To verify the orthogonality of SynKinase-mediated phosphorylation, we performed Western blot analysis using anti-pY antibodies specific for phosphorylated tyrosine residues within the ITAM motifs. Substrate presence in each sample was confirmed by Myc-tag fusion and subsequent anti-Myc staining (Fig. 1).

---

Figure 1: Schematic of SynKinase and SynSubstrate constructs used for phosphorylation validation. The cytosolic SynKinase consists of a truncated ABL kinase fused to a bZipEE dimerization domain. The SynSubstrate contains three tandem CD3ζ ITAM motifs (3x CD3ζ) fused to the complementary bZipRR domain and a C-terminal triple Myc tag for detection.

Western blot analysis confirmed that SynSubstrate phosphorylation occurs exclusively in the presence of SynKinase. The anti-pY antibody detected no phosphorylation signal in the SynSubstrate-only condition (Fig. 2, lane 1, top panel), whereas a strong signal appeared upon SynSubstrate + SynKinase cotransfection (lane 2, top panel). Myc-tag staining verified substrate expression in both conditions, confirming that the absence of phosphorylation in the control was not due to missing substrate. The α-Myc signal cannot be quantitatively compared between samples because residual anti-pY binding persisted after antibody reuse without stripping. Nevertheless, equal SynSubstrate plasmid amounts were transfected in all conditions, supporting the conclusion that SynKinase specifically and orthogonally phosphorylates SynSubstrate.

---

Figure 2: Western blot verification of SynSubstrate phosphorylation by SynKinase. HEK293T cells were transfected with equal amounts of SynSubstrate either alone or together with SynKinase. Phosphorylation was detected using a phospho-tyrosine-specific (anti-pY) antibody. A phosphorylation signal was observed exclusively in the presence of SynKinase, confirming kinase-dependent modification of the SynSubstrate.

To develop and test the basic circuit functionality, we established a split-fluorescent protein system enabling us to directly observe dimerization between CD3ζ and SH2. This strategy allowed us to investigate interaction dynamics before further engineering a dimerization dependent output. SplitFAST is a split version of RspA-FAST, a protein that fluorescences in the presence of the small molecule HMBR (Rakotoarison et al., 2024)(Fig. 3). Exhibiting low self-complementation and a high dynamic range it allows for effective studying of protein protein interactions.

---

Figure 3: Schematic of the CD3ζ-SH2 dimerization dependent splitFAST reporter read-out Upon coiled-coil interaction between RspA(N)-CD3ζ-bZipRR and the SynKinase, tyrosine residues within the CD3ζ are phosphorylated. This increases affinity to the SH2, allowing for phosphorylation-specific dimerization. Dimerization between RspA(N)-CD3ζ-bZipRR and RspA(C)-SH2 brings the splitFAST halves into proximity giving a fluorescent readout upon HMBR addition.

---

Figure 4: Validating phosphorylation dependent dimerization using splitFAST. HEK293T cells were transfected 24 hours after seeding. Y-axis description corresponds to the ng amounts of SynKinase, 100 ng of both RspA(N)-CD3𝜁-bZipRR and RspA(C)-SH2 were co-transfected. Dimerization dependent SplitFAST fluorescence was quantified by flow cytometry 48 hours after transfection, employing excitation lasers at 488 nm and 561 nm. Emission was recorded at 525 nm for splitFAST (Ligand: TF_Lime) and at 610 nm for constitutively expressed mCherry, serving as a transfection control. Data is depicted as the geometric mean of fluorescence +/- geometric SEM with ~30000 single cells

Reversibility of phosphorylation-dependent dimerization is crucial for dynamic signal processing and logic gate computation. To test this, we assessed whether SynPhosphatase could access and dephosphorylate CD3ζ, thereby disrupting CD3ζ-SH2 binding. Increasing SynPhosphatase expression progressively reduced splitFAST fluorescence, indicating dissociation (Fig. 5). The dose-dependent decline in signal, reaching background levels at high SynPhosphatase concentrations, confirms effective dephosphorylation and circuit reversibility.

---

Figure 5: Validating dephosphorylation dependent dissociation using splitFAST. HEK293T cells were transfected 24 hours after seeding. Y-axis description corresponds to the ng amounts of SynPhosphatase, 100 ng of RspA(N)-CD3𝜁-bZipRR, RspA(C)-SH2 and SynKinase were co-transfected. Dimerization dependent SplitFAST fluorescence was quantified by flow cytometry 48 h post transfection, employing excitation lasers at 488 nm and 561 nm. Emission was recorded at 525 nm for splitFAST (Ligand: TF_Lime) and at 610 nm for constitutively expressed mCherry, serving as a transfection control. Data is depicted as the geometric mean of fluorescence +/- geometric SEM with ~30000 single cells

To build on the previously validated phosphorylation and dimerization events, a transcriptional effector circuit was implemented that translates enzymatic activity into a measurable output. Therefore, Gal4DBD is fused to the synthetic substrate, while VP64 is attached to the SH2 domain. SynKinase, recruited via complementary coiled-coils (bZipEE/bZipRR), phosphorylates CD3ζ and enables SH2-VP64 binding. This interaction reconstitutes a functional Gal4DBD-VP64 transcription factor that activates NanoLuciferase expression. Conversely, SynPhosphatase dephosphorylates CD3ζ, preventing SH2-VP64 recruitment and thereby reporter activation (Fig. 7).

---

Figure 7: Schematic representation of the phosphorylation circuit and Nanoluciferase reporter assay. SynKinase (ABL-bZipEE) phosphorylates the SynSubstrate (Gal4DBD-CD3ζ-bZipRR) in a proximity-dependent manner, while SynPhosphatase (PTPN1-bZipEE) counteracts this modification through dephosphorylation. Phosphorylated SynSubstrate recruits the SH2-VP64 binder, which, together with Gal4DBD, activates transcription from a minimal promoter and induces NanoLuciferase expression. For clarity, CD3ζ (three ITAM repeats) and SH2 (two domains) are illustrated as single units.

---

Figure 8: Dose-response of SynKinase. HEK293T cells were transfected with SynSubstrate (5 ng) and SH2-VP64 (25 ng) in a 1:5 ratio together with increasing amounts of cytosolic SynKinase. The Dual-Luciferase assay was performed 48 hours post transfection, with NanoLuciferase normalized to Firefly. Data are presented as the mean ± SEM with n = 3 technical replicates. Statistical significance was determined with ordinary one-way ANOVA and Tukey's multiple comparisons test (single pooled variance).

