A practical guide to designing and running peptide pull-down assays with biotinylated baits. Covers bait length and modification placement, N-terminal versus C-terminal biotin chemistry, lysis and incubation conditions, elution, and detection by SDS-PAGE or mass spectrometry, including a…
A peptide pull-down assay uses a short synthetic peptide as bait to capture the proteins that bind it. The practical recipe for a researcher who wants to identify or confirm a protein-protein interaction is this: design a biotinylated peptide of about 15 to 20 amino acids centered on the region or modification of interest, immobilize it on avidin or streptavidin beads, incubate the beads with a clarified cell lysate at 4°C for about 6 hours, wash away unbound material, elute the captured proteins, and analyze them by SDS-PAGE or by trypsin digestion followed by mass spectrometry. A matched control peptide , identical except for the feature being tested, runs in parallel throughout.
The assay serves two distinct purposes. It can confirm an interaction suggested by other techniques, such as co-immunoprecipitation, by showing that a peptide spanning the suspected interface captures the predicted partner. It can also screen for novel interactions without a prior hypothesis about identity. A phosphorylation site is the classic bait modification, but the design generalizes to any post-translational modification that creates or removes a binding surface.
The rationale for the method is straightforward. A protein's function is largely defined by its interaction partners, so identifying the molecules that bind a protein reveals what that protein does. Peptide baits carry that logic one step further: if a short stretch of sequence contains the information needed for recognition, a synthetic peptide bearing that stretch should capture the same partners as the full-length protein.
The foundational evidence that short peptides can occupy a protein's binding site with real specificity comes from work on the chaperone hsp70/BiP . In 1991, Gregory C. Flynn, Jan Pohl, Mark T. Flocco, and James E. Rothman filled the peptide-binding site of BiP with a pool of random-sequence peptides and analyzed what bound Nature 353, 726-730; 24 October 1991; doi:10.1038/353726a0 . They found that BiP, a chaperone of relative molecular mass 70K, can distinguish only unfolded forms of proteins, and that its peptide-binding site shows considerable specificity. The study established that the information required for recognition by a peptide-binding protein can reside in a short linear sequence, which is the precondition for the entire pull-down approach.
A related use exploits competition rather than capture. A synthetic peptide corresponding to a suspected interaction region, added in excess, can competitively disrupt the binding between two full-length proteins. If the predicted interface is the real one, the soluble peptide blocks the interaction; if the proteins still associate, the interface lies elsewhere. This makes peptide competition a useful check on interactions predicted from sequence or from pull-down results.
Sequence choice comes first. When confirming a predicted interaction, the bait should span the predicted interface, which usually means a contiguous stretch of one partner's sequence. When screening for modification-specific binding, the bait should be centered on the modified residue. The two design decisions that then dominate the outcome are the length of the bait and the placement of the modification.
For pull-down baits, a length of about 15 to 20 amino acids is ideal, with the modification centered and at least 6 to 8 flanking residues on each side. The flanking residues matter because most recognition domains contact more than the modified side chain alone; they engage a short stretch of surrounding backbone and side chains. Too short a peptide loses binding affinity and specificity, while longer peptides are more expensive, synthesize less cleanly, and are more prone to aggregation and non-specific binding. Centering the modification matters for the same reason: a binding domain typically approaches its target from one side, and if the modification sits within a few residues of a terminus, part of the recognition surface is missing.
Biotin is attached at the N- or C-terminus, and the choice is practical as well as chemical. N-terminal biotin modification is recommended over C-terminal modification because it has a higher success rate, a shorter turnaround time, and is easier to perform. C-terminal modification requires the addition of an extra lysine residue, which provides the attachment point for the biotin on the side chain of that lysine. Placing the biotin at the terminus rather than at an internal position leaves the modification of interest and its flanking residues free to engage binding proteins.
The affinity chemistry explains why the system needs no covalent crosslink between bait and captured protein. Avidin and streptavidin are both tetrameric proteins, and each tetramer can bind up to four biotin ligands. The biotin-avidin and biotin-streptavidin pairs have exceptionally high binding affinity, high enough that the peptide-resin complex survives the washing steps of a pull-down while unbound proteins are removed. The bait is loaded onto the resin in excess, and the loaded resin is used directly in the assay.
Chemically synthesized biotinylated peptides are produced at 80% purity . For baits of 15 to 20 residues, that figure is within the routine range of solid-phase synthesis . The caveats attached to it are discussed in the final section; at the design stage, it means the bait is a defined chemical entity with a known sequence rather than a mixture of natural isoforms.
