Computational design of synthetic receptors with programmable signalling activity for enhanced cancer T cell therapy

Rath, Jan A.; Rudden, Lucas S. P.; Nouraee, Nazila; Que, Tiffany X. Y.; Von Gunten, Christine; Perez, Cynthia; Birch, Flora; Bhugowon, Yashashvi; Fueglistaler, Andreas; Chatzi Souleiman, Aisima; Barth, Patrick; Arber, Caroline

doi:10.1038/s41551-025-01532-3

Download PDF

Article
Open access
Published: 28 October 2025

Computational design of synthetic receptors with programmable signalling activity for enhanced cancer T cell therapy

Nature Biomedical Engineering (2025)Cite this article

4426 Accesses
93 Altmetric
Metrics details

Subjects

Abstract

The tumour microenvironment (TME) plays a key role in tumour progression, and soluble and cellular TME components can limit CAR-T cell function and persistence. Targeting soluble TME factors to enhance anti-tumour responses of engineered T cells through chimeric receptors is not broadly explored owing to the unpredictable signalling characteristics of synthetic protein receptors. Here we develop a computational protein design platform for the de novo bottom-up assembly of allosteric receptors with programmable input–output behaviours that respond to soluble TME factors with co-stimulation and cytokine signals in T cells, called TME-sensing switch receptor for enhanced response to tumours (T-SenSER). We develop two sets of T-SenSERs targeting vascular endothelial growth factor (VEGF) or colony-stimulating factor 1 (CSF1) that are both selectively enriched in a variety of tumours. Combination of CAR and T-SenSER in human T cells enhances anti-tumour responses in models of lung cancer and multiple myeloma, in a VEGF- or CSF1-dependent manner. Our study sets the stage for the accelerated development of synthetic biosensors with custom-built sensing and responses for basic and translational cell engineering applications.

Main

Adoptive T cell therapies for cancer using T cells engineered with chimeric antigen receptors (CAR-T cells) are at the forefront of clinically applied synthetic immunology^1,2,3. Direct cytotoxic targeting of tumour cells by adoptive transfer of CAR-T cells can produce remissions in chemo-refractory disease and has changed clinical practice for patients with B cell malignancies^4,5,6,7,8 and multiple myeloma (MM)^9,10,11. Nevertheless, in B cell malignancies, more than half of patients do not achieve long-term remissions, and durability of response is still limited in MM. In solid tumours, broad success is lacking^12,13, and so far, only one αβ-T cell receptor (TCR)-based adoptive T cell therapy has obtained regulatory approval in the United States¹⁴, and not a single CAR-T cell therapy has reached this stage.

To achieve sustained anti-tumour responses in vivo, CAR- or TCR-transgenic T cells must not only recognize and kill tumour cells but also receive distinct co-stimulation and cytokine signals from the environment. In most tumour microenvironments (TMEs) however, co-stimulation is dominated by co-inhibition¹⁵, and cytokine signals that sustain T stem cell and central memory populations with high proliferative capacity, as well as those supporting effector functions, are lacking or overruled by immunosuppressive signals¹⁶. Current TCR-based approaches only provide target antigen recognition and completely rely on environmental factors for additional signals. The co-stimulation built into second-generation CAR-T cell therapeutics is often insufficient and only partially compensates for the lack of stimulatory environmental signals. So far, most TCR, CAR and co-receptor constructs that equip therapeutic immune cells have been engineered empirically without mechanistic optimization and diversification of their signalling functions, thus limiting the development of more powerful and targeted cellular approaches.

In principle, these limitations could be overcome through the design of biosensing receptors with fully customized molecular properties and associated cellular functions^17,18. Modern computational protein design techniques can engineer proteins with a wide variety of structures and binding properties^19,20,21. However, except for protein-based materials that assemble into specific supramolecular architectures²², these methods have mostly been applied to single protein domains^23,24,25. The design of dimeric multi-domain signalling receptors remains challenging. The proper orchestration of these receptor functions requires specific ligand-induced protein association, structural switching and long-range communication between domains to ensure signal transduction^26,27,28,29. These properties are essential for the development of biosensors providing precise ligand control of cellular activities but have been neglected so far.

Here we developed a computational approach for the bottom-up assembly and design of multi-domain receptors with programmable input–output signalling functions. We applied the approach to engineer a class of receptors that we named T-SenSER (TME-sensing switch receptor for enhanced response to tumours). T-SenSERs function as allosteric biosensors and can be co-expressed with conventional CARs in T cells to considerably enhance their function (Fig. 1a). T-SenSERs are designed to detect and bind soluble factors in the TME and transmit signals of co-stimulation and common γ-chain cytokines to the engineered T cells. We exploited two distinct soluble factors, vascular endothelial growth factor A (VEGFA) and colony-stimulating factor 1 (CSF1), that are enriched in a broad variety of haematologic and solid cancers, respectively. When co-expressed as transgenes with different CARs, T-SenSERs delivered enhanced potency and specificity to CAR-T cells in a TME-responsive manner.

