GST Pull-Down for Protein Domain Interaction Mapping: Construct Design and Experimental Strategy

When Standard GST Pull-Down Is Not Enough: The Case for Domain-Level Mapping

A full-length GST pull-down tells you that protein A binds protein B somewhere within the full-length sequence of A. For many biological questions, this is the endpoint. But if protein A contains multiple recognizable domains — such as separate N-terminal and C-terminal cytoplasmic regions, SH2 or SH3 domains, or transcription factor activation domains — a positive full-length result raises the next question: which domain is responsible?

Domain-level pull-down answers this by splitting the bait protein into discrete fragments, each expressed as a separate GST fusion, and testing each fragment individually for binding to the prey. This approach has been applied to diverse systems — for example, GST-RACK1 fragments spanning the N-terminal domain (residues 1–180) and C-terminal domain (residues 137–317) were used to map the interaction surface with the TIMAP phosphatase (Kim et al., 2008). The approach is also routinely used for ion channels, receptor tyrosine kinases, and transcription factors to pinpoint the minimal interaction region.

For a general introduction to GST pull-down principles, see our guide to GST pull-down principles and applications.

Designing GST Fusion Constructs for Truncated Domains

Defining domain boundaries

The quality of the input domain definition determines the quality of the result. Several sources should be consulted:

• Sequence-based predictions: InterPro, Pfam, and SMART databases provide computationally predicted domain boundaries. These are good starting points but often over-simplify actual domain extents.
• Structural data: If a crystal structure or AlphaFold prediction is available, identify structurally compact units. Flexible linker regions (typically 5–20 amino acids of low complexity) make good domain termini.
• Published functional data: Previous deletion mutagenesis studies or paralogous domain boundaries provide experimental grounding.

For example, a channel protein with cytoplasmic N-terminal domain (residues 1–393) and C-terminal domain (residues 750–931) separated by a transmembrane core has naturally defined boundaries at the membrane–cytoplasm junctions.

N-terminal vs C-terminal GST fusion

For a first-pass experiment with a novel domain, express both N-GST and C-GST versions if resources allow. If only one orientation can be tested, N-GST is the safer starting point because of GST's strong solubilizing effect.

Construct design checklist

• Choose expression vector: pGEX series (pGEX-4T, pGEX-6P) as standard; include a protease cleavage site (thrombin or PreScission) between GST and the domain
• Design primers for the domain boundaries with restriction sites or Gibson assembly overhangs
• Validate the entire insert by sequencing, especially the GST-linker-domain junction
• Test expression in BL21(DE3) first; try Rosetta or CodonPlus for eukaryotic proteins; ArcticExpress for solubility improvement
• Include a GST-only control construct (empty pGEX vector) in every experiment

Construct design for domain-level GST pull-down: full-length protein is divided into individual domains, each cloned with N-terminal or C-terminal GST tag, with an empty GST-only control construct run in parallel.

Experimental Controls for Domain-Specific Pull-Down

If the GST-only control pulls down the prey at levels comparable to the GST-domain fusion, switch the tag (MBP or His) or reverse the experiment. For complementary validation strategies, see our Co-Immunoprecipitation guide.

From Pull-Down to Identification: What Comes After the IP

Targeted verification by Western blot

If a specific candidate binding partner is already hypothesized, analyze the pull-down eluate by Western blot with an antibody against the candidate. This is the fastest, most direct readout and is the standard for domain-mapping experiments.

Discovery by gel excision and mass spectrometry

When the goal is to identify novel binding partners for a specific domain:

• Run the pull-down eluate on SDS-PAGE alongside the GST-only control lane.
• Excise bands present only in the GST-domain lane. Digest with trypsin.
• Identify by LC-MS/MS. Filter candidates by presence in ≥2 biological replicates and consistency with the subcellular localization of the bait domain.

For details on selecting the appropriate mass spectrometry strategy, see our guide to IP-coupled mass spectrometry.

Fine-mapping by progressive truncation

Once a binding domain is confirmed (e.g., residues 750–931), create nested shorter fragments — for example, 750–850, 800–900, 850–931 — and repeat the pull-down. The fragment that retains binding defines the minimal interaction region, which can often be narrowed from ~180 residues to ~50–80 residues in two to three rounds.

Comparison: Full-Length vs Domain-Level GST Pull-Down

Decision workflow for domain-level GST pull-down experiments, mapping the path from initial question through construct design to downstream identification strategy.

Common Pitfalls and Troubleshooting

Decision Workflow for Domain-Level GST Pull-Down

Scenario A: Known interactor — "which domain binds?"

• Divide the bait protein into domains based on sequence, structure, and published boundaries
• Clone each domain as an N-GST fusion; add C-GST if first round is negative
• Run individual pull-downs against the known prey
• Identify the domain that retains binding; progressively truncate to map the minimal region

Scenario B: Unknown interactor — "what binds to this domain?"

• Select the domain of interest based on biological rationale
• Clone as N-GST and C-GST fusions; express and purify both
• Pull down from biologically relevant lysate alongside GST-only control
• Excise bands unique to GST-domain; identify by LC-MS/MS; validate by reverse pull-down or co-IP

Scenario C: Multi-domain protein — "map all interaction surfaces"

• Design 3–6 domain constructs covering the full protein with ~20 amino acid overlaps between adjacent fragments
• Express and purify all constructs in parallel using a standardized protocol
• Run pull-downs against the same prey in a single batch
• Binding domains confirm interaction surfaces; non-binding domains serve as internal negative controls