---

Figure 9: Titration of SynPhosphatase. HEK293T cells were transfected with 20 ng SynKinase, 25 ng SH2-VP64, and 5 ng Gal4DBD-CD3ζ-bZipRR (SynSubstrate), while SynPhosphatase (PTPN1-bZipEE) was titrated from 1 to 20 ng as indicated. Reporter activity was measured 48 h post transfection using a Dual Luciferase assay, with NanoLuciferase normalized to constitutively expressed Firefly luciferase. Data are shown as mean ± SEM (n = 3 technical replicates). Statistical significance was determined with ordinary one-way ANOVA with Tukey's multiple comparison test (single pooled variance).

Together, the kinase and phosphatase titration experiments demonstrate that the phosphorylation circuit enables precise, reversible, and tunable control of transcriptional activation. Reporter expression increased proportionally with SynKinase input until phosphorylation sites on the SynSubstrate became saturated, while rising SynPhosphatase levels progressively reduced the signal back to baseline. These findings confirm that circuit output directly reflects dynamic balance between kinase- and phosphatase-mediated modification states, establishing a controllable molecular switch that links enzymatic activity to gene expression.

To engineer a phosphorylation-inducible TEV protease, we developed two different designs. In the first one we used a split TEV system where the N-terminal half was linked to an SH2-domain, while the C-terminal half was linked to a SynSubstrate and bZipRR (Fig. 10A). In a second iteration, the two constructs were joined into one by linking them through a semi-rigid linker between the SH2 and bZipRR domains (Fig. 10B). In both versions, the association of the bZip coiled-coils causes phosphorylation of the SynSubstrate so that the SH2 domain binds to it, resulting in the reconstitution of the TEV protease. The activated protease in turn causes the secretion of a luminescent NanoLuc luciferase (NanoLuc) reporter, which could be measured in the supernatant.

---

Figure 10: Different PhosphoTEV designs. Reconstitution of split TEV halves depends on the phosphorylation-inducible binding of SH2 to CD3ζ. (A) In the split PhosphoTEV design, the interacting domains are located on different constructs. (B) In the linked PhosphoTEV, SH2 and bZipRR are connected through a variable hinge.

---

Figure 11: SynKinase activates PhosphoTEV and induces secretion. HEK293T cells were co-transfected with split and linked PhosphoTEV variants without (orange) and with (purple) SynKinase, and the NanLuc-RELEASE plasmid. Secreted NanoLuc in supernatant was measured 48h post-transfection. Positive and negative controls (yellow) with a full TEV and no TEV are shown for comparison. Data is depicted as the mean +/- SEM with n = 3 technical replicates. Statistical significance was calculated with an ordinary One-way ANOVA with Tukey's multiple comparisons test. **P<0.01, ****P<0.0001; ns, not significant.

A key advantage of phosphorylation-based signaling circuits lies in the reversibility dependent on kinase and phosphatase activity. To assess whether this behaviour applies to the RELEASE system, different amounts of PhosphoTEV were co-transfected with SynKinase alone or in combination with SynPhosphatase. After 48 hours, NanoLuc luminescence in the supernatant was measured (Fig. 12). As shown before, the background of PhosphoTEV without kinase or phosphatase did not differ significantly from the RELEASE background without any TEV. With only SynKinase, there was a 2.6-fold induction in the luminescence signal. This induction could be quenched significantly with the addition of SynPhosphatase, reducing the increase by about 50 %, showing that the activation of PhosphoTEV is reversible and sensitive to SynPhosphatase. With the successful (de-)activation of PhosphoTEV using SynKinase and -Phosphatase, we could establish the fully post-translational RELEASE response module, able to be integrated into more sophisticated circuits.

---

Figure 12: SynPhosphatase deactivates PhosphoTEV and reduces secretion. HEK293T cells were co-transfected with the linked PhosphoTEV and either no enzyme (yellow), only SynKinase (orange), or in combination with SynPhosphatase (purple), and the NanoLuc-RELEASE plasmid. Secreted NanoLuc in supernatant was measured 48h post-transfection. Data is depicted as the mean +/- SEM with n = 3 technical replicates. Statistical significance was calculated with an ordinary Two-way ANOVA with Tukey's multiple comparison's test. ***P<0.001, ****P<0.0001; Negative and positive controls were excluded from the test and only shown for reference.

To effectively drive gene expression in an inducible, ligand-specific manner, a receptor should be host-orthogonal, exhibit low background, and be readily inducible. In this context, we chose the 𝜅-Opioid receptor DREADD (KORD), an engineered GPCR inducible by the small-molecule Salvinorin B (SalB), as a receptor scaffold for our phosphorylation circuit (Fig. 13). By fusing its intracellular C-terminal tail with our SynKinase, a truncated ABL kinase able to induce specific, proximity-based phosphorylation, we created a novel receptor construct: KORD-ABL. Together with an engineered version of 𝛽-Arrestin2, an endogenous regulator of GPCR activity, it builds a system that is able to specifically recruit and phosphorylate our SynSubstrate upon receptor activation.

SalB binding triggers a conformational change that enables GRK-mediated phosphorylation of residues in the receptor's C-terminal tail (Fig. 13). This recruits β-Arrestin2-bZipEE that can bind Gal4DBD-Cd3ζ-bZipRR via a coiled-coil interaction. The resulting positioning of the CD3ζ near the ABL kinase fused to the GPCR leads to its phosphorylation. The phosphorylated Gal4DBD-CD3ζ-bZipRR subsequently dimerizes with SH2-VP64, forming a transcriptional activator that binds the 5xUAS motif and drives transgene expression.

---

Figure 13: Schematic overview of KORD-ABL coupled to the phosphorylation circuit. In the absence of SalB, 𝛽-Arrestin2-bZipEE cannot bind KORD-ABL. Receptor activation through SalB presence leads to a conformational change, allowing GRK binding and subsequent phosphorylation of the C-terminal tail of KORD. This allows for the binding of 𝛽-Arrestin2-bZipEE. Connection to the intracellular PHOENICS circuit is facilitated by the fused bZipEE domain on 𝛽-Arrestin2 and its spatial organization in the proximity of the ABL kinase upon KORD binding. Bound 𝛽-Arrestin2-bZipEE recruits Gal4DBD-CD3𝜁-bZipRR, which is subsequently phosphorylated by ABL. Dissociation from the KORD due to the transient coiled-coil interactions allows dimerization between the phosphorylated CD3𝜁 and SH2, reconstituting the split transcription factor Gal4DBD-VP64. Shuttling to the nucleus then facilitates transgene expression under control of the 5xUAS-minCMV promoter.