**Fig. 1: Concept and design of de novo assembled biosensors (T-SenSERs).**

Results

Computational receptor design with programmable signalling

A membrane receptor biosensor architecture can be schematically decomposed into two elements: (1) an extracellular (EC) ligand-binding sensor and (2) an intracellular (IC) signalling responder connected by a transmembrane (TM) domain. Communication between sensor and responder (that we define as coupling below) enables signal transduction across the membrane and activation of IC functions upon ligand binding. While the underlying structural mechanisms may vary, sensing-response behaviours rely on at least the following three sequential steps: the sensor changes conformation upon ligand binding, the sensor transmits this structural change to the responder, and the responder switches to an active state conformation and triggers receptor activity (Fig. 1b).

As in any allosteric system, sensing-response properties can be achieved through diverse design scenarios that impact the receptor’s basal activity, sensitivity and potency. These scenarios are outlined in Extended Data Figs. 1 and 2. Specifically, each element can independently switch between an inactive and active state and preferentially occupy one state in isolation (Extended Data Fig. 1a). When combined, different biosensor behaviours will be obtained depending on individual bias between inactive/active state and the mechanical coupling between the sensor and responder that will impact the state occupancies. For example, a programmable biosensor scaffold could involve sensor and responder elements that preferentially occupy an inactive and active state conformation, respectively, in isolation and absence of ligand (Extended Data Fig. 1b) or alternative scenarios (Extended Data Figs. 1c and 2b,c). If the sensor and responder are weakly coupled, the responder will readily access the active state and trigger high receptor basal activity (that is, without ligand). Conversely, strong coupling will maintain the responder mostly in the inactive state, turning off basal activity, while still enabling a strong ligand-induced response.

Studies of natural single-pass membrane receptors indicate that while the structural mechanisms underlying biosensing functions can be diverse, they usually involve two main structural modes of activation. In the pre-formed dimer (PFD) mode reported, for example, for the interleukin-7 receptor (IL-7R) or death receptor 5 (DR5) cytokine receptors^30,31,32, the receptor self-associates in the absence of ligand but mainly occupies inactive state dimer conformations. Ligand binding triggers intramolecular reorganization propagated allosterically to the cytoplasmic region through coupling, stabilizing active state conformations of the dimer structure. In the monomer–dimer equilibrium shift (MDE) mode initially described for the epidermal growth factor receptor (EGFR)³³, the ligand-free receptor mostly occupies a monomeric inactive state (Extended Data Fig. 1). Ligand binding triggers receptor association and the formation of active state dimer conformations. However, structural insights into receptor activation are sparse and the prevalence of each mechanism remains highly debated. In fact, several lines of evidence suggest that receptor activation may involve a combination of both allosteric and binding mechanisms³⁴.

Since the structural mechanisms of activation of single-pass TM receptors remain largely elusive, we reasoned that a simplified but effective approach for designing receptor biosensors should primarily focus on building optimal active state structures. While this positive design strategy neglects alternative inactive or pre-active states, optimizing structural features for a single target state has proven effective in many protein engineering studies. To account for the diverse activation mechanisms, optimization focuses on the two main structural properties driving signal transduction: dimerization and mechanical coupling, both underlying the MDE and allosteric PFD modes in the active state. In principle, a wide range of constitutive and ligand-inducible activities can be programmed through the modulation of dimerization propensity and communication in the active state. Together, they provide a rational blueprint for engineering receptor scaffolds with diverse and precise sensing and signalling functions^35,36 (Methods, Fig. 1c and Extended Data Figs. 1–3).

We developed a computational approach based on these rules to design dimeric biosensors that link the binding of a user-defined chemical input signal to modular cellular responses through genetically encoded fusions of protein domains. Overall, the approach proceeds in the following main steps (Methods, Fig. 1c and Extended Data Fig. 4): (1) selection of the structural elements defining input, signal transmission and output signals: (1.1) the sensor (EC ligand-binding domain and dimerizing domains that link to the responder) and (1.2) the responder (TM and IC signalling domains); (2) if necessary, self-association of individual domains into dimeric complexes through docking, and design of juxtamembrane linker sequence connecting the sensor and responder; (3) assembly of multi-domain dimeric scaffolds using structure prediction methods RoseTTAfold and AlphaFold2, and design protocols in Rosetta; (4) ranking of receptor scaffold structures based on their propensity for dimerization and long-range communication (that is, coupling) between ligand-binding and signalling domains that are calculated using Rosetta and Elastic Network models for investigating protein association and coupling, respectively (Methods). As mentioned above, this protocol only constructs an ensemble of conformations for the active ligand-bound state of the biosensor. It neglects the impact of the design decisions on alternative states or transitions between states that would require a detailed mechanistic understanding of the activation process. Nevertheless, this positive design selection for optimal oligomerization and coupling in active dimer structures should generate computational libraries of receptors enriched in constructs with desired sensing-response behaviours.

