DNA-Encoded Library Screening in Cells
DNA-encoded libraries can now be screened in cells — on the cell surface, in cell lysate, and inside a living cell — and the format you choose determines which targets are within reach. For two decades, DEL screening meant one thing in practice: a highly purified, correctly folded, catalytically active protein immobilised on a bead or a plate [1]. That requirement has quietly defined the boundary of the technology. Targets that cannot be purified in an active state, membrane proteins that lose their fold outside a lipid bilayer, and complexes that exist only transiently in a cytosol have all sat on the wrong side of that boundary.
Cellular DNA-encoded library selection removes the requirement. Vipergen’s Cellular Binder Trap Enrichment® (cBTE) was the first published technology to screen a multimillion-member DEL inside a living cell, with no highly purified active target protein needed anywhere in the workflow [2]. Other groups have independently developed selections on the surface of live cells and in cell lysate, and together these methods have moved DEL from a biochemical assay into a cellular one.
This article explains what cell-based DEL screening is, how the three formats differ, how cBTE works, which targets each format reaches, and what a cell-based DNA-encoded library service actually delivers. If you are evaluating a hit-finding CRO for a target that has resisted biochemical screening, the practical summary is in the services section.
Why screen a DNA-encoded library in cells?
A conventional DEL selection is an elegant piece of biochemistry: hundreds of millions to billions of DNA-tagged small molecules are incubated with an immobilised target, non-binders are washed away, and the surviving barcodes are amplified and sequenced [1]. The method is fast, cheap per compound, and needs very little protein. Its weakness is not the chemistry or the sequencing — it is the protein.
The target’s native environment
A purified protein on a bead is a simplified object. It has been stripped of its post-translational modifications, separated from its obligate binding partners, pulled out of the membrane or the complex that gives it its physiological conformation, and often tagged and immobilised in an orientation that occludes part of its surface. Screening in a cellular context preserves what purification removes. Methods that operate in cell lysate were explicitly designed around this argument: working in a lysate retains native post-translational modifications and interactions with endogenous binding partners, which both widens the range of accessible targets and raises the biological relevance of the interactions detected [5]. Screening inside an intact cell goes further still, retaining compartmentalisation, local concentration, and the crowded protein environment that a lysate dilutes away.
No purified active protein required
This is the single most consequential practical difference, and for many programmes it is the difference between a screen and no screen at all. Producing a target in a correctly folded, active, purified form can take months, cost a substantial fraction of a hit-finding budget, and sometimes simply fail. Membrane proteins are the obvious case: a natural cellular environment is often necessary to maintain their structure and function, which is precisely why ligand discovery against them has been so difficult [7]. The cell-based formats shift the burden from protein production to cell biology — expressing the target in a cell, or labelling it where it already sits, rather than purifying it.
Cellular target engagement built into the selection
In a biochemical campaign, binding and cellular activity are measured in series: find binders, resynthesise off-DNA, then discover in a cell assay how many of them engage the target in a living system. In an in-cell selection the order is reversed. A compound recovered from an intracellular screen has, by construction, found and bound its target in the presence of the entire cellular proteome. That is a meaningful filter applied before any resynthesis chemistry, and it is why target engagement in cells is one of the more defensible arguments for the approach: the selection itself is the engagement experiment.
Fewer artefacts from immobilisation and tagging
Immobilisation chemistry, affinity tags and bead matrices all introduce opportunities for false positives — compounds that bind the matrix, the tag, or a denatured surface exposed only because the protein was purified. Removing the bead removes a whole class of these artefacts. It does not remove all of them; cell-based formats introduce their own backgrounds, discussed below.
The honest limitation: getting DNA into a cell
DEL compounds are conjugated to oligonucleotides, and large pieces of charged DNA do not cross cell membranes. This, not the selection chemistry, is the central obstacle to intracellular DEL screening, and every published intracellular method is really a solution to it. One approach appends a cyclic cell-penetrating peptide to the library to carry it across the membrane [8]. Vipergen’s solution was different and, in hindsight, disarmingly direct: use a cell big enough to inject. Any honest account of this field should also note that cell-surface and lysate formats sidestep the delivery problem entirely, which is part of why they were developed first.
