Biology Atlas

Map / Assays

assay · intermediate · partly written

Target engagement in cells (CETSA, NanoBRET)

Assays that ask whether a compound reaches and physically binds its intended protein inside living cells, without asking whether the binding does anything.

Also called: CETSA, cellular thermal shift assay, NanoBRET target engagement, NanoBRET TE

What target engagement asks

Target engagement asks one narrow question: does the compound physically bind the intended protein inside a living cell? It does not ask whether the binding does anything. CETSA heats cells and sees whether the target survives higher heat when bound. NanoBRET lets the compound push a glowing tracer off a luciferase-tagged target, so light falls as it binds. 123

Reported, not yet replicated

Where it sits in the cascade more

A purified-protein assay says the compound can bind or inhibit. A cell viability assay says cells did something. Neither says the compound found its target inside the cell. Engagement data close that link, so a later "no effect" can be read as either "the target is not a vulnerability" (the STK33 story, see the linked scenario) or "the compound never got there".

This is orthogonal confirmation in the sense of the screening-cascade report: SPR shows binding to purified protein, these assays show binding in a cell, and both are still blind to function. 14

Reported, not yet replicated

How CETSA works more

Heating unfolds proteins, and unfolded protein clumps and drops out of solution. A bound ligand makes the fold harder to break, so more of the target stays soluble at a given temperature. The method was introduced by Martinez Molina, Nordlund and colleagues at Karolinska Institutet (Science 2013; only the abstract was read). Steps: treat cells or lysate with compound, heat, separate soluble from aggregated protein, measure the soluble fraction. 251

Reported, not yet replicated

How NanoBRET works more

The target is fused to NanoLuc, a 19 kDa luciferase. A cell-permeable fluorescent tracer, a known ligand with a dye attached, binds the target. When the tracer is close, the luciferase light excites the dye (BRET). An unlabelled compound that binds the same site displaces the tracer, so the BRET signal falls with dose. IC50 values were converted to apparent affinities with the Cheng-Prusoff equation. The two-wavelength ratio makes the signal less sensitive to cell number. (That energy transfer needs nanometre-scale closeness is general physics; the paper does not discuss distance.) 31

Reported, not yet replicated

Stated limits and how it fails more

CETSA: a missing shift does not prove no binding; heating time shifts the apparent AC50; a luciferase-fragment tag may change stability. NanoBRET: the tracer may bind differently from the parent compound; ligands that engage the target without competing with the tracer are false negatives. In the Takeda work, compound quenching made an ELISA readout unusable and forced a switch to Western blot, and reversible compounds had to be kept in buffer so they would not dissociate during processing. The ex vivo PBMC EC50 for compound 22 (660-700 nM) was about 5-fold above the L929 cell value (120 nM); the authors attribute this probably to plasma protein binding (PBMCs in 100% plasma, L929 in 10% FBS). The paper text says "lower" but its own table shows higher.

No opened source quantifies how much a NanoBRET fusion tag changes target behaviour; Robers tested N- and C-terminal fusions. No opened source documents misuse by practitioners; reading a shift or displacement as "the compound works" is the report author reading. 5361

Reported, not yet replicated

Cost: what was and was not found more

No public price for either assay was found on an opened page. Promega kit and tracer prices are behind a login; no service-lab fee was found. A search snippet quoting a price for a 100-assay kit was not confirmed and is not used here. 781

Reported, not yet replicated

Question answered

Does the compound reach the intended protein inside a living cell and physically bind it? It separates "the compound never got there" from "it got there and the target did not matter", two failures the cascade otherwise merges. 1

Reported, not yet replicated

Readout physical

CETSA: the fraction of the target protein still soluble after heating cells or lysate, read by Western blot, mass spectrometry or a luminescence readout. NanoBRET: a ratio of two emission wavelengths (450 and 610 nm filters) from a luciferase-tagged target and a fluorescent tracer; the ratio falls as the compound displaces the tracer. 593