VEGFA and CSF1 were chosen as ligands, two soluble factors that are highly enriched in a variety of TMEs and critically involved in tumour progression. As VEGFs promote neovascularization and CSF1 supports tumour-associated macrophage development and polarization in various TMEs, these targets open broad fields of applications and offer high translational impact for our T-SenSERs^37,38. c-MPL signalling was selected as the output signal since we have previously shown that c-MPL activates beneficial co-stimulatory, cytokine and type I interferon pathways when expressed in TCR-transgenic T cells^39,40. We aimed to design robust VEGF-MPL-receptor (VMR) scaffolds that are entirely VEGF dependent, and CSF1-MPL-receptor (CMR) scaffolds that have a low but significant basal activity with full CSF1 ligand inducibility (‘low constitutive-inducible’). We hypothesized that a low constitutive-inducible CMR has the capacity to counterbalance an immune-suppressive TME by enhancing T cell homeostasis and proliferative capacity in the absence of T cell stimulatory cytokines. Thus, low constitutive-inducible CMR activity is expected to sustain local CAR-T cell persistence and anti-tumour function in the TME (Fig. 1d).

We first analysed the topology and individual domain structures of the native vascular endothelial growth factor receptor 2 (VEGFR2), colony-stimulating factor 1 receptor (CSF1R) and c-MPL receptors. While the structures of the full-length receptors remain elusive, several domain structures have been characterized^27,41,42. Both VEGFR2 and CSF1R contain immunoglobulin-like domains, out of which domains D2 and D3 strongly bind to their cognate ligand. These two domains were selected as the input signalling region of the VMR or CMR, respectively. For the output signalling, the structure and activation mechanism of cytokine receptor homologues of c-MPL indicate that strong ligand regulation and potent JAK/STAT signalling are achieved through the intricate coupling between the cytokine TM, juxtamembrane (JM) and the cytoplasmic (CT) regions^43,44. Hence, we reasoned that an optimal biosensor scaffold should couple the native TM region of c-MPL and not that of VEGFR2 or CSF1R to the c-MPL CT domain. In the absence of structural information, we modelled the c-MPL TM domain in a dimer active signalling state from sequence using the method EFDOCK-TM and then assembled the entire signalling (TM + JM + CT) region using our assembly approach (Methods). We next curated a library of all known native VEGFR and CSF1R EC domain structures, the recombination of which could modulate coupling between the input and output signalling domains in engineered receptors. Unlike c-MPL, all seven IgG-like VEGFR2 EC domains (D1–7) have been structurally characterized and the isolated D4, D5 and D7 domains are known to homodimerize²⁷. Likewise, the D5 and D6 domains are critical for CSF1R homodimerization⁴².

Next, we created a diverse set of chimeras to stringently test our ability to rationally design full-length receptor scaffold structures with fine-tuned signal transduction propensity. Our computational pipeline enables the arbitrary combination of VEGFR or CSF1R EC domains with pre-defined linkers to sample different densities of contacts across the receptor dimerization interface and encode different levels of mechanical coupling. Final designs are returned as a dynamic ensemble of conformations. Thus, the method offers a computationally inexpensive means of obtaining biophysically relevant states from which to assess the modulation of signal transmission triggered by VEGF or CSF1 binding (Methods and Fig. 1c).

We ultimately designed a total of 18 chimeric receptor scaffolds, with 9 constructs generated for each family of sensors (Methods, Extended Data Fig. 5a,b and Supplementary Table 1). For VMR, these included both intuitive topologies, in which the domain order was preserved as found in natural receptors, as well as non-intuitive designs, where the domain arrangement and combination differed from VEGFR (for example, VMR_INT, VMR_D7D6). For CMR, synthetic linkers were designed with properties—such as length, structure and sequence—that differed from those of CSF1R. We initially explored an extensive space of de novo linker structures and sequences—examining more than 700,000 possibilities—using advanced deep learning methods, including ProteinMPNN and S4PRED, in combination with fragment assembly approaches (Methods). From this initial in silico screening, we found that the average helicity of the linkers—critical for dictating coupling properties—was generally low (Methods and Extended Data Fig. 5c,d). Our nine designed CMR chimeras combined a diverse range of these ProteinMPNN de novo sequences with a more targeted, structure-informed approach that integrated fragments from native TpoR and CSF1R sequences, leading to linkers with higher helicity and the design of CMR_FL, CMR_INT, CMR_SHORT and CMR_VeryShort. Except for CMR_{FL_MPNN_helix}, the only chimera from ProteinMPNN to feature helicity, the calculated couplings for the ProteinMPNN linkers fell below the thresholds required to effectively programme signalling responses in the CMR constructs (Extended Data Fig. 5c,d and Supplementary Table 1).