To evaluate the synthetic KORD-ABL receptor, we investigated its ability to drive NanoLuciferase expression upon induction by its small molecule ligand SalB. We co-transfected HEK293T cells with KORD-ABL, 𝛽-Arrestin2-bZipEE, as well as our phosphorylation circuit and constitutively expressed Firefly luciferase to normalize to cell viability. Cells were induced with 500 nM SalB 24 hours after transfection. Nano- and Firefly Luciferase activity was determined 48 hours after transfection using the Nano-Glo Dual-Luciferase Reporter Assay System. As controls, we first used the entire phosphorylation circuit without KORD-ABL, as well as separately without 𝛽-Arrestin2-bZipEE. In three biological replicates, induction of KORD-ABL significantly increases NanoLuciferase expression by 28-fold compared to its uninduced condition (p < 0.0001), exhibiting a switch-like behavior to drive gene expression (Fig. 14). Moreover, KORD-ABL does not exhibit significant leakiness or background noise compared to control conditions without KORD-ABL or 𝛽-Arrestin2-bZipEE (p > 0.99). This demonstrates the functionality of a novel receptor architecture that can drive transgene expression in a switch-like manner, without exhibiting significant background or interference in native signaling pathways. This data was further corroborated by measuring mCherry expression upon KORD-ABL induction on a single-cell basis using flow cytometry.

---

Figure 14: KORD-ABL represents a novel receptor platform that senses and responds to extracellular cues in a switch-like manner. HEK293T cells were transfected 24 hours after seeding. NanoLuc and Firefly Luciferase expression was measured 48 hours after transfection using the Dual Luciferase assay. Induction occurred 24 hours transfection using DMSO or SalB [500 nM]. Negative 1 corresponds to KORD-ABL and the full phosphorylation-dependent expression circuit without 𝛽-Arrestin2-bZipEE. Negative 2 corresponds to 𝛽-Arrestin2-bZipEE and the full phosphorylation-dependent expression circuit without KORD-ABL. 30/5 ng refers to the condition containing the full receptor construct, with 30 ng of KORD-ABL, 5 ng of 𝛽-Arrestin2-bZipEE, and the full phosphorylation-dependent expression circuit. Data is depicted as the mean +/- SEM with n = 3 biological replicates. Two-way ANOVA with Sidak's multiple comparison test. ****P < 0.0001; NS, not significant.

A critical variable in synthetic systems is the ability to respond to extracellular cues in a fast and steady manner. To evaluate this property, we characterized the response kinetics of our combined sensing, processing, and output system. For this purpose, SalB induction was performed in cells co-transfected with KORD-ABL, β-Arrestin2-bZipEE, the phosphorylation circuit, and a normalization plasmid. Induction occurred at different time points before measurement to capture dynamic changes in system activity. NanoLuciferase expression increases significantly by 10-fold compared to its uninduced condition 5 hours after induction (p = 0.0049). Expression also correlates linearly with time (R² = 0.97), reaching a total fold change of 151-fold compared to its uninduced condition measured 32 h after induction (Fig. 15). This highlights how KORD-ABL, in conjunction with our phosphorylation circuit, drives rapid gene expression after sensing a small molecule and processing the signal at the post-translational level.

---

Figure 15: KORD-ABL drives fast gene expression upon ligand sensing . Line plot showing an increase in expression over time. HEK293T cells were transfected 24 hours after seeding. NanoLuc and Firefly Luciferase expression was measured 48 hours after transfection using the Dual Luciferase assay. Induction occurred at the indicated time points before measurement with either DMSO or SalB [2 µM]. Each condition was transfected with 50 ng of KORD-ABL and 5 ng of 𝛽-Arrestin2-bZipEE, Overlaid with a linear regression curve (R² = 0.97). Data is depicted as the mean +/- SEM with n = 3 technical replicates. Due to technical errors during sample induction, one value is missing for all induced conditions. Two-way ANOVA with Sidak's multiple comparison test. **P < 0.01; ****P < 0.0001.

To expand the existing receptor repertoire to go beyond gene-expression mediated repression, we set out to establish a receptor construct that is able to directly interact with, and quench our phosphorylation circuit on a post-translational level. Fusing our SynPhosphatase, a truncated version of the PTPN1 phosphatase, to the C-terminal tail of the KORD, we aimed to enable ligand-inducible dephosphorylation and subsequent quenching of expression (Fig. 16)

---

Figure 16: Schematic overview of KORD-PTPN1 coupled to the phosphorylation circuit. In the absence of SalB, 𝛽-Arrestin2-bZipEE is unable to bind KORD-PTPN1. Receptor activation through SalB presence leads to a conformational change, allowing GRK binding and subsequent phosphorylation of the C-terminal tail of KORD. This allows for the binding of β-Arrestin2-bZipEE. Connection to the intracellular PHOENICS circuit is facilitated by the fused bZipEE domain on β-Arrestin2 and its spatial organization in the proximity of the PTPN1 phosphatase upon KORD binding. Bound 𝛽-Arrestin2-bZipEE recruits Gal4DBD-CD3ζ-bZipRR, which is subsequently dephosphorylated by PTPN1. Dephosphorylation of CD3ζ leads to dissociation of Gal4DBD-CD3ζ-bZipRR and Sh2-VP64, thereby splitting the transcription factor and quenching gene expression.

To evaluate the ability of KORD-PTPN1 to interact with and facilitate dephosphorylation of CD3𝜁, as well as to interrogate full circuit compatibility, we co-transfected KORD-ABL, KORD-PTPN1, 𝛽-Arrestin2-bZipEE, and our phosphorylation circuit with a 5x UAS NanoLuciferase expression cassette as well as a normalization plasmid. Cells were induced with 500 nM SalB 24 hours after transfection. Expressed NanoLuciferase was measured 48 hours after transfection using a plate reader. Titrating increasing amounts of KORD-PTPN1 against KORD-ABL led to a significant decrease in phosphorylation-mediated expression, indicating a dose-dependent ability to dephosphorylate and quench expression starting at a 2:3 ratio (KORD-PTPN1:KORD-ABL; p = 0.004) (Fig. 17). Increasing KORD-PTPN1 amounts relative to KORD-ABL further resulted in a 4.3-fold reduction in the total expressed NanoLuciferase (p < 0.0001). Overall, this indicates that KORD-PTPN1 can facilitate a dose-dependent, linear suppression of total output, confirming its inhibitory role within the circuit.

---

Figure 17: Titration of KORD-PTPN1 against KORD-ABL. HEK293T cells were transfected 24 hours after seeding. NanoLuc and Firefly Luciferase expression was measured using the Dual Luciferase assay 48 hours after transfection. Induction occurred 24 h after transfection, using either DMSO or SalB [500 nM]. Each condition was transfected with 30 ng of KORD-ABL and 5 ng of 𝛽-Arrestin2-bZipEE, as well as the amount of KORD-PTPN1 indicated on the X-axis. Data is depicted as the mean +/- SEM with n = 3 technical replicates. Two-way ANOVA with Sidak's multiple comparison test. ***P<0.001, ****P<0.0001.