Reported, not yet replicated

Not measured

Function: neither assay shows that binding changes what the protein does. CETSA gives no direct affinity and can miss real binders that do not shift the melt; the Takeda authors found no consensus formula to turn isothermal dose-response values into true affinity, so they report apparent affinity. NanoBRET misses ligands that engage the target without competing with the tracer, and works only for targets a tracer can reach and bind in live cells. 563

Reported, not yet replicated

Setting

Where in cascade

Sample prep

CETSA: treat cells or lysate with compound, heat, then separate soluble from aggregated protein (traditionally by centrifugation) and measure what is soluble. NanoBRET: express the target fused to NanoLuc, a 19 kDa luciferase, in cells, add a cell-permeable fluorescent tracer and the compound. Western and bead formats need a good antibody. 593

Reported, not yet replicated

Conditions

CETSA runs as a melt curve (range of temperatures, fixed dose) or an isothermal dose-response (one temperature, range of doses, giving a half-maximal concentration). Heating time shifts the apparent AC50, and the CASRAI guide advises caution above about 55 C in intact cells. NanoBRET equilibrium assays used about 2 x 10^4 cells per well in white non-binding plates. 53

Reported, not yet replicated

Throughput

Depends on format. Mass-spectrometry CETSA: one person can produce four 16-plex sets per day, but mass spectrometry takes about 24 h per set, so 3 replicates of 52 compounds take about two weeks on one instrument (the authors own setup). Full 2D proteome-wide profiling needs about 10^8 cells per experiment per compound. A vendor page says AstraZeneca screened 500,000 compounds on one protein target with a bead-based high-throughput CETSA; that is a vendor claim with no hit-rate figure. NanoBRET kits are sold in 100, 1,000 and 10,000 assay sizes, covering over 340 kinases. 9107

Reported, not yet replicated

Time per run

Mass-spectrometry CETSA is limited by about 24 h of instrument time per 16-plex set. No other run-time figure was found. 9

Reported, not yet replicated

Engineerable variants

CETSA readout formats

Western blot (one protein, needs a good antibody); bead-based high-throughput CETSA (automatable); mass spectrometry reading the proteome in one run (thermal proteome profiling, compressed CETSA-MS quantifying 7000-8000 proteins, PISA, iTSA); cell-surface CETSA-MS; single-cell CETSA reducing cells from millions to a few hundred. Also intact cells versus lysate, and tissue. 9

Reported, not yet replicated

NanoBRET tracers and wash-out

A different tracer per kinase family (Promega lists eight: K-3, K-4, K-5, K-8, K-9, K-10, K-11, K-12) and profiling panels. A residence-time variant pre-treats cells, washes, then adds saturating tracer; rising BRET means the compound is leaving. In Robers 2015, romidepsin left HDAC1 more slowly than mocetinostat, while SAHA left too fast to measure. 38

Reported, not yet replicated

Use cases

Takeda RIPK1 inhibitors (CETSA)

CETSA isothermal dose-response in HT-29 cells gave an EC50 of 6.5 nM for the lead, compound 22 (a reversible allosteric pyrazolo-pyridinone), against 1,100 nM for Nec-1, a reference compound. Cell EC50 correlated with necroptosis inhibition (R2 = 0.9120) and more weakly with recombinant binding (R2 = 0.7028). In mice dosed 50 mg/kg orally, blood-cell (PBMC) occupancy averaged 63.4% at 1 h and the compound stabilised RIPK1 in spleen and brain. Preclinical, not a clinical candidate. 6

Reported, not yet replicated

Crizotinib in live cells (NanoBRET)

Vasta et al. profiled kinase inhibitor selectivity across 178 kinases in live cells. Cellular ATP reduced engagement, and for crizotinib several putative targets were not engaged at a clinically relevant dose; assays on purified protein can fail to predict cellular engagement. Abstract only was read. 11

