Iterative capture, wash, elution and re-amplification of an M13 scFv/Fab library against immobilized antigen — progressively enriching rare high-affinity binders round over round
Phage display antibody panning begins with a physical link between genotype and phenotype: each M13 filamentous phage particle carries the DNA encoding an antibody fragment (scFv or Fab) packaged inside its capsid while displaying that same fragment fused to the pIII minor coat protein on its surface. This "one bug, one bug" logic is what makes iterative selection possible — bind the phenotype, recover the genotype, sequence it, and you know exactly which antibody you selected.
A phagemid is a hybrid plasmid: it carries an M13 origin of replication, a phage packaging signal, and an antibody fragment gene fused in-frame to a truncated pIII gene, but it lacks the rest of the M13 genome needed to make infectious particles on its own. VH and VL (or VH-CH1/VL-CL for Fab) genes are amplified from peripheral blood lymphocytes (naive/synthetic libraries) or from immunized donors/animals (immune libraries), assembled by overlap-extension PCR with a flexible (Gly4Ser)3 linker for scFv format, and cloned via SfiI/NotI restriction sites into vectors such as pCANTAB6, pComb3X, or pHEN1/pHEN2.
Ligated DNA is electroporated into E. coli TG1 (genotype supE, F'pilus+) at high efficiency (>10⁸ transformants/µg), and library diversity is typically capped by transformation efficiency rather than PCR diversity — a single large-scale electroporation of 1 mL of concentrated cells and 1 µg DNA in a 2 mm cuvette (2.5 kV, 25 µF, 200 Ω) can yield 10⁹–10¹⁰ independent clones.
Because the phagemid lacks the full M13 genome, transformed TG1 cells are superinfected with M13KO7 helper phage (multiplicity of infection ~20:1, 37°C non-shaking for 30 min to allow pilus-mediated adsorption, then shaking growth with kanamycin selection). M13KO7 carries a defective origin (p15A-derived, weakened packaging signal) so that phagemid genomes — with an intact, higher-affinity M13 packaging signal — are preferentially packaged over the helper genome itself, at typically 10–100:1 phagemid:helper ratio in the resulting virion pool.
Overnight secretion (30°C, 16–18 h, shaking) into 2×TY/carb/kan medium yields phage-containing supernatant that is PEG/NaCl precipitated (20% PEG-8000, 2.5 M NaCl) and resuspended in PBS to give library stocks of 10¹²–10¹⁴ cfu/mL, ready for the first round of selection. Library quality is checked before panning by colony PCR insert-check (>90% full-length insert expected) and by Sanger sequencing 24–48 random clones to confirm diversity and absence of stop codons or frameshifts.
The first panning round is a blunt instrument: with roughly 10⁹–10¹⁰ competing clones and only a handful of true binders present at very low frequency (often <1 in 10⁶), Round 1 exists mainly to remove the overwhelming majority of phage that display irrelevant or non-binding fragments, not to isolate the best binder outright. Losing a good clone here to statistical bad luck is common, which is why replicate immunotubes and moderate — not harsh — wash stringency are used.
Two dominant immobilization strategies are used. Direct adsorption panning coats antigen (10 µg/mL in carbonate buffer, pH 9.6, 4°C overnight) onto a Nunc MaxiSorp immunotube or 96-well plate via passive hydrophobic/electrostatic adsorption — simple but can partially denature the antigen and expose cryptic, non-native epitopes. Solution-phase panning instead biotinylates the antigen (EZ-Link NHS-biotin, ~1–2 biotins per antigen molecule to avoid steric interference) and captures it on streptavidin- or neutravidin-coated paramagnetic beads (Dynabeads M-280) after a brief pre-incubation, preserving native conformation and allowing precise control of antigen concentration and valency — critical for later affinity-driven rounds.
After blocking (2% milk-PBS or 3% BSA-PBS, 1 hour, room temperature, on both the plate/beads AND the phage library separately to remove plastic/streptavidin-binding "sticky" clones), the library (typically 10¹²–10¹³ cfu in 1 mL) is added and incubated 1–2 hours with gentle rotation, allowing diffusion-limited binding of the rare cognate phage among a vast excess of non-binders.
