Programmable DARPin-based receptors for the detection of thrombotic markers

Cell-based therapies lack a broad repertoire of cellular receptors to sense clinically relevant disease markers, limiting programmability and therapeutic control. Existing engineered receptors—such as CARs, chimeric receptors fusing native extracellular and intracellular domains, and synthetic dimerization-dependent receptors—require extensive ligand-specific optimization and are constrained by the ligand space of available binders. Non-antibody scaffolds (e.g., monobodies, aReps, nanobodies, DARPins) can be engineered to recognize diverse targets, and selection technologies (phage, yeast, ribosome display) enable high-throughput discovery of binders with tailored properties. Building on the GEMS platform, which links extracellular binding to EpoR-mediated STAT3 signaling, the authors sought to expand receptor specificity through integration with high-throughput DARPin selection, creating AMBER receptors. The research question is whether an integrated DARPin-to-receptor pipeline can generate programmable, bispecific receptors for soluble ligands—specifically thrombotic markers such as fibrin degradation products—with sufficient sensitivity, specificity, and modularity for diagnostic and therapeutic applications.

Prior receptor engineering in mammalian synthetic biology has largely relied on native receptor–ligand pairs and antibody-derived scFvs, limiting accessible epitope space and requiring per-ligand tuning. Alternative binding scaffolds like fibronectin-III monobodies, aRep HEAT-like repeats, camelid nanobodies, and DARPins have been successfully diversified to bind non-native targets, broadening options. Selection platforms—phage display and yeast display—facilitate discovery of disulfide-containing proteins, whereas ribosome display supports large libraries and affinity maturation via error-prone PCR and is particularly well-suited to well-folding scaffolds like DARPins (14–17 kDa). The GEMS platform previously demonstrated modular receptor construction with sandwich binding requirements but remained dependent on available binders. Clinically, D-dimer/FDPs are established biomarkers to rule out pathological coagulation, motivating their selection as targets to validate a generalized receptor pipeline. Structural constraints from EpoR activation and epitope geometries imply that sandwich-pair binders and spatial arrangement critically determine receptor function.

Overall pipeline and receptor design: The AMBER system couples extracellular binders to an EpoR-based scaffold that activates STAT3-driven transcription of reporters (SEAP, NLuc) or therapeutic outputs (tenecteplase TNK, hirudin). The pipeline comprises: (1) selection of DARPins against target antigens via ribosome display; (2) sequence analysis and classification; (3) receptor assembly and screening in homodimeric, heterodimeric, and tandem configurations; and (4) application in vitro and in vivo. Spatial constraints were estimated from the crystal structure of activated EpoR (PDB 1EER), targeting an inter-binder C-terminus spacing under ~8 nm.

DARPin selection by ribosome display: Cross-linked fibrin degradation products (xFDPs) enriched in D-dimers were biotinylated and immobilized on streptavidin/neutravidin beads. A synthetic N3C-DARPin library (theoretical diversity ~3.8×10^12; three randomized internal repeats with variable cap randomization) was used. Four selection rounds with decreasing xFDP concentrations, one off-rate selection (non-biotinylated xFDPs), and a recovery round were performed on a semi-automated platform. After enrichment, clones were expressed in E. coli, and 380 single clones were screened by HTRF binding (donor: streptavidin-Tb; acceptor: anti-FLAG-d2) versus xFDPs with specificity controls (no target, human serum albumin). Thirty-two hits were sequenced; 19 unique DARPins were advanced. Analytical SEC identified 15/19 as monomers.

Receptor configurations and cloning: Three architectures were built on the GEMS scaffold: (i) homodimeric (identical binder on both chains); (ii) heterodimeric (pairwise different binders on separate chains); and (iii) tandem (two binders fused in head-to-tail orientation on a single chain to reduce payload for vectorization). Receptors were fused using appropriate linkers; benchmarking used GFP binders (clamp-DARPin, single DARPin, nanobody) with GFP-GFP analytes bearing long (GGGGS)5, short (GGGGS)2, or no linkers. An MBP sensor used a published anti-MBP scFv (PDB 7JTR chain B) combined with a DARPin (PDB 1SVX) connected by a rigid (EAAAK)5 linker. Constructs were expressed under constitutive CMV promoters.

Cell culture and screening assays: HEK-293T cells were transiently transfected with receptor plasmids (2 ng each chain), constitutive STAT3 (pLS15), and a STAT3-responsive reporter (PLS13, RE2-PhCMVmin-SEAP). Insensitivity to endogenous EPO was confirmed. Toxicity and viability of xFDPs and human plasma were evaluated (constitutive SEAP and resazurin assays). A comprehensive pairwise screen of 210 receptor combinations (including 20 homodimers) was conducted, testing induction by purified xFDPs and human citrate-treated plasma. Positive responses were categorized as weak (2<fold<5) or strong (>5).

