterobacteriaceae infection, illustrating how colistin now serves as a backbone in last-resort combination regimens.^[7]

Nephrotoxicity: The Dose-Limiting Problem

Kidney damage remains the primary barrier to effective polymyxin therapy. Colistin-associated acute kidney injury (AKI) occurs in up to 38% of patients in some clinical series. The mechanism involves accumulation of polymyxin molecules in renal proximal tubular epithelial cells through megalin receptor-mediated endocytosis and PEPT2 transport. Once inside these cells, polymyxins trigger multiple damage pathways: mitochondrial dysfunction, endoplasmic reticulum stress, oxidative stress through ROS generation, and activation of both caspase-dependent and caspase-independent apoptosis.^[8]

Nephrotoxicity is dose-dependent but also time-dependent. Longer treatment courses produce higher rates of kidney injury. This creates a therapeutic tension: polymyxins are typically needed for severe infections that require extended treatment, but extended treatment is precisely what causes the most kidney damage. There is no approved nephroprotective agent that reliably prevents polymyxin-induced kidney injury, though preclinical studies have shown that antioxidants (N-acetylcysteine, vitamin C, melatonin) can attenuate damage in animal models.

Polymyxin B and CMS-derived colistin differ in their nephrotoxic profiles. Some evidence suggests polymyxin B produces more predictable drug levels and potentially less kidney toxicity than CMS, because CMS requires unpredictable in vivo conversion to active colistin. A 2025 multicenter retrospective cohort comparing colistin sulfate to polymyxin B for carbapenem-resistant Acinetobacter baumannii infections found comparable efficacy but differing renal toxicity profiles, though the study had important confounders.

The MCR-1 Crisis: Transferable Resistance

Before 2015, polymyxin resistance was known but considered manageable. Resistance arose through chromosomal mutations that modified the lipid A portion of LPS, reducing its negative charge and therefore its affinity for the cationic polymyxin ring. These mutations imposed fitness costs on bacteria and could not spread between species.

The 2015 discovery of mcr-1 changed the calculus entirely. Liu et al. identified the mcr-1 gene on a plasmid in E. coli isolates from pig farms and hospital patients in China. The gene encodes a phosphoethanolamine transferase enzyme that adds a phosphoethanolamine group to lipid A, reducing the net negative charge of the outer membrane. Because mcr-1 sits on a mobile genetic element, it can transfer horizontally between bacterial species through conjugation at frequencies of 10^-1 to 10^-3 cells per recipient cell.^[1]

Since the initial report, mcr-1 has been detected on all six inhabited continents: in humans, livestock, wildlife, food products, and environmental samples including wastewater and soil. A 2025 study detected genes associated with both polymyxin and antimicrobial peptide resistance in clinical Pseudomonas aeruginosa isolates, demonstrating that resistance mechanisms are spreading beyond Enterobacteriaceae.^[9] At least ten mcr gene variants (mcr-1 through mcr-10) have now been identified, each encoding a phosphoethanolamine transferase with slightly different properties.

A 2025 study in Nature Communications showed that plasmid-driven virulence and antibiotic resistance are converging in E. coli, with virulence plasmids increasingly co-carrying resistance genes including those conferring polymyxin resistance.^[10] This convergence means that the most dangerous bacteria (those causing the most severe infections) are also becoming the most drug-resistant.

The discovery prompted China to ban colistin as an animal feed additive in 2017. Several other countries followed with restrictions on agricultural colistin use. Whether these policy changes will slow the spread of mcr genes remains an open question; the genetic elements are already widely disseminated, and horizontal gene transfer does not require ongoing selective pressure to maintain plasmid carriage in all populations.

Polymyxin Synergy: Extending the Last Resort

Given the toxicity and resistance challenges, researchers have focused on combination strategies that lower the required polymyxin dose while maintaining or enhancing bactericidal activity. The logic is pharmacological: if polymyxins permeabilize the outer membrane at sub-inhibitory concentrations, other antimicrobials can penetrate and kill.

Polymyxin-Antibiotic Combinations

A 2026 study by Story et al. in ACS Infectious Diseases demonstrated that peptide-linked amikacin conjugates showed broad-spectrum activity against extensively drug-resistant (XDR) and pandrug-resistant bacteria when combined with polymyxin B at sub-MIC (minimum inhibitory concentration) levels.^[11] The conjugates exploited polymyxin's membrane-permeabilizing action to deliver the aminoglycoside payload directly into bacteria that would normally exclude it. This study specifically targeted WHO priority pathogens including carbapenem-resistant A. baumannii and P. aeruginosa.