Reported, not yet replicated

Teach this

In one sentence

Target engagement assays show that a compound sits on its target inside a cell, which separates "never arrived" from "arrived and did not matter". 1

Reported, not yet replicated

The five-minute version

A compound that inhibits a purified protein may still fail in cells. CETSA tests whether the protein resists heat-induced unfolding when the compound is present; NanoBRET tests whether the compound pushes a glowing tracer off a tagged protein in living cells. Both can miss real binders and neither shows function, so they are paired with a functional assay such as the necroptosis readout in the Takeda RIPK1 work. 163

Reported, not yet replicated

What a listener will ask

If CETSA shows a shift, is the compound working?

It is binding in the cell; stability is not function, and there is no consensus formula converting the shift into true affinity, so pair it with a functional assay (Ishii 2017; CASRAI). 65

Reported, not yet replicated

If there is no shift or no displacement, does the compound not bind?

Not necessarily: a ligand may not shift the melt, or may bind without competing with the tracer. Use a second method (CASRAI; Robers 2015). 53

Reported, not yet replicated

Why do cell results differ from enzyme results?

Cellular ATP reduces engagement of kinase inhibitors, and in the Vasta study several crizotinib targets were not engaged at a clinically relevant dose. 11

Reported, not yet replicated

The common wrong idea

A thermal shift or a tracer displacement means the compound works.

It means the compound binds the target in the cell; whether that changes the protein function needs a separate functional assay.

The one citation to hand over: 3

Still open here

Not yet written: Cost, Before after, Capital, ML role, Data availability.

Linked from

Sources

  1. HowStuffWorks: target engagement in cells (our own earlier write-up, unpublished). Kept in the atlas repository; not published.
  2. Martinez Molina, D., Nordlund, P., et al.. Martinez Molina, Nordlund et al. 2013: cellular thermal shift assay (CETSA), Science. Science (record via Europe PMC) (2013). paper; opened 2026-11-17 all records citing this
  3. Robers, M.B., et al.. Robers et al. 2015: NanoBRET target engagement in living cells, Nature Communications. Nature Communications (full text via Europe PMC) (2015). paper; opened 2026-11-17 all records citing this
  4. The screening cascade: concept breakdown (our own earlier write-up, unpublished). Kept in the atlas repository; not published.
  5. CETSA guide (CASRAI). CASRAI. web, not peer reviewed; opened 2026-11-17 all records citing this
  6. Ishii, T., et al.. Ishii et al. 2017: CETSA for RIPK1 inhibitors in cells and mouse tissue, Scientific Reports. Scientific Reports (full text via Europe PMC) (2017). paper; opened 2026-11-17 all records citing this
  7. NanoBRET TE kinase assay kit, catalogue N2520 (Promega). Promega. vendor-page, not peer reviewed; opened 2026-11-17 all records citing this
  8. Cell-permeable kinase NanoBRET tracers (Promega). Promega. vendor-page, not peer reviewed; opened 2026-11-17 all records citing this
  9. CETSA formats and throughput, Frontiers in Molecular Biosciences 2022 (DOI 10.3389/fmolb.2022.866764). Frontiers in Molecular Biosciences (2022). review; opened 2026-11-17 all records citing this
  10. AstraZeneca enhances drug discovery with Pelago Biosciences CETSA assay (SelectScience). SelectScience (2024). vendor-page, not peer reviewed; opened 2026-11-17 all records citing this
  11. Vasta, J.D., et al.. Vasta et al. 2018: live-cell kinase inhibitor engagement profiling, Cell Chemical Biology. Cell Chemical Biology (record via Europe PMC) (2018). paper; opened 2026-11-17 all records citing this
Where it sits on the map
History of this record (6)

This record has been changed 6 times, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.

Throughput class
medium
Emerging
no
Status
partly written
Last updated
2026-11-19
Record
content/assays/target-engagement-in-cells.json