Biophysical characterization: Surface plasmon resonance (SPR) on a Biacore T200 with CMD200M chip coated with xFDPs (205 RU) measured binder kinetics using a heterogeneous ligand model, yielding two KDs per analyte. Electrochemical impedance spectroscopy (EIS) measured charge transfer resistance (Rct) changes upon ligand binding in live cells seeded on modified Au electrodes in the presence of [Fe(CN)6]3−/[Fe(CN)6]4−.

Specificity mapping and coagulation pathway dissection: Minimal systems using purified fibrinogen ± thrombin tested receptor activation prerequisites. Plasma fractionation by SEC and mass spectrometry, purified fibrinogen fragments D/E, and plasmin digestion experiments assessed the identity of inducing species. Pharmacologic modulators—heparin, hirudin (thrombin inhibitors), tranexamic acid (plasmin inhibitor), and tridegin (FXIIIa inhibitor)—were used to probe dependence on fibrin cross-linking and fibrinolysis.

Stable designer cell lines and outputs: Sleeping Beauty transposon integrations created stable HEK-293T lines with STAT3-responsive reporters and AMBER receptors. For therapeutic coupling, a secreted NLuc fused via a furin-cleavable site (RARYKR) to TNK (Nluc-mFc–TNK-mFc) ensured equimolar secretion and allowed NLuc activity as a proxy for TNK. Separate lines expressed secreted hirudin-HM2 (Nluc–p2a–hirudin) under STAT3 control. Reporter assays measured SEAP or NLuc in supernatants.

Clinical samples and in vivo models: Human citrate plasma samples (ethical approvals obtained) were tested at defined v/v percentages with or without heparin supplementation (20 U ml−1) to assess detection of elevated D-dimer and hypercoagulability (e.g., DIC). For in vivo validation, hydrodynamic tail-vein injection transduced mouse liver with AMBERB4/scFv and a STAT3-NLuc reporter. Systemic coagulation was induced by intravenous thrombin (0–1000 U kg−1), and blood D-dimer and NLuc were quantified. An acetaminophen (APAP) acute liver injury model (160 mg kg−1 IP) assessed local coagulation activation; AST/ALT and NLuc were measured at 8 h and 12 h.

Benchmarking with model ligands:

• GFP-targeting AMBERs showed nanomolar sensitivities and sensitivity to epitope spacing. EC50 values (95% CI): clamp-DARPin homodimer: 1.14 nM (long linker), 0.77 nM (short), 0.84 nM (no linker); single DARPin: 2.27, 2.00, 1.83 nM; nanobody: 0.49, 0.34, 0.35 nM. An MBP AMBER (scFv + DARPin) achieved EC50 ≈ 0.85 nM.

DARPin selection and receptor screening against fibrin targets:

• HTRF screen of 380 DARPin clones yielded 160 binders (>1.15-fold) and 118 strong binders (>1.45-fold). Nineteen unique DARPins advanced; 15 were monomeric by SEC.
• Comprehensive AMBER screen (210 receptor pairs): With xFDPs, 4 weak (1.9%) and 3 strong (1.4%) activations; all strong responders were heterodimers combining a DARPin with an anti-D-dimer scFv. With human plasma, 37 weak (17.6%) and 49 strong (23%) responses; 77% of active combinations included DARPins A6, A7, B4, or F6.

Receptor performance and sensitivity:

• Homodimeric DARPin A6/A7/B4 receptors showed minimal xFDP dose dependence, but displayed ON-type switching with up to ~110-fold induction in human plasma, suggesting activation by thrombin-processed fibrinogen/FDP species present in plasma but underrepresented in the xFDP prep.
• Heterodimeric receptors (DARPin + scFv) exhibited xFDP dose responses with EC50 below the clinical D-dimer threshold (0.5 µg ml−1): A6+scFv 0.15 µg ml−1, A7+scFv 0.19 µg ml−1, B4+scFv 0.12 µg ml−1. Plasma induction profiles were also strong.
• Tandem receptors retained functionality. DARPin-only tandems detected xFDPs with EC50s of 4.3 µg ml−1 (G1–A7) and 1.2 µg ml−1 (G1–B4). Tandems combining scFv with DARPins matched heterodimers: scFv–A7 EC50 0.31 µg ml−1 (vs 0.19), scFv–B4 0.25 µg ml−1 (vs 0.12). Plasma increased signals across tandems, including G1–G1.

Biophysics and binding:

• SPR KDs (heterogeneous model) versus xFDPs: scFv 244/169 nM; DARPins A6 14/37 nM, A7 91/44 nM, B4 23/59 nM, G1 602/1030 nM. EIS corroborated ligand binding and avidity effects: at 1 µg ml−1 xFDPs, homodimeric Rct (kΩ): B4 5.58, A6 3.59, A7 1.48, G1 0.362, scFv 0.62; heterodimeric B4+scFv 21.88 kΩ.