The three formats of cell-based DEL screening
| Format | Where binding happens | Targets it reaches | Purified protein needed? | Main limitation |
|---|---|---|---|---|
| Cell-surface / whole-cell selection | On the outside of intact cells | Receptors, transporters, surface antigens | No | Surface targets only; high non-specific background from the cell surface |
| Cell lysate screening | In solution, in a crude lysate | Soluble intracellular proteins, modified proteins | No | Lysis dilutes the proteome and loses membrane context and labile complexes |
| In-cell (intracellular) selection | Inside the living cell | Soluble intracellular proteins, PPIs, poorly behaved targets | No | The library has to be delivered across the membrane |
DEL screening on the cell surface and on whole cells
Cell-surface selection was the first cellular DEL format to be demonstrated at scale. The principle is straightforward panning: incubate the library with intact cells displaying the target, wash, and sequence what remains. The complication is that a cell surface is a crowded and heterogeneous place, so the signal from a genuine target binder competes with a large non-specific background.
The field solved this chiefly through counter-selection and through target labelling. The first substantial demonstration extended DEL technology to integral membrane proteins by selecting against a GPCR displayed on cells, identifying novel antagonists of the NK3 tachykinin receptor and establishing that DNA-tagged small molecules could be selected from complex combinatorial libraries against a target in a heterogeneous milieu such as a cell surface [6]. Parallel selections against cells that do not express the target, or against a knockout line, are what separate target-specific enrichment from surface stickiness.
A more refined generation of methods labels the target itself. DNA-programmed affinity labelling attaches a DNA tag specifically to the protein of interest on a live cell, and that tag then directs library selection to the right protein — enabling selections against endogenous membrane proteins without overexpression or other genetic manipulation. A 30.42-million-compound library screened this way against the folate receptor, carbonic anhydrase 12 and EGFR on live cells yielded validated ligands for each [7]. This is the current state of the art for DEL screening on whole cells, and it is the format of choice when the target is a cell-surface protein.
DEL screening in cell lysate
Lysate screening is the halfway house between biochemistry and cell biology, and it was conceived specifically to escape the purification requirement. Interaction determination using unpurified proteins (IDUP) identifies ligand–target pairs from one-pot mixtures of DNA-linked ligands and unpurified protein targets in cell lysates, using the formation of a ternary complex to trigger a DNA extension event that records the pairing. Because it operates in lysate, it preserves post-translational modifications and endogenous binding partners, and it can even register the influence of affinity-modulating adaptor proteins that would simply be absent from a purified-protein experiment [5].
The trade-off is lysis itself. Breaking the cell dilutes the target — typically ten-fold or more relative to its intracellular concentration [8] — destroys compartmentalisation, dissolves the membrane environment that membrane proteins require, and gives labile complexes an opportunity to fall apart. Lysate screening is a real and useful option for soluble intracellular proteins that resist purification. It is not a substitute for screening in an intact cell.
DEL screening in live cells: intracellular selection
Intracellular DEL screening is the most demanding format and the one with the broadest reach, because it is the only one that interrogates a target in the environment where the eventual drug will have to work.
Two published solutions to the DNA-delivery problem exist. The first appends a cyclic cell-penetrating peptide to the DNA-encoded library to carry it into the cytosol, then uses affinity cross-linking to covalently trap ligand–target interactions inside the cell, after which the target is purified and the trapped barcodes recovered. This established that DEL selections against targets within live cells were possible, and the authors framed it explicitly as a strategy for challenging targets that cannot be expressed and purified as active [8]. The second is Vipergen’s cBTE, which solves delivery by microinjection into a cell large enough to accept it [2].
