Phage Display: How Viruses Find the Best Peptides
Peptide Discovery Technologies
10 billion peptide variants
A single phage display library can contain over 10 billion unique peptide sequences, each displayed on the surface of a bacteriophage for simultaneous screening against a target.
Cwirla et al., PNAS, 1990
Cwirla et al., PNAS, 1990
View as imageMost peptide drugs are designed rationally: researchers study a protein target, model the binding interface, and synthesize candidate sequences. Phage display inverts this approach. Instead of designing one peptide at a time, it tests billions simultaneously by genetically fusing random peptide sequences to the coat proteins of bacteriophages (viruses that infect bacteria). The phages that display peptides binding the target are physically captured, amplified, and sequenced. In a few rounds of this "biopanning" process, a library of 10 billion random sequences can be winnowed to a handful of high-affinity binders that no rational design process would have predicted.[1] Phage display is one of several peptide discovery methods covered in our pillar article on combinatorial peptide libraries.
Key Takeaways
- A single phage display library can screen over 10 billion unique peptide sequences against a target in a single experiment (Cwirla et al., PNAS, 1990)
- George P. Smith and Sir Gregory P. Winter shared the 2018 Nobel Prize in Chemistry for developing phage display of peptides and antibodies (Barderas et al., Analytical and Bioanalytical Chemistry, 2019)
- Phage-displayed cyclic peptide libraries achieve higher target affinities than linear libraries due to reduced conformational entropy (Deyle et al., Accounts of Chemical Research, 2017)
- Phage display has produced at least six FDA-approved antibody drugs, including adalimumab (the world's best-selling drug as of 2017)
- Noncanonical amino acid incorporation into phage-displayed peptides expands chemical diversity beyond the 20 natural amino acids (Hampton et al., Chemical Reviews, 2024)
- Phage display-based tumor-homing peptides like iRGD improve drug delivery to solid tumors in preclinical models
How phage display works
Phage display relies on a simple principle: if you insert a DNA sequence encoding a peptide into a gene for a bacteriophage coat protein, the peptide will be displayed on the phage's outer surface while the encoding DNA remains packaged inside. Each phage particle becomes a physical link between a peptide sequence (the phenotype) and its genetic blueprint (the genotype).[2]
George P. Smith first demonstrated this in 1985 using the filamentous bacteriophage M13. He inserted a foreign gene fragment into the gene for the pIII minor coat protein of M13, showing that the resulting fusion protein was displayed on the phage surface and could be recognized by antibodies. This proof-of-concept established the foundation for all subsequent phage display work.
Building the library
A phage display library is constructed by inserting randomized DNA sequences into the pIII or pVIII coat protein gene of M13 phage. Randomization is achieved by synthesizing oligonucleotides with degenerate codons (NNK or NNS) at each position, where N represents any nucleotide and K or S limits the third position to reduce stop codons. A library of peptides with 7 random amino acids (7-mer) can theoretically contain 20 to the 7th power, or 1.28 billion, unique sequences.[1]
In practice, library diversity is limited by transformation efficiency (the number of unique phage clones that can be produced in a single experiment). The largest M13 libraries reach 10 to 11 billion unique clones. This is large enough to sample a substantial fraction of all possible short peptide sequences but far short of the theoretical diversity of longer peptides.
Biopanning: the selection process
Biopanning is the iterative enrichment cycle that identifies target-binding peptides from the library.
Round 1: The entire phage library is incubated with the target protein, which is immobilized on a surface (plate, beads, or column). Phages displaying peptides that bind the target are retained; non-binders are washed away. The retained phages are eluted (typically by pH shift or competitive displacement) and amplified by infecting E. coli bacteria, producing billions of copies of each selected clone.
Rounds 2-4: The amplified selected pool is subjected to additional rounds of binding, washing, and amplification, with increasingly stringent wash conditions. Each round enriches the population for higher-affinity binders. After 3 to 5 rounds, individual phage clones are isolated, sequenced, and tested for target binding.