To further evaluate the total quenching ability of KORD-PTPN1 on a single cell level, we co-transfected KORD-ABL, KORD-PTPN1, 𝛽-Arrestin2-bZipEE, and our phosphorylation circuit with a 5x UAS mCherry expression cassette as well as a normalization plasmid. Cells were induced with 500 nM SalB 24 hours after transfection. Expressed mCherry fluorescence was determined 48 hours after transfection using Flow cytometry. Cells were gated based on GFP fluorescence from a transfected normalization plasmid with constitutive expression. As a control, 180 ng of KORD-ABL was transfected together with the complete phosphorylation circuit and measured under both uninduced (DMSO) and induced (SalB) conditions. For simplicity, only induced conditions of KORD-ABL + KORD-PTPN1 are shown. mCherry fluorescence of DMSO control conditions was equivalent to that of the only KORD-ABL condition, confirming minimal basal activation. Co-transfection of equal amounts of KORD-ABL and KORD-PTPN1 with the expression circuit quenched expression by 6.8-fold. At high KORD-PTPN1 levels (2:1 ratio of KORD-PTPN1 to KORD-ABL), fluorescence approached background intensity (1.4-fold above full quenching achieved by the cytosolic SynPhosphatase), indicating that KORD-PTPN1 efficiently quenches KORD-ABL–mediated signaling (Fig. 18).

---

Figure 18: KORD-PTPN1 quenches expression on a post-translational level. HEK293T cells were transfected with the complete expression-coupled PHOENICS circuit 24 hours after seeding, with varying amounts of KORD-ABL and KORD-PTPN1 as indicated. Cells were induced 24 hours after transfection with DMSO or SalB [500 nM]. mCherry fluorescence was quantified with flow cytometry 48 hours after transfection. 180/0+ DMSO: Uninduced control exhibiting only background fluorescence in the absence of SalB. 180/0+ SalB: Induced positive control showing the signal achievable without quenching by KORD-PTPN1. Test conditions show the plasmid amounts between KORD-ABL and KORD-PTPN1 (180/180: 180 ng KORD-ABL and 180 ng KORD-PTPN1; intracellular components remain constant ). Fluorescence profiles of viable single cell populations (>13000 per condition) are displayed in the ridgeline plot on the left, geometric mean fluorescence (RFU) ± geometric SEM are shown in the barplot to the right.

Building on an extension of the KORD architecture, the Programmable Antigen-Gated Engineered G Protein-coupled Receptor (PAGER) (Kalogriopoulos et al., 2025), we engineered a receptor that detects both a small molecule and a soluble or surface-bound ligand. Targeting the tumor-relevant antigen vascular epithelial growth factor (VEGF), we show the ability to sense, process, and respond to physiologically relevant cues using the PAGER-ABL construct in conjunction with the PHOENICS circuit. This approach enables the PAGER architecture to overcome the existing limitations of synthetic receptors by introducing dual modularity: the extracellular nanobody domain can be readily exchanged to detect diverse soluble or membrane-associated ligands, while the intracellular GPCR backbone can be replaced with β-arrestin-recruiting receptors, ensuring signaling orthogonality.

Furthermore, this design enables flexible reprogramming of input specificity and allows combinatorial processing of signals, while supporting adaptable rewiring of downstream outputs, allowing for seamless integration into the PHOENICS circuit with fully reversible signaling dynamics.

The PAGER-ABL construct consists of the KORD-ABL backbone and intracellular domains. In addition, it contains an extracellular nanobody N-terminally fused with Arodyn, a peptide antagonist of KORD, facilitating constant auto-inhibition. Ligand binding by the nanobody displaces Arodyn from the orthosteric site of KORD, allowing SalB to bind, triggering a conformational change that enables GRK-mediated phosphorylation of residues in the receptor's C-terminal tail (Fig. 19). This recruits β-Arrestin2-bZipEE that can bind Gal4DBD-Cd3ζ-bZipRR via a coiled-coil interaction. The resulting positioning of the CD3ζ near the ABL kinase fused to the GPCR leads to its phosphorylation. The phosphorylated Gal4DBD-CD3ζ-bZipRR subsequently dimerizes with SH2-VP64, forming a transcriptional activator that binds the 5xUAS motif and drives transgene expression.

---

Figure 19: Schematic overview of VEGF sensing PAGER-ABL coupled to the phosphorylation circuit. In the absence of SalB or VEGF, 𝛽-Arrestin2-bZipEE is unable to bind KORD-ABL. Presence of VEGF results in nanobody binding, displacing Arodyn from the orthosteric pocket of PAGER-ABL. This sensitizes the PAGER to SalB, allowing SalB-dependent activation. Receptor activation leads to a conformational change, allowing GRK binding and subsequent phosphorylation of the C-terminal tail of KORD. This allows for the binding of β-Arrestin2-bZipEE. Connection to the intracellular PHOENICS circuit is facilitated by the fused bZipEE domain on β-Arrestin2 and its spatial organization in the proximity of the ABL kinase upon KORD binding. Bound β-Arrestin2-bZipEE recruits Gal4DBD-CD3ζ-bZipRR, which ABL subsequently phosphorylates. Dissociation from the KORD due to the transient coiled-coil interactions allows dimerization between the phosphorylated CD3𝜁 and SH2, reconstituting the split transcription factor Gal4DBD-VP64. Shuttling to the nucleus then facilitates transgene expression under control of the 5xUAS-minCMV promoter.

To test whether the VEGF-sensing PAGER-ABL receptor could act as an AND gate and drive expression through the PHOENICS circuit, we co-transfected cells with VEGF-PAGER-ABL, the phosphorylation circuit, a NanoLuciferase reporter, and a constitutive control for normalization. HEK293T cells were stimulated 24 hours after transfection with SalB (100 nM, 250 nM, or 500 nM) and VEGF₁₆₅ (500 nM). Nano- and Firefly Luciferase activity was determined 48 hours after transfection using the Nano-Glo Dual-Luciferase Reporter Assay System. Induction of PAGER-ABL with either DMSO, SalB, or VEGF did not induce a significant increase in NanoLuciferase expression (p > 0.05). Induction with SalB + VEGF resulted in significant, dose-dependent increases of NanoLuciferase expression. Compared to SalB only, SalB + VEGF induced a 44.2-fold, 48.8-fold, and 28-fold change in expression (Fig. 20). This shows the functionality of a novel receptor architecture that can drive transgene expression in an AND logic gate, binary switch-like manner, without exhibiting significant background or interference in native signaling pathways. Furthermore, the receptor architecture can sense tumor-relevant antigens and induce expression in a dose-dependent manner.