How Cellular Binder Trap Enrichment (cBTE) works
cBTE is built on two earlier Vipergen technologies. YoctoReactor® synthesises the library: the centre of a self-assembled DNA three-way junction, a volume of roughly a yoctolitre (10⁻²⁴ L), serves as the reaction vessel in which building blocks are brought into proximity and coupled, with the assembled DNA encoding the product [3]. Binder Trap Enrichment® screens it: instead of washing non-binders off a solid support, the target and library are diluted and emulsified so that a binding pair is trapped together in a droplet under dissociation-dominated kinetics, and the two DNA strands are then fused into a single readable molecule [4]. Both were designed around fidelity rather than raw library size, and the combination has produced single-digit nanomolar inhibitors directly from a library without hit maturation [9].
cBTE extends this logic into the cell. The published workflow runs as follows [2]:
- Express the target as a prey fusion. mRNA encoding the protein of interest fused to a prey protein is injected into the cell, which translates it. The prey module provides the handle for specific DNA labelling and is what allows the system to distinguish library members bound to the target from those bound to the cell’s own endogenous proteins.
- Inject the library and the bait-DNA. The DEL is microinjected into the same cell, together with a bait-DNA: a DNA molecule bearing a small molecule that binds the prey module. Xenopus laevis oocytes are used because their volume of about 1 µL — roughly 100,000 times that of a typical somatic cell — makes injection straightforward, which solves the problem of delivering DELs across cell membranes for in-cell screening.
- Let binding occur inside the living cell. Library members find their targets under physiological conditions, in a cytosol with its full complement of proteins, metabolites and ions.
- Fuse genotype to phenotype. Where a library member and the bait are bound to the same target protein, the bait-DNA and the library member’s barcode are joined into one DNA strand that records the binding event.
- Lyse, dilute, emulsify, sequence. The cell is broken, the contents are diluted heavily and emulsified, and the trapped pairs are read out by amplification and high-throughput sequencing.
The validation data are what make this more than a concept. A 194-million-member DEL was screened inside live Xenopus laevis oocytes against three pharmaceutically relevant targets — p38α (MAPK14), ACSS2 and DOCK5 — with hits identified and validated for all three, and for p38α the validated hits reached single-digit nanomolar potency [2]. No highly purified active target protein was used.
Two design features deserve emphasis for anyone comparing methods. First, cBTE is a homogeneous selection with no solid support anywhere in the process, which removes bead and immobilisation artefacts rather than merely controlling for them. Second, the system tolerates either N- or C-terminal prey fusions, so the orientation of the target is a matter of convenience rather than a constraint imposed by the screening chemistry.
Which targets suit cell-based DEL screening?
The useful question is not whether a target can be screened in cells, but which format fits it.
Membrane proteins
Membrane proteins are the textbook case for cell-based selection. They account for a large share of approved small-molecule drug targets, yet they are notoriously hard to purify in a functional state because a natural cellular environment is often required to maintain their structure and function [7]. For these targets, cell-surface selection is usually the right format: the protein stays in its bilayer, and modern target-labelling approaches give the selection the specificity that naive panning lacks. Intracellular formats add little here, since the binding site of interest is typically extracellular.
GPCRs and ion channels
GPCRs have been screened with DELs in three distinct ways, and the progression is instructive. Purified-receptor selection works and has produced genuinely novel pharmacology: panning DELs of roughly 190 million compounds against purified human β2-adrenergic receptor yielded a small-molecule negative allosteric modulator, an outcome conventional screening against the orthosteric site had not delivered [10]. Cell-surface selection against receptor-expressing cells came next, with the NK3 tachykinin receptor antagonists [6]. Most recently, selections have moved beyond binding to function: a split-protein complementation strategy applied to opioid receptors on living cells directly selected DEL members that induce specific signalling outcomes, yielding a selective, G-protein-biased κ-opioid receptor agonist with an EC₅₀ of 100 nM [11]. A related live-cell approach has been used to identify agonists for membrane proteins including the insulin receptor, with sub- to low-nanomolar affinity and potency [12].