Washing at this stage uses only 5–10 cycles of PBS containing 0.05–0.1% Tween-20, each a fill-and-dump or fill-and-aspirate cycle of roughly 5 minutes. This is deliberately gentle: with 10⁹ clones competing, even weak specific binders can be lost if wash stringency is set too high in Round 1, and there is no way yet to know which few clones in the pool are true positives. Typical output after Round 1 is 10⁶–10⁷ cfu — an apparent recovery of only 0.001–0.1% of input, reflecting that most of the library never touched the antigen and most of what did was low affinity and washed away. The eluted pool at this stage is still enormously diverse (essentially unchanged from the naive library, ~10⁹ clones) but is now enriched perhaps 2–5-fold for antigen-reactive sequences relative to background — a signal too weak to detect by sequencing alone, which is why 3–4 total rounds are standard.
Between successive panning rounds, selective pressure is deliberately ratcheted upward: more wash cycles, higher detergent concentration, shorter incubation times, lower antigen coating density, and competition with excess soluble antigen all combine to shift selection from "does this phage bind at all" toward "does this phage bind with a slow enough off-rate (koff) to survive extended, competitive washing." This is the round-over-round engine of enrichment.
The central biophysical lever in escalating rounds is dissociation kinetics. During a wash cycle, bound phage face repeated dilution into fresh buffer; the probability a given clone remains bound after n wash cycles of duration t scales with exp(-n·koff·t). A clone with koff = 10⁻² s⁻¹ (weak, fast-dissociating) is washed away almost completely after 20 cycles of 5-minute soaks, while a clone with koff = 10⁻⁴ s⁻¹ (slow, tight-binding) survives. Extending wash duration and cycle count therefore acts as a kinetic off-rate filter largely independent of on-rate (kon), which is valuable because therapeutic antibody utility correlates more with long residence time than with fast association.
A second lever is antigen valency and density. Multivalent phage (each virion can display several copies of the antibody fragment when helper-phage-derived wild-type pIII is limiting, or when using phage-display formats with higher pIII copy number) can achieve high apparent avidity against densely coated antigen even with intrinsically weak monomeric affinity — a false positive that Round 1 tolerates but later rounds must eliminate. Reducing antigen coating concentration in Rounds 2–3 (from ~10 µg/mL down to 0.1–1 µg/mL, or reducing bead-bound antigen amount) forces phage to rely on true monovalent affinity rather than multivalent avidity, since sparse antigen spacing makes simultaneous bivalent engagement geometrically unlikely.
Competitive elution is a third tool: instead of (or in addition to) low-pH elution, an excess of soluble, non-biotinylated antigen (often 100–1,000× molar excess) is added during the final wash or elution step. This directly competes off bound phage in a koff-dependent manner and, because it is a specific rather than a global disruption (pH/protease), can preferentially recover phage bound at the antigen's native epitope rather than at a denatured or plastic-associated artifact. Pre-adsorption ("negative selection" or "deselection") against a closely related off-target antigen, or against blank plate/beads, is typically layered into Rounds 2–3 to remove cross-reactive or matrix-binding clones before positive selection, sharpening specificity in addition to affinity.
By the end of Round 3, cumulative enrichment of specific binders is typically 100- to 1,000-fold over the naive frequency, and the underlying clonal diversity has usually collapsed from ~10⁹ starting sequences down to a few hundred to a few thousand dominant clones, several of which begin to be detectable by simple colony PCR fingerprinting (BstNI/HinfI restriction digest patterns) even before sequencing.
Selection only has value if the genotype of a bound phage can be recovered and propagated. Elution physically or chemically disrupts the antibody-antigen interaction to release bound phage into solution, after which infection of fresh E. coli and helper-phage superinfection regenerates a phage pool for the next round — or, at the final round, for direct plating and monoclonal picking.
Acid elution with 0.1 M glycine-HCl (pH 2.2, 5–10 minutes at room temperature with gentle agitation) is the most widely used method: low pH protonates ionizable side chains at the paratope-epitope interface, disrupting the hydrogen bonds and salt bridges that stabilize binding, and releases essentially all bound phage regardless of clone identity. Eluate is immediately neutralized with 1 M Tris-HCl pH 9.1 (roughly 1/10 volume) since M13 particles lose infectivity below pH 3 with prolonged exposure. An aliquot of residual antigen-bound phage can also be recovered by directly adding log-phase TG1 to the washed tube/beads ("on-bead infection"), which captures very tight binders that acid elution alone might not fully release.