---

Figure 20: VEGF sensing PAGER-ABL represents a novel receptor platform, sensing and responding to a surface-bound or soluble ligand and a small molecule extracellular cues in an AND logic, switch-like manner . Barplot showing NanoLuciferase expression in the presence of different ligands and at different SalB concentrations. HEK293T cells were transfected 24 hours after seeding. NanoLuc and Firefly Luciferase expression was measured 48 hours after transfection using the Dual Luciferase assay. Induction occurred 24 hours after transfection with DMSO, DMSO + VEGF [500 nM], SalB or VEGF [500 nM] + SalB. Each condition was transfected with 60 ng of KORD-ABL and 5 ng of 𝛽-Arrestin2-bZipEE. Data is depicted as the mean +/- SEM with n = 3 technical replicates. Two-way ANOVA with Tukey's multiple comparisons test was performed. ****P < 0.0001; NS, not significant.

Our MESA receptor design was inspired by a recent publication (Yang et al., 2025) and consists of a rapalog binding domain-fused ABL kinase chain (FRB-ABL) and a rapalog binding domain-attached bZipEE coiled coil chain (FKBP-bZipEE), which heterodimerize upon addition of the rapalog (AP21967). Proximity between receptor-fused kinase and substrate-recruiting coiled coil results in the phosphorylation of our regulatory substrate CD3ζ. This activates our previously established expression platform: an SH2-Gal4DBD reader protein and a bZipRR-VP64 activator, which is fused to CD3ζ. SH2 selective binding to phosphorylated CD3ζ enables the reconstitution of a functional Gal4DBD-VP64 transcription factor, driving NanoLuciferase reporter expression (Fig. 21).

---

Figure 21: NanoLuciferase expression coupled PHOENICS circuit with rapalog induction via MESA kinase receptors. Adapted MESA-receptors consist of a bZipEE coiled coil-carrying and a ABL kinase fused chain. Receptors bind rapalog via extracellular FRB/FKBP domains, mediating receptor dimerization and inducing proximity between receptors fused ABL kinase and coiled coil bZipEE. bZipEE recruits the substrate-construct via a complementary coiled coil bZipRR. Proximity-induced phosphorylation of CD3ζ results in binding of SH2. Gal4DBD and VP64 reconstitute. This drives NanoLuciferase reporter expression upon nuclear shuttling of the reconstituted transcription factor complex, producing a measurable luminescence signal.

To verify that ligand-dependent receptor-dimerization leads to substrate-recruitment and phosphorylation, cells were co-transfected with equal amounts of FRB-ABL and FKBP-bZipEE receptor constructs in combination with a bZipRR-fused CD3ζ-substrate. To assess ligand-dependent substrate-phosphorylation, western blot with phosphospecific and anti-Myc Antibody-staining was performed for rapalog-induced and uninduced conditions (Fig. 22A). To confirm ligand-inducible expression via our Nanoluciferase platform, cells were co-transfected with optimal ratios of FRB-ABL and FKBP-bZipEE, alongside the phosphorylation-responsive protein-circuit components and NanoLuciferase reporter construct. NanoLuciferase activity was quantified by luminometry and normalized by Firefly luciferase, which was co-transfected at equimolar amounts across all conditions, accounting for possible variations in cell viability and transfection efficiency (Fig. 22B).

---

Figure 22: Verification of rapalog-induced phosphorylation and SH2-mediated reporter activation. (A) Western blot of myc-tagged substrate using anti-phosphotyrosine (pY) and anti-myc antibodies. The strong myc signal in the positive control results from reusing the same secondary antibody without stripping after pY detection, leading to signal accumulation. (B) Quantification of SH2-mediated reporter activation using the Nano-Glo assay . Cells were transfected with the complete expression-coupled PHOENICS circuit, varying only the receptor configuration as indicated. Receptor conditions contained a 1:9 ratio of ABL to bZipEE halves (10 ng total DNA). Pos. ctrl: cytosolic kinase fused to a coiled coil; neg. ctrl: FRB-ABL receptor half lacking the coiled coil half. NanoLuciferase luminescence values (AU) are normalized by Firefly luminescence (AU) and display mean ± SEM. Normalized luminescence was analyzed by two-way ANOVA. P values: ns > 0.05; * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001.

In the Western Blot, cytosolic ABL in combination with the coiled coil-fused substrate serves as a positive control for robust phosphorylation. As a negative control, we used a receptor construct without coiled coil, disrupting substrate-recruitment to the receptor fused ABL kinase. Anti-Myc-staining shows clear bands for both rapalog-induced and uninduced receptors at the length of our CD3ζ substrate, confirming successful protein loading. Anti-phosphorylation-staining results in a clear band for the induced condition, while uninduced receptors did not exhibit detectable phosphorylation.

In luminometry, cytosolic ABL in combination with intracellular circuit components again serves as a positive control. As a negative control, the FRB-ABL receptor chain and circuit components were transfected without the FKBP partner to assess background phosphorylation. Rapalog induction produced a 3.4-fold increase in NanoLuciferase expression compared to uninduced conditions, demonstrating robust ligand-dependent transcriptional activation through our synthetic phosphorylation receptor platform.

This demonstrates ligand-dependent phosphorylation of our CD3ζ substrate using our receptor designs and verifies ligand-inducible expression via our MESA-receptor-coupled phosphorylation circuit.

After establishing ligand-dependent activation of gene expression, we engineered what is, to our knowledge, an entirely novel platform for inhibitory ligand sensing and integrated it into our circuit. For this, we replaced the truncated ABL kinase with the truncated phosphatase PTPN1 on the FRB receptor chain. This MESA-receptor now dephosphorylates our regulatory substrate upon ligand-induced dimerization, which prevents SH2-mediated transcription factor reconstitution and inhibits expression (Fig. 23).

---

Figure 23: Rapalog-mediated suppression of NanoLuciferase expression via the PHOENICS circuit coupled to MESA phosphatase receptors. Ligand-dependent control of proximity-based dephosphorylation was achieved by replacing the intracellular kinase domain of one MESA receptor with a phosphatase, while retaining the bZipEE motif on the other receptor half. Co-transfection with a cytosolic kinase ensures constitutive phosphorylation of the substrate, enabling detection of ligand-induced dephosphorylation events. Upon rapalog binding, MESA receptor dimerization brings the phosphatase into proximity with the substrate, promoting dephosphorylation. Loss of substrate phosphorylation disrupts SH2 binding and prevents reconstitution of the Gal4DBD and VP64 transcriptional activator, thereby suppressing NanoLuciferase reporter expression and reducing luminescence output.

To demonstrate ligand-dependent inhibition of expression, we co-transfected optimal ratios of our inhibitory receptor constructs alongside cytosolic ABL to maintain basal substrate phosphorylation, the intracellular circuit components, and the NanoLuciferase reporter construct. NanoLuciferase activity was again quantified by luminometry and Firefly Luciferase normalization (Fig. 24).