The practical message for a GPCR programme: if you want binders and have a purifiable construct, biochemical screening remains efficient. If the receptor will not behave outside the membrane, or if you want ligands selected for a functional outcome rather than affinity alone, a cell-based format is the only route.
Protein–protein interactions and intracellular targets
Soluble intracellular proteins, scaffolding proteins and PPI interfaces are where in-cell selection earns its place. These targets often depend on partners, modifications or a particular conformational state that purification disturbs, and their interfaces are typically flat and shallow — the classic “undruggable” profile. The cBTE validation set reflects this: DOCK5 is a guanine nucleotide exchange factor, and ACSS2 a metabolic enzyme, neither of which is a conventional small-molecule target [2].
Targets that resist purification
The broadest category, and the one that most often brings a programme to a CRO, is simply: targets where protein production has failed or has consumed too much time. Large multidomain proteins, proteins requiring specific post-translational modifications, proteins that aggregate on purification, proteins that lose activity on immobilisation. If the protein cannot be made active and pure, biochemical DEL screening has nothing to offer; a cellular format may.
Target-agnostic and phenotypic selections
Two newer directions are worth tracking. Selections can be run against whole cancer cells without a predefined target: a 104.96-million-member DEL selected against a matched pair of breast cancer lines with high and low metastatic character produced cell-specific ligands, and the DNA tag was then used to identify the surface receptor of one of them — making DEL a route to biomarker discovery as well as ligand discovery [13]. Separately, methods are emerging that record a compound’s induced phenotype onto its own barcode in living cells, linking library selection to cellular activity rather than binding [14]. Neither is a routine service offering yet, but both suggest where cell-based DEL is heading.
Cell-based versus biochemical DEL screening
Biochemical DEL screening is not obsolete, and anyone who tells you otherwise is selling something. It remains the faster and cheaper option whenever the target cooperates.
| Parameter | Biochemical DEL screening (incl. BTE) | Cell-based DEL screening (incl. cBTE) |
|---|---|---|
| Target protein | Purified, active, often immobilised [1] | Expressed in or on the cell, or labelled in situ |
| Protein production burden | High; frequently the rate-limiting step | Shifted to cell biology and construct design |
| Target classes | Soluble, well-behaved, purifiable | Adds membrane proteins, PPIs, unpurifiable targets |
| Physiological relevance | Buffer conditions | Native or near-native environment |
| Artefact profile | Bead, tag and immobilisation binders | Non-specific cell binding; delivery-related losses |
| Throughput per target | Higher; well established | Lower; more set-up per target |
A workable decision rule: if the target can easily be produced pure and active and the binding site is not context-dependent, screen biochemically. If protein production is difficult, if the target is a membrane protein, or if the biology that matters only exists inside a cell, screen in cells. In practice the two are complementary, and some programmes run both — a biochemical screen for breadth and an in-cell screen to confirm that the chemotype engages the target in a physiological setting.
It is also worth separating two things that get conflated in marketing material. Homogeneous, solution-phase screening such as Binder Trap Enrichment® removes the solid support but is still biochemical: the target is purified, it is simply not immobilised [4]. Cell-based screening removes the purification requirement itself. These are different claims, and only the second one changes which targets you can screen.
Cell-based DNA-encoded library services at Vipergen
What a cell-based DEL screening campaign includes
A typical engagement runs end to end:
- Target assessment and feasibility. We establish which format fits — in-cell, cell-surface or biochemical — before any screening work is quoted. This conversation is free and usually short.
- Construct design and expression (about four weeks). For cBTE, design of the target–prey fusion and an expression study confirming that it expresses and behaves in the oocyte system. You decide whether to proceed to the screen on the strength of that result.