Enzymatic elution offers an epitope-agnostic alternative: a protease cleavage site (commonly for trypsin or a tobacco etch virus [TEV] site) engineered into the linker between the antibody fragment and pIII allows brief protease treatment to sever the displayed fragment from the phage particle, releasing phage that carries no risk of chemical damage. This is especially useful for antigens or antibody frameworks sensitive to low pH, though it does cleave off the antibody fragment itself (leaving genotype intact, since the encoding DNA remains packaged) and is used only when acid elution is contraindicated.
Eluted phage are used to infect a fresh culture of exponentially growing E. coli TG1 (OD600 ≈ 0.4–0.6, ensuring abundant F-pili for efficient M13 adsorption): eluate and cells are mixed and incubated static at 37°C for 30 minutes (allowing phage adsorption without shearing pili), then shaken at 37°C for an additional 30 minutes before plating serial dilutions on TYE agar with carbenicillin/ampicillin and 1–2% glucose (to repress leaky antibody expression before selection) for titering, and inoculating the bulk infected culture into 2×TY/carb/glucose for overnight growth. The next morning, cells are sub-cultured, glucose is removed, and M13KO7 helper phage is added to rescue a new phage pool — regenerating an enriched sub-library ready for the next, more stringent round. Output titers typically climb from ~10⁶–10⁷ cfu after Round 1 to 10⁸–10⁹ cfu by Round 3–4, and this titer increase itself is one of the simplest real-time readouts that panning is converging on true binders rather than background.
The final round of output is a polyclonal population, even after aggressive enrichment — it must be resolved into individual monoclonal antibody fragments, each screened, sequenced, and biophysically ranked before any single clone can be called a "hit." This last stage converts an enriched phage pool into an actual short list of candidate antibodies with measured affinities, ready for reformatting into IgG and functional characterization.
Individual colonies from the final round's titering plate (typically 96 to ~380, spanning one to four 96-well blocks) are picked into deep-well plates containing 2×TY/carb/glucose, grown to log phase, then induced to secrete monoclonal phage by adding M13KO7 helper phage (or, for phagemid-only formats, IPTG to induce soluble scFv/Fab expression directly). Culture supernatants containing monoclonal phage (or periplasmic/secreted soluble fragment) are transferred to ELISA plates pre-coated with antigen in one set of wells and BSA or blocking buffer alone in a parallel set, then detected with an anti-M13 coat protein (pVIII)-HRP conjugate (for phage ELISA) or an anti-tag (His/FLAG/Myc)-HRP conjugate (for soluble fragment ELISA), developed with TMB substrate and read at OD450.
A clone is scored positive when its antigen-well signal exceeds roughly 3–5× the BSA-well background (or a fixed absolute OD450 cutoff, commonly >1.0 after 10–15 minutes of TMB development) — false positives from plastic-binding or streptavidin-binding phage are excluded by this parallel background subtraction. Typical hit rates in a well-executed 3–4 round campaign range from 20% to 70% of picked colonies, depending on target druggability and how aggressively stringency was escalated.
Positive clones are Sanger-sequenced across the VH/VL or Fab insert; CDR-H3 sequence (the most diverse and typically epitope-determining loop) is used to cluster clones into sequence families, collapsing what might be 100+ ELISA-positive colonies into a much smaller number (often 5–30) of truly unique binders — the remainder are frequently siblings from a single dominant clone that outcompeted others during amplification.
Unique clones are then reformatted as soluble scFv, Fab, or full IgG and kinetically ranked using label-free biophysical methods: biolayer interferometry (Octet, immobilizing antigen on an AR2G or Ni-NTA biosensor and dipping into serial antibody dilutions) gives rapid, medium-throughput koff/kon/Kd estimates in a single afternoon for dozens of clones, while surface plasmon resonance (Biacore, typically 1:1 Langmuir fit) provides a higher-precision, lower-throughput confirmatory measurement for the shortlisted top candidates — often the final 3–10 clones before progressing to developability assessment (thermal stability, aggregation propensity, polyspecificity reagent [PSR] binding, and expression titer).
In a representative synthetic Fab-library campaign against a checkpoint-receptor extracellular domain, output titer rose from 2.1×10⁶ cfu after Round 1 to 3.8×10⁹ cfu by Round 4 — roughly 1,800-fold enrichment. Of 190 colonies screened by monoclonal ELISA, 61% (116/190) were antigen-positive; CDR-H3 clustering collapsed these to 14 unique sequence families, and Octet ranking identified the top clone at Kd = 0.9 nM with a koff of 4×10⁻⁵ s⁻¹ — a binder subsequently reformatted to full IgG1 and advanced into affinity maturation and developability screening.