Antimicrobial Peptide Enhancement

A 2026 study found that temporin-GHa-derived peptides (short alpha-helical AMPs originally isolated from frog skin) enhanced both the antibacterial and antibiofilm activities of polymyxin B against Pseudomonas aeruginosa and E. coli.^[12] The mechanism appears to involve complementary membrane disruption: temporin attacks the outer membrane from a different angle than polymyxin, creating synergistic damage that exceeds what either peptide achieves alone.

Novel Synergy Partners

A 2025 study demonstrated synergistic membrane disruption when colistin was combined with cannabidiol (CBD) against gram-negative bacteria. The polycationic peptide colistin disrupted the outer membrane, while the lipophilic CBD molecule destabilized the inner membrane through a complementary mechanism, lowering the effective dose of both compounds.^[4]

Geitani et al.'s 2019 study in BMC Microbiology examined cationic antimicrobial peptides as adjuvants to antibiotics against both methicillin-resistant Staphylococcus aureus and multidrug-resistant Pseudomonas aeruginosa, finding that certain AMP-antibiotic combinations produced synergistic killing that neither agent achieved independently.^[6]

Beyond Polymyxins: Next-Generation Peptide Antibiotics

The limitations of polymyxins have driven intense research into improved peptide antibiotics that retain broad-spectrum gram-negative activity while reducing toxicity. Several approaches are showing progress.

Engineered AMPs Targeting Gram-Negative Superbugs

A scorpion-derived antimicrobial peptide, AaeAP2a, demonstrated activity against carbapenem-resistant Acinetobacter baumannii through membrane disruption and triggered metabolic collapse in the pathogen. The peptide achieved bactericidal concentrations with minimal hemolytic activity, a critical safety parameter for systemic use.^[13]

Preclinical assessment of peptide-based antimicrobials against carbapenem-resistant A. baumannii published in 2026 showed both in vitro and in vivo efficacy, with the authors describing the approach as having "translational potential" for clinical development.^[14]

A plant-derived antimicrobial peptide demonstrated antibacterial and antibiofilm activities comparable to or exceeding polymyxin B, operating through multiple mechanisms of action that may reduce the likelihood of resistance development.^[15]

AI-Designed Peptide Antibiotics

Machine learning is accelerating peptide antibiotic discovery. A 2026 study used a pre-trained and fine-tuned few-shot learning pipeline to discover novel antimicrobial peptides specifically targeting Acinetobacter baumannii.^[16] The approach generated candidates from limited training data, addressing a bottleneck in peptide drug discovery where experimental validation is slow and expensive.

In silico design and evaluation of hybrid antimicrobial peptides for combating multidrug-resistant environmental bacteria combined features from multiple natural AMP templates to create chimeric molecules with enhanced activity profiles.^[17] The computational approach allowed rapid screening of structural variants that would take years to test experimentally.

For the current state of antimicrobial peptides in clinical development, see Which AMPs Are in Clinical Trials? A 2026 Pipeline Tracker.

Natural AMP Discovery

The natural world continues to yield promising candidates. Novel alligator cathelicidin As-CATH8 demonstrated anti-infective activity against clinically relevant and crocodylian bacterial pathogens, including species with intrinsic polymyxin resistance.^[18] Cathelicidins from reptiles are of particular interest because these animals thrive in microbe-rich environments without developing frequent infections, suggesting their antimicrobial peptides have been optimized by millions of years of selection pressure.

Dosler et al. demonstrated that antimicrobial peptides can inhibit and destroy Pseudomonas aeruginosa biofilms, a property that conventional antibiotics (including polymyxins) struggle with because biofilm-embedded bacteria are shielded from drug penetration.^[19] Biofilm formation is a major driver of chronic and device-associated infections, and peptide-based biofilm disruption represents a therapeutic avenue that polymyxins alone cannot address.

For more on how nature's peptide pharmacies are being explored, see Marine Antimicrobial Peptides: The Ocean's Untapped Pharmacy and Nisin: The Food-Grade Antimicrobial Peptide in Your Cheese.