- Library selection (about four weeks, through to the hit list). Screening of YoctoReactor®-derived DELs under cBTE or BTE conditions, with appropriate controls and counter-selections.
- Sequencing and data analysis. High-throughput sequencing of recovered barcodes with enrichment analysis and chemotype clustering.
- Off-DNA resynthesis. Synthesis of selected hits without the DNA tag.
- Hit validation. Orthogonal confirmation of binding and potency on the resynthesised compounds.
Why library design matters more than library size
The industry has tended to compete on compound count. We have argued the opposite case, and designed for fidelity: in a library spread across billions of barcodes, each compound is sampled so sparsely that genuine binders can vanish into statistical noise [4]. YoctoReactor® libraries are built in the hundreds of millions, drug-like in size and lipophilicity, and synthesised with the DNA junction enforcing correct assembly [3]. The cBTE validation screen used 194 million members and returned single-digit nanomolar validated hits for p38α [2] — which is the relevant benchmark, not the headline library number.
Who this is for
- Biotechs without an in-house DEL platform that need hit matter against a defined target.
- Pharma teams with a stalled programme where protein production or biochemical screening has failed.
- Groups working on membrane proteins, GPCRs or PPIs that have exhausted conventional high-throughput screening.
Intellectual property
Partners retain rights to the hits arising from their campaigns; specific terms are agreed per contract before work begins. We recommend settling this in the first conversation rather than the last.
What you get back
Validated, off-DNA small-molecule hits against your target, with the sequencing data, enrichment analysis and structure–activity information behind them. For a hit-finding CRO engagement, that — not a list of enriched barcodes — is the deliverable.
Frequently asked questions
Yes. Three formats are established: selection on the surface of intact cells, selection in cell lysate, and selection inside a living cell. All three remove the requirement for highly purified active target protein. Vipergen published the first screen of a multimillion-member DEL inside a living cell in 2021 [2].
Cell-based DEL screening is the selection of a DNA-encoded small-molecule library against a target in a cellular context rather than against a purified protein immobilised on a solid support. The target stays in its native environment, keeping its modifications, partners and conformation.
DNA does not cross membranes readily, so this is the central technical problem. Two published solutions exist: appending a cyclic cell-penetrating peptide to the library [8], or microinjecting the library into a cell large enough to accept it. cBTE uses Xenopus laevis oocytes, whose ~1 µL volume is roughly 100,000 times that of a somatic cell [2].
No. That is the defining advantage. The target is expressed in or on the cell, or labelled in situ, and no highly purified active protein is required at any stage [2].
Yes, by several routes: against purified receptor [10], against receptor-expressing cells [6], against endogenous membrane proteins using DNA-tagged targeting on live cells [7], and in functional formats that select for biased agonism rather than binding [11][12]. Which route fits depends on whether your receptor can be purified and on whether you want binders or functional ligands.
Lysate screening keeps post-translational modifications and endogenous binding partners but dilutes the target roughly ten-fold or more and destroys membrane context and compartmentalisation [5][8]. Live-cell screening preserves all of it, at the cost of having to deliver the library into the cell.
Several CROs offer DEL screening, and some offer cell-surface or lysate-based selections. Vipergen is the company that published and commercialised intracellular DEL screening through Cellular Binder Trap Enrichment® [15]. For cell-surface work, academic methods from several groups are well documented and increasingly available through service providers.
About eight weeks from receipt of the target sequence. The first four weeks are an expression study confirming that the target expresses and behaves in the cell system; if you then decide to proceed, the screen itself takes roughly another four weeks and ends in a hit list with chemical structures. No protein isolation or purification is required at any stage [15].
Getting started
If your target has resisted purification, sits in a membrane, or only behaves the way you need it to inside a cell, cell-based DEL screening is worth a conversation. To get a feasibility answer quickly, send us two things: the identity of the target and the sequence then we can express it in cells and check expression levels before entering a campaign.
Contact Vipergen to discuss your target.