Control baits are not optional. To identify modification-specific binding, the pull-down must be run in parallel with a control peptide identical in sequence and biotin placement but lacking the modification of interest. Comparing pull-downs performed with modified and unmodified control peptides identifies binding functions that depend on the modification. The specific bait and the control bait should come from the same synthesis scale and purification round, so that their purity is comparable.
| Parameter | Recommendation |
|---|---|
| Bait length | 15 to 20 amino acids |
| Modification position | Centered, 6 to 8 flanking residues on each side |
| Biotin site | N-terminal preferred |
| C-terminal biotin | Requires an additional lysine residue |
| Peptide purity | 80% |
| Control bait | Same sequence, no modification |
The conditions below follow the protocol used in the quantitative peptide pull-down work and in common practice. Cells are lysed in a buffer containing 1% v/v Nonidet P-40 NP-40 , 150 mM NaCl, and 50 mM Tris-HCl at pH 7.5, supplemented with protease and phosphatase inhibitors. Nonidet P-40 is a non-ionic detergent that solubilizes membranes and proteins while preserving non-covalent interactions. The 150 mM NaCl approximates physiological ionic strength, which keeps specific interactions intact while discouraging the electrostatic stickiness that plagues low-salt buffers. The inhibitors protect the modification on the bait and the captured proteins from degradation during the long incubation.
| Component | Concentration |
|---|---|
| Nonidet P-40 | 1% v/v |
| NaCl | 150 mM |
| Tris-HCl, pH 7.5 | 50 mM |
| Protease and phosphatase inhibitors | As required |
The lysate should be clarified by centrifugation before use, and the resin should be washed free of its storage buffer before the peptide is loaded. For a meaningful comparison, the specific bait and the control bait pull-downs must start from equal amounts of total protein. Load the biotinylated peptides onto avidin or streptavidin resin, then incubate the clarified lysate with the immobilized bait at 4°C for about 6 hours . The low temperature slows proteolysis and preserves labile modifications. The six-hour window balances two opposing pressures: long enough for low-abundance binders to associate, short enough to limit the non-specific binding that accumulates with time.
After incubation, wash the resin to remove unbound proteins. Wash stringency is the main lever on background. More washes or higher salt remove more non-specific proteins, but also risk stripping weak but genuine interactions. Elute the captured proteins by boiling in SDS sample buffer, which denatures everything on the resin, or by cleavage with 50 mM dithiothreitol DTT , which releases the bait from the resin through a reducible linker and carries the bound proteins with it. DTT elution leaves the resin behind and produces a cleaner sample for downstream analysis. A resin-only sample, incubated alongside the baits with the same resin but no peptide, helps assign bands that appear in both the bait and control lanes.
A failed pull-down usually fails in one of a few places. If no bands appear in either the bait or control lane, suspect that the bait never reached the resin, that the lysate was not properly solubilized or clarified, or that the wash was too aggressive for the interaction under study. If both lanes are full of identical bands, the problem is non-specific binding, and the first response is to increase wash stringency or shorten the incubation. If the bait and control lanes look alike but contain a few distinct bands, those proteins may bind the sequence regardless of the modification; that is interpretable information, not a failure.
| Step | Condition |
|---|---|
| Lysis | 1% NP-40, 150 mM NaCl, 50 mM Tris-HCl pH 7.5, inhibitors |
| Immobilize bait | Biotinylated peptide on avidin or streptavidin resin |
| Incubate | Clarified lysate with resin, 4°C, about 6 hours |
| Wash | Remove unbound proteins |
| Elute | Boiling SDS sample buffer, or 50 mM DTT |
| Analyze | SDS-PAGE, or trypsin digestion with mass spectrometry |
Detection usually starts with SDS-PAGE. Run the modified-bait eluate and the control-bait eluate side by side on a gel and stain it. Bands present only in the modified sample are modification-specific candidates. Bands common to both lanes are more likely to reflect non-specific binding to the peptide backbone, the biotin, or the resin, although a shared band can also be a partner that binds the sequence regardless of the modification, which is itself a useful result.
For identification, the captured proteins are digested with trypsin and analyzed by mass spectrometry. Peptide pull-downs typically retrieve a limited set of proteins from whole-cell lysates, and that simplicity is an advantage: the eluate is sparse enough that mass spectrometry identification is tractable, and the limited background is itself evidence that the bait is not capturing everything in the lysate.