Specificities within coagulation pathway:

• Activation required thrombin-mediated fibrinogen processing in the presence of FCS-derived thrombin; heparin and hirudin blocked activation. Tridegin (FXIIIa inhibitor) and tranexamic acid modulated responses, indicating dependence on cross-linking and plasmin activity. Mapping suggests: AMBERB4 recognizes fragment E or complexes (D–E, DD–E); scFv prefers cross-linked D-dimers; tandem scFv–B4 favors larger cross-linked FDPs; G1–B4 favors non-cross-linked fragments.

Clinical validation and therapeutic coupling:

• Patient plasma (5% v/v): AMBERB4 detected elevated D-dimer samples (P<0.0001 vs negative control), and AMBERB4/scFv discriminated DIC samples from controls based on D-dimer with heparin (P=0.0049) and hypercoagulability without heparin (P<0.0001).
• Designer cells secreted TNK upon xFDP induction; NLuc reporter correlated with TNK activity; detection down to 0.1 µg ml−1 xFDPs; onset ~6 h. Hirudin-producing cells reduced downstream coagulation reporter signal 3.3-fold and prolonged thrombin time by ~10% (P=0.0016).

In vivo proof-of-concept:

• Hydrodynamic liver delivery of AMBERB4/scFv in mice followed by IV thrombin (500–1000 U kg−1) increased blood D-dimer and NLuc in a dose-dependent manner (significant vs 0 U kg−1). In APAP-induced liver injury, AST/ALT and NLuc were elevated at 8–12 h, indicating in vivo activation of the sensor.

The integrated DARPin-to-AMBER pipeline successfully produces programmable, bispecific receptors for soluble targets with tunable specificity and sensitivity. By expanding the binder repertoire beyond scFvs to include DARPins selected via ribosome display, the platform accesses a broader epitope landscape and overcomes prior dependence on pre-existing antibody binders. The study demonstrates detection of clinically relevant thrombotic markers at or below pathologic thresholds and distinguishes different coagulation states (e.g., cross-linked versus non–cross-linked FDPs) by combining binder types or architectures (heterodimeric versus tandem). Coupling receptor activation to therapeutic protein secretion (TNK or hirudin) establishes closed-loop, potentially autologous, anti-thrombotic responses, while dual output (reporter + therapeutic) enables real-time monitoring of efficacy. Biophysical analyses (SPR/EIS) explain functional performance via affinity and avidity, highlighting non-linear gains in heterodimeric and tandem formats. The findings validate that spatial constraints and epitope geometry determine receptor function; sandwich-pair complementarity between DARPin concave interfaces and scFv loop-based paratopes enables synergistic binding. Clinically, the receptors detect elevated D-dimer in patient samples and respond in mouse models of systemic and local thrombosis, underscoring translational potential for diagnostics and cell-based therapies targeting thrombotic disorders.

This work introduces AMBER, a modular receptor platform integrating high-throughput DARPin selection with a GEMS-based signaling scaffold to create programmable sensors for soluble ligands. Using fibrin degradation products as a proof-of-concept target, the pipeline yielded functional homodimeric, heterodimeric, and compact tandem receptors with sensitivities below clinical thresholds and distinct specificities across the coagulation cascade. Designer cells equipped with AMBERs both report and counteract coagulation by secreting TNK or hirudin, and the system operates in patient samples and mouse models. The approach broadens access to custom receptors for targets lacking natural sensors and is adaptable to other soluble biomarkers, cytokines, and potentially cell-surface antigens. Future research should expand binder libraries to additional scaffolds, refine epitope-pairing rules to increase hit rates, optimize intracellular signaling modules for alternative outputs and cell types, and develop vectorized delivery (e.g., rAAV) for in vivo therapeutic applications and real-time thrombosis monitoring.

Only a subset of selected DARPins yielded functional receptors, reflecting stringent spatial and affinity requirements for sandwich-pair activation. Success rates depend on binder complementarity and epitope geometry; monomeric targets are expected to be more challenging for homodimeric designs. Some homodimeric receptors exhibited OFF-type behavior upon xFDP binding, indicating possible inactive dimer stabilization. The xFDP preparation underrepresents certain native plasma species, leading to differences between purified-protein and plasma responses. Species specificity varied (e.g., some receptors responded to human but not mouse plasma), potentially complicating in vivo translation. Although tandem receptors reduce genetic payload, the combinatorial configuration space is large, necessitating substantial screening. Finally, therapeutic efficacy and safety were demonstrated in vitro and in mouse models; comprehensive preclinical evaluation, immunogenicity assessment, and long-term performance remain to be established.