References
[1] Satz AL, Brunschweiger A, Flanagan ME, et al. DNA-encoded chemical libraries. Nat Rev Methods Primers (2022), 2(1), 3. https://doi.org/10.1038/s43586-021-00084-5
[2] Petersen LK, Christensen AB, Andersen J, et al. Screening of DNA-encoded small molecule libraries inside a living cell. J Am Chem Soc (2021), 143(7), 2751-2756. https://doi.org/10.1021/jacs.0c09213
[3] Hansen MH, Blakskjær P, Petersen LK, et al. A yoctoliter-scale DNA reactor for small-molecule evolution. J Am Chem Soc (2009), 131(3), 1322-1327. https://doi.org/10.1021/ja808558a
[4] Blakskjær P, Heitner T, Hansen NJV. Fidelity by design: yoctoReactor and binder trap enrichment for small-molecule DNA-encoded libraries and drug discovery. Curr Opin Chem Biol (2015), 26, 62-71. https://doi.org/10.1016/j.cbpa.2015.02.003
[5] McGregor LM, Jain T, Liu DR. Identification of ligand-target pairs from combined libraries of small molecules and unpurified protein targets in cell lysates. J Am Chem Soc (2014), 136(8), 3264-3270. https://doi.org/10.1021/ja412934t
[6] Wu Z, Graybill TL, Zeng X, et al. Cell-based selection expands the utility of DNA-encoded small-molecule library technology to cell surface drug targets: identification of novel antagonists of the NK3 tachykinin receptor. ACS Comb Sci (2015), 17(12), 722-731. https://doi.org/10.1021/acscombsci.5b00124
[7] Huang Y, Meng L, Nie Q, et al. Selection of DNA-encoded chemical libraries against endogenous membrane proteins on live cells. Nat Chem (2021), 13(1), 77-88. https://doi.org/10.1038/s41557-020-00605-x
[8] Cai B, Kim D, Akhand S, et al. Selection of DNA-encoded libraries to protein targets within and on living cells. J Am Chem Soc (2019), 141(43), 17057-17061. https://doi.org/10.1021/jacs.9b08085
[9] Petersen LK, Blakskjær P, Chaikuad A, et al. Novel p38α MAP kinase inhibitors identified from yoctoReactor DNA-encoded small molecule library. MedChemComm (2016), 7(7), 1332-1339. https://doi.org/10.1039/C6MD00241B
[10] Ahn S, Kahsai AW, Pani B, et al. Allosteric “beta-blocker” isolated from a DNA-encoded small molecule library. Proc Natl Acad Sci U S A (2017), 114(7), 1708-1713. https://doi.org/10.1073/pnas.1620645114
[11] Cai B, El Daibani A, Bai Y, Che T, Krusemark CJ. Direct selection of DNA-encoded libraries for biased agonists of GPCRs on live cells. JACS Au (2023), 3(4), 1076-1088. https://doi.org/10.1021/jacsau.2c00674
[12] Huang Y, Hou R, Lam FS, et al. Agonist discovery for membrane proteins on live cells by using DNA-encoded libraries. J Am Chem Soc (2024), 146(35), 24638-24653. https://doi.org/10.1021/jacs.4c08624
[13] Gui Y, Hou R, Huang Y, et al. Discovering cell-targeting ligands and cell-surface receptors by selection of DNA-encoded chemical libraries against cancer cells without predefined targets. Angew Chem Int Ed (2025), 64, e202421172. https://doi.org/10.1002/anie.202421172
[14] Sander PN, Gillen Miller JT, Lairson LL. Induced cell phenotype activity recording of DNA-tagged ligands. RSC Chem Biol (2025), 6(2), 273-280. https://doi.org/10.1039/d4cb00137k
[15] Vipergen. Discovery of inhibitors, binders and modulators. Vipergen A/S. Accessed November 2026. https://www.vipergen.com/discovery-of-inhibitors-binders-and-modulators/