---

Figure 24: Rapalog-induced reporter expression quenching by phosphatase receptor. Quantification of SH2-mediated reporter activation using the Nano-Glo assay. Cells were transfected with the complete expression-coupled PHOENICS circuit, varying only the receptor configuration as indicated. Neg. ctrl: FKBP-bZipEE receptor half lacking the phosphatase-fused construct. Receptor conditions contained a 1:4 ratio of PTPN1 to bZipEE halves (10 ng total DNA). NanoLuciferase luminescence values (AU) are normalized by Firefly luminescence (AU) and display mean ± SEM. Luminescence was analyzed by two-way ANOVA. P values: ns > 0.05; * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001.

As a negative control for background gene expression, we transfected intracellular circuit components and our FRBP-bZipEE construct. Rapalog treatment of functional receptor conditions resulted in a 2.8-fold reduction in normalized NanoLuciferase expression compared to uninduced states, verifying reliable, ligand-dependent inhibition of gene expression.

This demonstrates functionality of our phosphatase-fused receptors and enables intracellular processing of two opposing ligand signals with our MESA-architectures coupled to our phosphorylation circuit.

After integrating stimulatory and inhibitory ligand sensing into our circuit, we set out to verify target flexibility of our synthetic receptor architecture. By exchanging the extracellular domains of our receptor constructs with anti-TNFα scFvs, we established soluble protein-dependent activation of gene expression (Fig. 25).

---

Figure 25: NanoLuciferase expression coupled PHOENICS circuit with TNFα induciton via MESA kinase receptor. Adaptation of the initial reporter assay to assess TNFα-mediated reporter expression via anti-TNFα-scFv-fused MESA-receptors. MESA-receptors bind TNFα via extracellular anti-TNFα-scFvs. TNFα-binding mediates receptor dimerization, inducing proximity between receptors fused ABL kinase and coiled coil bZipEE. bZipEE recruits the substrate-construct via a complementary coiled coil bZipRR. Proximity-induced phosphorylation of CD3ζ results in binding of SH2. Gal4DBD and VP64 reconstitute. This drives NanoLuciferase reporter expression upon nuclear shuttling of the reconstituted transcription factor complex, producing a measurable luminescence signal.

To show induction of gene expression by TNFα-sensing, we co-transfected our adapted receptor constructs, intracellular circuit components, and the NanoLuciferase reporter construct. NanoLuciferase activity was quantified by luminometry and Firefly Luciferase normalization (Fig. 26).

---

Figure 26: TNF-induced reporter activation by MESA kinase receptor. Quantification of SH2-mediated reporter activation using the Nano-Glo assay. Cells were transfected with the complete expression-coupled PHOENICS circuit, varying only the receptor configuration as indicated. Neg. ctrl: RAPA-ABL receptor half lacking the bZipEE half. Receptor conditions contained a 1:4 ratio of ABL to bZipEE halves (10 ng total DNA). NanoLuciferase luminescence values (AU) are normalized by Firefly luminescence (AU) and display mean ± SEM. Luminescence was analyzed by two-way ANOVA. P values: ns > 0.05; * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001.

As a positive control, cytosolic ABL kinase was added to facilitate strong substrate phosphorylation. Conditions including only the intracellular circuit components alone as well as the ABL receptor monomer served as negative controls of background gene expression and background phosphorylation byFRB-ABL without its FKBP-bZipEE-partner. For functional receptor conditions, treatment with TNFα resulted in a 1.8-fold increase of NanoLuciferase expression, confirming soluble-protein-dependent activation of our circuit, albeit needing optimization to reduce background expression.

To refine expression control for minimized background activity, we utilized our dry lab model SPARC and optimized receptor transmembrane domains (TMDs) with coarse-grained molecular dynamics. To reduce spontaneous dimerization we exchanged CD28 and CD28-M3 TMDs with FGFR1 TMDs. Additionally we fine tuned receptor construct ratios for precise TNFα-inducible phosphorylation of our regulatory substrate. To assess phosphorylation efficiency of different designs at single-cell resolution, we adapted our circuit for a flow cytometry readout. We replaced Gal4DBD and VP64 with halves of the splitFast fluorophore system to quantify SH2-mediated assembly of the intracellular circuit components. Upon reconstitution of circuit components and addition of HMBR, SplitFast becomes fluorescently active (Fig. 27).

---

Figure 27: TNFα-inducted split fluorophore reconstitution in PHOENICS circuit via MESA kinase receptor. MESA-receptors bind TNFα via extracellular anti-TNFα-scFvs. TNFα-binding mediates receptor dimerization, inducing proximity between receptors fused ABL kinase and coiled coil bZipEE. bZipEE recruits the substrate-construct via a complementary coiled coil bZipRR. Proximity-induced phosphorylation of CD3ζ results in binding of SH2. Split fluorophore constructs RspA(C) and RspA(N) reconstitute, yielding fluorescence upon addition of HMBR, producing a measurable reporter for CD3ζ-SH2 association.

To demonstrate reduced background activity of our improved TNFα-MESA, we compared the previous design with FGFR1 TMD constructs, which was transfected at optimal ratios of ABL-chain and bZipEE-chain, as predicted by SPARC, and quantified CD3ζ phosphorylation with flow cytometry (Fig. 28).

---

Figure 28: TNFα-induced CD3ζ-SH2 constitution via MESA-like kinase receptor. Quantification of TNFα-induced phosphorylation-mediated CD3ζ-SH2 constitution. Cells were transfected with the complete expression-coupled PHOENICS circuit, varying only the receptor configuration as indicated. Neg. ctrl: No receptor. Receptor conditions contained specified ratios of ABL to bZipEE halves (170 ng total DNA in 12-well plate wells). SplitFast fluorescence was quantified with flow cytometry. Dimerization dependent SplitFAST fluorescence was quantified by flow cytometry 48 hours after transfection, employing excitation lasers at 488 nm and 561 nm. Emission was recorded at 525 nm for splitFAST (Ligand: TFLime) and at 610 nm for constitutively expressed mCherry, serving as a transfection control. (A) Fluorescence profiles of viable single cell populations are displayed in ridgeline plot, (B) geometric mean fluorescence (RFU) ± geometric SEM with ~9000 single cells are shown in barplot.

As a negative control, we transfected our SplitFast-fused intracellular components without receptor constructs to display background fluorescence without substrate-phosphorylation. For functional receptor conditions, FGFR1-transmembrane domains and an optimized ratio of receptor constructs for reduced background activity exhibit a clear shift of the cell population towards higher fluorescence upon TNFα treatment. With these changes in place we move from a 1.5-fold increase of geometric mean of fluorescence between uninduced and induced states for the old ratios and transmembrane domains to a 3.7-fold increase. Optimized receptor construct ratios and transmembrane domains result in significantly lower background phosphorylation in the uninduced state, likely due to minimized spontaneous dimerization. We expect to transfer these improvements in dynamic range to our expression system, which is mediated by the constitution of the intracellular circuit components measured in this assay, in the near future.