Polymyxins in Context: How They Compare to Other Peptide Antibiotics

Polymyxins are one of several peptide-based antibiotics in clinical use, each targeting bacteria through distinct structural strategies. Understanding where polymyxins fit in this landscape clarifies both their unique value and their limitations.

Daptomycin is a cyclic lipopeptide like polymyxins, but it targets gram-positive bacteria exclusively. Daptomycin inserts into bacterial membranes in a calcium-dependent manner, forming oligomeric pores that depolarize the membrane. Its gram-positive specificity arises from its reliance on phosphatidylglycerol-rich membranes, which are more accessible in gram-positive organisms that lack an outer membrane barrier. Polymyxins and daptomycin are complementary rather than overlapping: polymyxins cover the gram-negative pathogens that daptomycin cannot reach, and vice versa.

Vancomycin takes a fundamentally different approach. This glycopeptide antibiotic inhibits cell wall synthesis by binding to the D-Ala-D-Ala terminus of peptidoglycan precursors, preventing cross-linking. Vancomycin does not disrupt membranes at all. Its peptide backbone serves a structural role in target recognition rather than membrane insertion. Vancomycin is also gram-positive-specific, effective against MRSA and enterococci but useless against gram-negative organisms whose outer membrane blocks its entry.

Nisin, a bacteriocin produced by Lactococcus lactis, bridges the gap between food preservation and antimicrobial therapy. Nisin forms pores in gram-positive bacterial membranes and also binds lipid II, a cell wall synthesis precursor, giving it a dual mechanism. Unlike polymyxins, nisin has an extensive safety record in food applications spanning decades, though its clinical therapeutic use remains limited.

What unifies these peptide antibiotics is their reliance on physical interaction with bacterial structures rather than inhibition of single enzymatic targets. This structural approach makes resistance development slower than for conventional antibiotics, though as mcr-1 demonstrates for polymyxins and vanA demonstrates for vancomycin, no antibiotic is permanently resistance-proof.

Current Clinical Use: Indications and Dosing Challenges

Polymyxins are currently used in three main clinical contexts. Intravenous colistin (as CMS) or polymyxin B treats systemic infections caused by carbapenem-resistant gram-negative bacteria, primarily Acinetobacter baumannii, Pseudomonas aeruginosa, and Klebsiella pneumoniae. Inhaled colistin is used as adjunctive therapy for ventilator-associated pneumonia caused by multidrug-resistant gram-negatives, delivering high local concentrations to the lungs while reducing systemic exposure and nephrotoxicity risk. Topical polymyxin B appears in ophthalmic and otic preparations, often combined with other antibiotics, where systemic toxicity is not a concern.

Dosing remains a challenge. The 2019 International Consensus Guidelines for the Optimal Use of Polymyxins recommended loading doses for both colistin and polymyxin B to achieve therapeutic levels more rapidly, followed by maintenance dosing adjusted for renal function. However, achieving adequate drug exposure at the infection site while minimizing kidney damage requires therapeutic drug monitoring, a service not available in many hospitals, particularly in low-resource settings where polymyxin use is most common. The gap between recommended practice and real-world implementation is substantial.

The pharmacokinetic differences between CMS and polymyxin B have practical implications. CMS produces unpredictable colistin levels because the prodrug conversion varies between patients and is affected by renal function. Polymyxin B achieves more consistent plasma levels because it is administered as the active compound. Some experts now advocate for polymyxin B over CMS in systemic infections, though both formulations remain in widespread use and head-to-head randomized trials are limited.

The Evidence Landscape: What We Know and What We Do Not

Polymyxins occupy an unusual position in the evidence hierarchy. They are among the most studied peptide antibiotics in terms of mechanism, with decades of structural biology, biophysics, and microbiology research. They are also among the most poorly studied in terms of modern clinical trial data. Most contemporary clinical evidence comes from retrospective cohort studies, case series, and observational data rather than randomized controlled trials. This evidence gap exists because polymyxins are typically used as last-resort agents in critically ill patients, a population difficult to enroll in placebo-controlled studies and ethically impossible to randomize to no-treatment arms.

The pharmacokinetic understanding of polymyxins has improved substantially since the early 2000s, but significant gaps remain. CMS-to-colistin conversion rates vary widely between patients, making dose optimization difficult. Population pharmacokinetic models exist but have been validated primarily in ICU populations, with limited data for other clinical settings.