The beauty of biopanning is its practicality. A single researcher can screen billions of peptide sequences in a few weeks using standard molecular biology equipment. No computational prediction, structural knowledge, or target crystal structure is required. The phages do the work of finding what binds.[3]
From peptides to antibodies: the Nobel Prize
George P. Smith's original 1985 work demonstrated peptide display. Sir Gregory P. Winter at the MRC Laboratory of Molecular Biology in Cambridge then adapted the technology to display antibody fragments, enabling the directed evolution of antibodies with therapeutic properties. Their combined contributions earned one-half of the 2018 Nobel Prize in Chemistry.[2]
Winter's adaptation was transformative for drug development. Phage display of antibody libraries enabled the discovery of fully human therapeutic antibodies without the need for animal immunization. The first and most commercially successful phage display-derived antibody was adalimumab (Humira), an anti-TNF antibody that became the world's best-selling drug, reaching $20 billion in annual sales at its peak. At least six FDA-approved antibodies were discovered through phage display technology.
The peptide side of phage display, while less commercially prominent than antibody display, has produced critical research tools and drug leads across oncology, infectious disease, and targeted drug delivery.
Cyclic peptide libraries: higher affinity through constraint
Linear peptide libraries suffer from conformational flexibility. A short linear peptide can adopt many conformations in solution, and only a fraction of those conformations will bind the target. This flexibility reduces effective binding affinity and makes linear phage-displayed peptides poor drug candidates.
Cyclic peptide phage display addresses this limitation. Deyle et al. reviewed the field in a 2017 Accounts of Chemical Research article, documenting how cyclization through disulfide bonds, thioether linkages, or chemical crosslinkers constrains the displayed peptide into a defined shape. The reduced conformational entropy translates directly into tighter binding: cyclic phage-displayed peptides routinely achieve 10- to 100-fold higher affinities than their linear counterparts for the same target.[4]
Several approaches to cyclization have been developed:
- Disulfide-constrained libraries incorporate two cysteine residues flanking the random region, forming an intramolecular disulfide bond
- Bicyclic peptides use a chemical scaffold (such as tris-bromomethyl benzene) to link three cysteine residues, creating two loops on a rigid scaffold
- Thioether cyclization uses non-natural amino acids or chemical modification to form non-reducible cyclic structures
Bicyclic peptides from phage display have been particularly successful. The company Bicycle Therapeutics has built a clinical pipeline of "Bicycle" molecules discovered through phage display, with candidates in trials for solid tumors and other indications. For more on the broader significance of cyclic peptide structures, see our article on macrocyclic peptides.
Applications in peptide drug discovery
Tumor-homing peptides
Phage display has been used extensively to discover peptides that home to tumor vasculature or tumor cells after intravenous injection. The most clinically relevant example is iRGD (CRGDKGPDC), a cyclic peptide that binds integrin receptors on tumor blood vessels and, after proteolytic cleavage, binds neuropilin-1 to trigger transcytosis of co-administered drugs into tumor tissue.[5]
The iRGD peptide was discovered by Erkki Ruoslahti's group through in vivo phage display, where phage libraries were injected intravenously and phages that accumulated in tumors were recovered. This in vivo biopanning approach identifies peptides that not only bind tumor markers but survive circulation, extravasate, and penetrate tumor tissue. For a deeper look at how these peptides guide drug delivery, see how phage display discovers tumor-homing peptides.
Antiviral peptide discovery
Castel et al. reviewed phage display applications in antiviral research, documenting the discovery of peptides that block viral entry, inhibit viral proteases, and interfere with viral assembly. The approach is particularly valuable for rapidly emerging viruses where structural data may be limited, since phage display requires only the target protein, not a detailed understanding of its function.[3]
Antimicrobial peptide discovery
Jakob et al. reported in 2022 on phage display-based discovery of cyclic peptides targeting bacterial anti-virulence mechanisms. Rather than killing bacteria directly, these peptides block virulence factors, reducing the selective pressure for resistance development. The cyclic constraint improved both target affinity and proteolytic stability compared to linear hits from the same screen.[6]
Cancer immunotherapy
Goracci et al. reviewed phage display applications in cancer immunotherapy in 2020, including the discovery of peptide mimotopes (peptides that mimic tumor antigens for vaccine development), targeting ligands for nanoparticle drug delivery, and peptide inhibitors of immune checkpoint proteins.[5]
Expanding chemical space: noncanonical amino acids
Traditional phage display is limited to the 20 genetically encoded amino acids. This constraint excludes the vast chemical diversity available through non-natural amino acids, which are often critical for achieving drug-like properties (protease resistance, cell permeability, oral bioavailability).