Overall, this serves as a proof-of-concept for processing soluble-protein ligands in our phosphorylation circuit via our MESA-receptor platform. We showcase its target flexibility by readily exchangeable extracellular binding domains, opening up various retargeting options to tumor-microenvironment associated soluble proteins.

After establishing both, small-molecule and soluble-protein gated synthetic MESA-receptors as a sensing layer, we proceeded to validate simultaneous processing of multiple ligand-signals in our phosphorylation circuit. Guided by SPARCs mathematical modeling predicting optimal receptor ratios, we integrated both our kinase-carrying TNFα-receptors and our ralapog-gated phosphatase-receptors within one cell, forming a cellular processing unit: Gene expression is only triggered when the stimulatory TNFα-signal outweighs the inhibitory rapalog input, providing an additional safety layer and enabling precise environmental control of cellular outputs (Fig. 29).

---

Figure 29: NanoLuciferase expression coupled PHOENICS circuit with TNFα induction and rapalog inhibition via MESA receptors. Adaptation of the initial reporter assays to assess TNFα-induced reporter expression via anti-TNFα-scFv-fused MESA-receptors and rapalog-controlled quenching. ABL kinase carrying MESA-receptors bind TNFα via extracellular anti-TNFα-scFvs. TNFα-binding mediates receptor dimerization, inducing proximity between ABL and coiled coil bZipEE. bZipEE recruits the substrate-construct via a complementary coiled coil bZipRR, resulting in proximity-induced phosphorylation of CD3ζ. SH2 binds phosphorylated CD3ζ, with subsequent Gal4DBD-VP64 transcription factor constitution and NanoLuciferase reporter expression. Rapalog-binding of PTPN1 phosphatase-carrying receptor induced dimerization, resulting in substrate dephosphorylation, quenching reporter expression. NanoLuciferase expression can be quantified by luminometry.

To showcase effective expression regulation in a two-ligand-setting, we co-transfected stimulatory and inhibitory receptors alongside our intracellular circuit components and the NanoLuciferase reporter construct. NanoLuciferase activity was quantified by luminometry and normalized by Firefly Luciferase (Fig. 30).

---

Figure 30: Dual-receptor logic for processing dual-ligand-specific reporter activation. Quantification of SH2-mediated reporter activation using the Nano-Glo assay. Cells were transfected with the complete expression-coupled PHOENICS circuit, varying only the receptor configuration as indicated. Neg. ctrl: No receptor. Receptors were co-transfected with 10 ng total DNA using a 1:4 ratio of TNF to rapalog receptors. For the rapalog receptor, a 1:3 ratio of phosphatase to bZipEE halves was used; for the TNF receptor, a 1:5 ratio of TNF-ABL to bZipEE halves. NanoLuciferase luminescence values (AU) are normalized by Firefly luminescence (AU) and display mean ± SEM. Luminescence was analyzed by two-way ANOVA. P values: ns > 0.05; * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001.

Circuit components without receptor constructs serve as a negative control displaying only background expression. For conditions including both, stimulatory TNFα-receptor and inhibitory rapalog-receptor, TNFα treatment produced a 2.0-fold increase relative to the uninduced state. Simultaneous TNFα and rapalog exposure fully suppressed activation, returning NanoLuciferase to baseline expression of the uninduced system. This demonstrates functional ligand-dependent AND-NOT gating of expression. Future experiments will focus on the reduction of background activity by optimizing transmembrane domains and calibrating receptor stoichiometry, as shown above. By adjusting the relative abundance of stimulatory versus inhibitory receptors, the switching threshold of stimulatory versus inhibitory ligand concentrations can be positioned at defined ligand ratios. This maintains dormancy in healthy tissues and robust expression of therapeutically relevant effectors when tumor microenvironment specific ligand profiles are encountered.

Overall, these results serve as a proof-of-concept for the computation of opposing signals with our phosphorylation circuit, which can be tailored to produce precise cellular outputs. By exchanging extracellular binding domains on our MESA-receptors, we have showcased straightforward retargetability of our receptors to cancer-relevant soluble-proteins. This unlocks various potential targets that shape and are indicative of the tumor microenvironment. We have further expanded the accessible ligand space with our dry lab efforts by successfully generated several de novo protein binders for cancer-relevant ligands with experimental validation demonstrating our pipeline's reliability. This enables personalized applications of the PHOENICS system, which can be re-tuned to programmed ligand combinations, allowing therapeutic effector functions to be gated by ligand ratios characteristic of specific tumor microenvironments.

Having designed protein binders for retargeting our sensory receptor platforms by exchange of extracellular binding domains on MESAs or nanobody-replacement on our GPCRs, we set out to achieve experimental validation of our binder designs. To rapidly assess computationally designed binders against GDF-15, a protein upregulated in cancers as shown by (Zhao et al., 2020). We implemented the Cell-Free Two-Hybrid (CF2H) system as described by (Capin et al., 2025). This approach enables protein-protein interaction screening without requiring cloning, protein purification, or specialized equipment. This method utilizes a cell-free expression system to selectively express reporter sfGFP by binder-mediated reconstitution of split transcription factors. Biotinylated GDF-15-Fc was complexed with streptavidin to induce multimerization and then spiked into cell-free reactions containing linear DNA encoding CIopt-fused binder constructs and a pRM-sfGFP reporter plasmid. After expression, binder-target interaction induces CIopt dimerization, which activates pRM promoter-driven sfGFP expression, providing a fluorescent readout to functionally assess binding capabilities of the binder design to its designated target (Fig. 31)

---

Figure 31: Cell-Free Two-Hybrid (CF2H) assay for functional binder validation. Functional validation of protein-binders with sfGFP expression readout. Biotinylated target proteins complex with the streptavidin tetramer. This complex is added to a cell-free expression system in combination with linearized DNA encoding for a CIoptDBD-fused binder and a reporter plasmid, encoding sfGFP under pRM-promotor control. CIoptDBD-binder expression results in binding to target protein with subsequent CIopt-dimerization. Two dimerized CIopt domains bind to Operon 1 and Operon 2 of the pRM-promoter inducing expression and subsequently yielding a measurable fluorescence signal.

We observed some basal promoter leakage for the negative control, indicating moderate residual pRM activity in the absence of CIopt activation. This was however significantly lower compared to our positive control, confirming the viability of this assay. Binder B3 demonstrated highest sfGFP fluorescence upon ligand induction out of all tested candidates, while in the absence of biotinylated GDF-15 it produced fluorescence not higher than the autofluorescence of the cell-free extract (Fig. 32). This confirms that sfGFP was specifically expressed due to binder-target interaction rather than non-specific aggregation. Binder-3 exhibited higher fluorescence than our positive control, however direct comparison of binding affinities is limited by varying protein concentrations as binders were expressed in situ by the cell-free system. Additionally the molar amount of GDF-15 binder DNA slightly exceeded the amount of the positive control binder.