The two elution routes feed different analytical workflows. Boiling in SDS sample buffer is the natural preparation for the gel-based route: the SDS is removed during electrophoresis, and the resolved bands are excised and digested with trypsin in the gel. DTT elution produces a soluble sample that can be digested directly in solution and cleaned up for mass spectrometry without a gel step. The choice affects sensitivity and convenience more than the identity of the results, and some protocols combine both, reserving a portion of the DTT eluate for a quick gel check before committing the rest to mass spectrometry.
A quantitative version of the screen, described in the Journal of Biological Chemistry in 2004 J. Biol. Chem. 279:10756-10764; doi:10.1074/jbc.M309909200 , uses stable isotope labeling to tell specific binders from background. Cells are grown in medium containing 13C so that every cellular protein carries heavy isotopes. The labeled lysate is incubated with one bait, and the equivalent unlabeled lysate with the other, typically the phosphorylated bait against the non-phosphorylated control. The two samples are analyzed together by mass spectrometry; a protein that binds one bait specifically appears with higher peak intensity for the labeled form. The method is unbiased, meaning no prior hypothesis about which proteins bind is required. It is highly specific and reproducible, requires only that the binding partner be present in the lysate, and retrieves only a limited set of proteins from whole-cell lysates.
What such a screen establishes, and what it does not, is worth stating precisely. The heavy-to-light ratio identifies proteins that bind the modified peptide in the context of a whole-cell lysate and at the local concentrations presented by the resin. Because the bait is a short peptide, a captured protein is usually in direct contact with it, although a bridging protein cannot be formally excluded. The result is not evidence that the interaction occurs in cells at endogenous concentrations, or that it has a function. Those questions require follow-up. A candidate can be checked by mutating the modification site and showing that the interaction is lost, by co-immunoprecipitation of the full-length proteins, or by competition with a soluble peptide of the same sequence, which should block capture. The experimenter designing such a screen today is not limited to 13C metabolic labeling; modern isobaric labeling schemes achieve the same logic, but the design principle, heavy and light lysates against modified and control baits, remains the assay's core.
Peptide pull-downs have hard limits that follow from their design. They can detect only binding partners that are present and accessible in the cell lysate. A partner that is not expressed in the chosen cell type, confined to a different compartment, or present below the detection threshold will not appear. An empty result therefore does not disprove an interaction; it shows only that no lysate protein bound the bait under these conditions.
The specificity of the result depends entirely on the controls. Without a matched control peptide, there is no way to distinguish specific interactions from non-specific binding to the peptide backbone, the biotin moiety, or the resin. Even with a control, the assay reports binding under non-physiological conditions: the bait is presented at high local density on a solid support, which can push weak or otherwise irrelevant interactions above the detection threshold.
The purity figure deserves a cautious reading. A claimed purity of 80% applies to the synthetic peptide preparation, not to the biological assay. In practice, a bait at that purity can contain a substantial fraction of failure sequences, truncations, and deletion products from solid-phase synthesis, and those impurities can act as additional baits with their own binding properties. For a pull-down, consistency between the bait and control preparations matters as much as the absolute purity number.
It also matters where the design recommendations come from. The guidance above, the 15 to 20 residue length, the 6 to 8 flanking residues, the preference for N-terminal biotin, and the purity figure, reflects the recommendations and capabilities of a commercial peptide synthesis provider LifeTein , not an independent systematic comparison of conditions. The underlying logic is well supported by the BiP study and the 13C screen, and the protocol is a sound starting point, but peptide length and biotin placement may need optimization for individual targets.
Several questions remain open. How peptide pull-downs compare quantitatively with co-immunoprecipitation for confirming a predicted interaction has not been systematically evaluated. The standard recommendations come with no decision tree for what to change when they fail, for example whether to lengthen the peptide, shift the modification, or switch biotin placement when a bait captures no specific partners. And beyond the examples cited here, the range of proteins captured in common peptide pull-down experiments is not well documented. The practical conclusion for a researcher adopting the method is to treat the protocol as a validated starting point, run the controls, interpret empty results cautiously, and verify every candidate interaction by an independent method.
Related reading: Epimerization Risk in Peptide Synthesis: Pathways and Control, Peptide Calculator: Molecular Weight, Charge, pI and Hydrophilicity, Detecting and Quantifying Small Peptides by SDS-PAGE, Best Canadian Peptide Vendors 2026: Ranked by Published Evidence.