Resistance surveillance data is improving but inconsistent globally. mcr gene prevalence varies enormously by geography, reflecting different agricultural colistin usage patterns. Many low- and middle-income countries, where polymyxin use is highest and resistance is likely most prevalent, have the least robust surveillance systems.

The Bottom Line

Polymyxins are cyclic lipopeptide antibiotics that kill gram-negative bacteria through membrane destruction, a mechanism rooted in their peptide structure. Colistin's forced return to clinical use after decades of abandonment reflects the severity of the antimicrobial resistance crisis, while the global spread of plasmid-borne mcr resistance genes demonstrates that even last-resort peptide antibiotics face evolutionary pressure. The next generation of peptide antibiotics, informed by structural understanding of polymyxin mechanism and limitation, aims to retain the membrane-targeting strategy while engineering out the nephrotoxicity that makes polymyxin therapy a calculated risk.

Frequently Asked Questions

What is colistin and why is it called a last-resort antibiotic?

Colistin (polymyxin E) is a cyclic lipopeptide antibiotic that kills gram-negative bacteria by binding to and disrupting their cell membranes. It is called a last-resort antibiotic because it is typically reserved for infections caused by multidrug-resistant bacteria that are resistant to all other available antibiotics, including carbapenems. Colistin was largely abandoned in the 1980s due to kidney toxicity but was brought back into clinical use in the early 2000s when resistant superbugs left physicians with no alternatives.

How do polymyxins work against bacteria?

Polymyxins kill bacteria through a physical process of membrane destruction. The positively charged peptide ring binds to negatively charged lipopolysaccharide (LPS) in the bacterial outer membrane, displacing stabilizing calcium and magnesium ions. The fatty acid tail then inserts into the membrane, creating packing defects. This disrupts both outer and inner membranes, causing uncontrolled ion flux, loss of cellular contents, and cell death. The process is rapid and does not require bacteria to be actively dividing, unlike many conventional antibiotics.

What is the difference between polymyxin B and colistin?

Polymyxin B and colistin (polymyxin E) differ by a single amino acid at position 6 of their peptide ring: D-phenylalanine in polymyxin B versus D-leucine in colistin. The clinically significant difference is in administration. Polymyxin B is given as the active sulfate salt, while colistin is administered as the inactive prodrug colistimethate sodium (CMS), which must convert to active colistin in the body. This conversion is slow and variable, making polymyxin B pharmacokinetically more predictable.

Why does colistin cause kidney damage?

Colistin accumulates in renal proximal tubular epithelial cells through receptor-mediated endocytosis, primarily via the megalin receptor and PEPT2 transporter. Inside these cells, colistin triggers mitochondrial dysfunction, endoplasmic reticulum stress, oxidative stress through reactive oxygen species generation, and activation of apoptotic cell death pathways. Acute kidney injury occurs in up to 38% of patients receiving polymyxin therapy, making nephrotoxicity the primary dose-limiting factor.

What is the mcr-1 gene and why does it matter?

The mcr-1 gene, first identified in 2015 in China, encodes a phosphoethanolamine transferase enzyme that modifies the lipid A component of bacterial LPS. This modification reduces the negative charge of the outer membrane, decreasing polymyxin binding affinity. The gene sits on a mobile plasmid, meaning it can transfer between bacterial species through horizontal gene transfer at high frequencies. Since its discovery, mcr-1 has been detected on all six inhabited continents in humans, animals, and environmental samples. At least ten mcr variants (mcr-1 through mcr-10) have been identified.

Are there alternatives to polymyxins for drug-resistant gram-negative infections?

Research is actively developing alternatives. Engineered antimicrobial peptides from scorpion venom, plant extracts, and alligator cathelicidins have shown activity against carbapenem-resistant bacteria in preclinical studies. AI-designed peptides are accelerating discovery timelines. Combination strategies pairing polymyxins with other antimicrobials or novel peptides at sub-inhibitory concentrations can enhance efficacy while reducing the required polymyxin dose. Cefiderocol, a siderophore cephalosporin, and ceftazidime-avibactam are newer conventional antibiotics active against some resistant gram-negative organisms, though resistance to these agents is already emerging.