---

Fig 32: Functional validation of de novo binders with the CF2H assay. Functional assessment of in silico generated de novo binders against GDF-15 using the CF2H assay. Neg. ctrl: No DBD-binder encoding DNA. Pos. ctrl: DNA encoding for NbALFA nanobody paired cognate ALFA-tag epitope. sfGFP fluorescence [AU] was measured for induction with GDF-15 and compared to fluorescence without target protein, displaying mean ± SEM. Fluorescence was analyzed by two-way ANOVA. P values: ns > 0.05; * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001.

While quantitative comparison is thus limited, these results qualitatively confirm the binding capabilities of our de novo designed binder to GDF-15. Binders B2 and B7 also showed detectable binding activity, albeit without or reduced significant difference compared to background sfGFP expression. B1 and B5 did not produce detectable sfGFP fluorescence over the autofluorescence of the extract measured in Neg. CT2. Overall, these results demonstrate that our de novo binder actively targets cancer-relevant GDF-15. This indicates the potency of our dry lab model SPARC in designing functional binders for select target proteins and highlights the utility of CF2H as an accessible and rapid validation platform for de novo designed protein binders. Going forward, this established pipeline linking in silico design and experimental validation enables us to engineer custom binding moieties for retargeting our receptor platforms.

We created our synthetic GPCR sensing platform by fusing a truncated version of the ABL kinase or PTPN1 phosphatase to the κ-opioid receptor DREADD (KORD) and recruiting the CD3ζ substrate to β-Arrestin2 via complementary coiled coils. When the GPCR is activated by ligand binding, it undergoes a conformational change that mediates 𝛽-Arrestin2-bZipEE binding, thus ensuring that proximity-induced (de–)phosphorylation events occur exclusively after ligand binding.

This architecture exhibited highly accurate ligand-inducible control over reporter expression with robust fold changes in both luminometry and flow cytometry. By integrating either SynKinase or SynPhosphatase fusions, we could readily switch between ligand-inducible activation (KORD-ABL) and inhibition (KORD-PTPN1), enabling processing of both stimulatory and inhibitory signals. Motivated by these results, we further extended this GPCR platform by implementing PAGER (Programmable Antigen-gated G-Protein coupled Engineered Receptors) designs, where extracellular nanobodies carrying an arodyn-blocking peptide enabled dual-input AND-gating of small-molecule and protein ligands. After achieving proof-of-concept for the PAGER design as a sensory component in our phosphorylation circuit, we retargeted PAGER to VEGF, successfully establishing cancer-relevant soluble protein sensing coupled to an expression output. Collectively, these results establish GPCR-based (de-)phosphorylation receptors as a highly versatile, tight platform for precise environmental control of cellular activity in the PHOENICS circuit.

Our MESA designs were engineered by fusing ABL kinase or PTPN1 phosphatase to receptors carrying rapalog-binding domains and recruiting CD3ζ via a coiled coil. Ligand binding induces receptor dimerization and brings the substrate into proximity to the kinase or phosphatase, yielding ligand-inducible transcriptional activation. Furthermore, by exchanging extracellular domains (ECDs) for scFvs, we showed that MESAs are readily retargetable to soluble proteins such as TNFα, confirming their modularity as a sensing layer. Finally, by integrating kinase- and phosphatase-fused MESA receptors in one cell, we demonstrated PHOENICS' ability to process opposing extracellular cues, resulting in flexible switching between active and dormant states.

Our current MESA designs exhibit moderate spontaneous background dimerization, indicated by elevated reporter expression compared to baseline levels in the absence of an inducing ligand. Despite optimization cycles guided by our SPARC modeling, including transmembrane domain engineering and construct ratio tuning, leaky expression in the OFF-state could only be partially suppressed. This resulted in smaller fold changes compared to our synthetic GPCRs and limits the dynamic range of our MESA receptors. While our MESA platform provides a versatile and modular architecture, we plan to continue optimization efforts, which will likely prove advantageous for applications requiring highly precise control of cellular outputs.

Using the previously established VEGF-PAGER-ABL, our optimized PhosphoTEV and the NanoLuc-RELEASE reporter, we implemented a circuit to make RELEASE VEGF-inducible (Fig. 33). To simulate an application in which SalB is present systematically and circuit assembly should occur only in the presence of VEGF, NanoLuc secretion was compared between SalB alone and with VEGF. Co-stimulation 24 hours post-transfection with SalB led to a significant increase in secretion (p=0.03), corresponding to a modest 1.5-fold change. To estimate the basal activity of SalB alone, a separate no-kinase RELEASE condition was included for reference. In this comparison, SalB alone appeared to generate elevated background relative to the no-kinase baseline, suggesting partial leakiness of PAGER gating in the absence of VEGF.

---

Figure 33: VEGF-inducible NanoLuc secretion. VEGF-inducible NanoLuc secretion. HEK293T cells were transfected with VEGF-PAGER-ABL, PhosphoTEV, and NanoLuc-RELEASE plasmids 24 hours post-seeding. 24 hours post-transfection, they were induced with either SalB [500 nM] or SalB [500 nM] + VEGF [500 nM]. The data is shown as the mean +/- SEM with n = 3 technical replicates. A 1-tailed Student's t-test was performed to test for an increase with VEGF induction. Positive and negative controls were excluded from testing and are shown as a reference. *P<0.05.

To prove the modularity of our secretion module, we exchanged the NanoLuc-RELEASE reporter with Interleukin-12 (IL-12). IL-12 is a potent cytokine that activates natural killer and cytotoxic T cells and exhibits anti-angiogenic effects, inhibiting tumor growth and counteracting the immune-suppressive tumor microenvironment. We validated it on a simple circuit using two different amounts of the KORD-ABL construct. The cells were induced with SalB after 24 hours and secreted IL-12 was measured in the supernatant another 24 hours later using a sandwich ELISA with a chemiluminescent readout (Fig. 34). For both KORD-ABL amounts, a significant increase in secreted IL-12 could be observed. However, there was a substantial increase without SalB induction compared to the negative control without any protease, indicating leakage of the system. Nevertheless, this proves the system working as planned in principle.

---

Figure 34: SalB induced IL-12 secretion using PhosphoTEV-RELEASE and KORD-ABL. HEK293T cells were co-transfected with β-Arrestin2-bZipEE, IL-12-RELEASE, PhosphoTEV and KORD-ABL plasmids. Cells were induced with 3 µM SalB 24h post-transfection. IL-12 in 100 µL of supernatant was measured by ELISA with a chemiluminescence readout 48h post-transfection. Data is depicted as the mean +/- SEM with n = 3 technical replicates. Statistical significance was calculated with a Two-way ANOVA with Sidak's multiple comparisons test. *P<0.05, **P<0.01.