Hampton et al. published a comprehensive 2024 Chemical Reviews article documenting methods to incorporate noncanonical amino acids into phage-displayed peptide libraries. Approaches include amber suppression (using engineered tRNA/aminoacyl-tRNA synthetase pairs to incorporate non-natural amino acids at amber stop codons), selective pressure incorporation (starving bacteria of a natural amino acid and providing a non-natural analog), and post-translational chemical modification of displayed peptides.[7]
These methods expand the chemical space accessible to phage display from 20 amino acid building blocks to potentially hundreds, bridging the gap between biological peptide libraries and synthetic chemical libraries. The integration of noncanonical amino acids into phage display is converging with AI-driven peptide design to create increasingly powerful discovery platforms.
Phage display vs. other display technologies
Phage display is the oldest and most widely used peptide display technology, but alternatives have emerged.
mRNA display achieves library diversities of 10 to the 13th power (10 trillion), roughly 1,000-fold larger than the largest phage libraries. It uses an in vitro translation system, avoiding the biological bottleneck of bacterial transformation. For a comparison of these technologies, see our article on mRNA display.
Ribosome display similarly operates in vitro but maintains the ribosome-mRNA-peptide complex throughout selection, allowing continuous evolution without cloning steps.
Bacterial and yeast surface display present peptides on living cell surfaces, enabling flow cytometry-based sorting that provides quantitative binding data during selection rather than the binary bound/unbound readout of biopanning.
Each technology has trade-offs. Phage display's advantages are simplicity, robustness, and decades of optimization. Its limitations are library size (capped by bacterial transformation) and restriction to natural amino acids in standard protocols. For applications requiring very large libraries or non-natural amino acids, mRNA display or synthetic library approaches may be preferred. For standard target-binding peptide discovery, phage display remains the workhorse technology.[8]
Limitations and ongoing challenges
Library bias. Not all peptide sequences are equally well-tolerated by M13 phage. Sequences that are toxic to E. coli, that interfere with phage assembly, or that are rapidly proteolyzed inside bacterial cells are underrepresented or absent from libraries. This means phage display systematically misses certain classes of peptides.
Target-unrelated peptides. Biopanning can enrich peptides that bind the plastic plate, the blocking agent, or other components of the selection system rather than the intended target. These "target-unrelated peptides" or "parasitic sequences" require careful negative selection controls to identify and exclude.
Affinity ceiling. Monovalent peptide-target interactions discovered by phage display typically plateau at low-micromolar to high-nanomolar affinities. Reaching the low-nanomolar affinities needed for drug candidates usually requires post-selection affinity maturation through mutagenesis, cyclization, or chemical modification.
Translation to drugs. A peptide that binds a target on a phage surface may not function as a drug. Cell permeability, metabolic stability, pharmacokinetics, and off-target effects must all be optimized after discovery, through separate medicinal chemistry campaigns. Phage display identifies the starting point, not the finished drug.
The Bottom Line
Phage display uses bacteriophages to screen billions of peptide sequences against molecular targets, identifying binders that rational design cannot predict. The technology earned the 2018 Nobel Prize in Chemistry and has produced at least six FDA-approved antibody drugs. For peptide drug discovery specifically, cyclic peptide phage display, noncanonical amino acid incorporation, and in vivo biopanning for tumor-homing peptides represent the most active research frontiers. Phage display remains the workhorse for peptide lead discovery, with mRNA display emerging as a complementary technology for ultra-large library screens.