To conduct in-depth experimental validation, we selected six constructs representative of the dataset and the range of predicted outcomes. For the VMR chimeras, we selected (1) VMR_FL, which incorporates all native ectodomains (D1–7), and exhibited the most optimal predicted VEGF response while maintaining the lowest propensity for constitutive activity; (2) VMR_SHORT, the minimal version of the chimera, where the ligand-binding domain (D1–3) is directly linked to the TM region (it was chosen as a negative control, as it is predicted to exhibit a weak signalling response to VEGF binding); and (3) VMR_INT, a non-intuitive design where D4 is directly connected to D7 (D1–4 + D7), chosen for its intermediate properties. Our calculations predict a gradient of increasing dimerization and coupling properties from VMR_SHORT to VMR_INT to VMR_FL, indicating a progressive enhancement in the ability of VMR chimeric scaffolds to redirect VEGF sensing into potent c-MPL signalling (Methods, Fig. 1e and Extended Data Fig. 5). Among the CMR designs, we selected (1) CMR_INT, our top-ranked design in terms of signal transduction propensity; (2) CMR_FL, which provided a well-balanced compromise between constitutive and ligand-induced activities, aligning with our design goals; and (3) CMR_SHORT, as it exhibited one of the lowest dimerization propensities. Our calculations predicted CMR_FL to have the weakest coupling according to our activation model (Extended Data Fig. 3), and therefore should show the highest level of basal activity (Fig. 1e). All three CMR variants had strong dimerization propensity and were predicted to provide potent signalling responses to CSF1 sensing.

Molecular dynamics simulation of VMR_FL

To validate our design approach and the positive design hypothesis, we next characterized the impact of the ligand on the receptor structure and dynamics using molecular dynamics (MD) simulations (Fig. 2a). Since these calculations are very time-consuming, we selected the most optimal construct in terms of predicted dynamic response, VMR_FL, and carried out a large-scale simulation both with (total 0.75 μs) and without ligand (1 μs) (Methods). While these simulations are at least one order of magnitude too short to explore the entire receptor activation process⁴⁵, they revealed distinct conformational properties of the ligand-free and ligand-bound forms that aligned with the expected native behaviour of c-MPL^44,46. Principal component analysis (PCA) and subsequent K-means clustering (Methods) of both the receptor and TM coordinates revealed a unique space occupied by only the ligand-bound state (Fig. 2b–d). These conformations corresponded to the receptor adopting an ‘upright’ position, with the representative cluster centre showing an angle of 84° between the lipid membrane and the D2’s centre of mass. The remaining conformational spaces shared by both the ligand-bound and ligand-unbound tended towards lower angles of around 55° on average. Calculated mechanical coupling of these representative cluster centres correlated with these angles, with the upright position returning a much higher coupling score than the lower angle ligand-bound or unbound conformations. This finding aligns well with the consensus that inactive receptor tyrosine kinase conformations adopt a bent configuration⁴⁷, and indeed can form direct interactions with the membrane itself⁴⁸, before becoming upright when ligand bound (Fig. 2c,d). Overall, our results imply that only conformations accessible to the ligand-bound state can confer the necessary coupling required for a potent response, and that our coupling metric is sensitive enough to capture this behaviour.

**Fig. 2: Signalling state distribution of VMR_FL by MD simulations.**

Consistent with previous experimental findings on c-MPL, the ligand impacts also the conformation of the JM region that flanks the TM on the CT side of the membrane, participating in the receptor activation. For instance, a cation–π interaction between R514 and W515 known to stabilize a helical motif in the inactive state⁴⁶ is more often observed in the ligand-free simulations (Fig. 2e, light orange square). Conversely, Q516, known to stabilize the interface of the active state⁴⁴, is found with higher frequencies at the interface in the ligand-bound simulation (Fig. 2e,f, dark orange rectangles). Overall, while we have not modelled the entire activation process and our simulations have unlikely reached a true inactive state, our ligand-unbound simulations revealed several known inactive-like structural features. The shifts in coupling behaviours and TM–JM motif interactions between the ligand-bound and unbound simulations therefore validate our constructed model and strongly suggest that our assembly protocol can design reasonable receptors with predictable behaviours.

T-SenSER signal transduction in human T cells

To experimentally explore the computationally predicted signal transduction propensity of the designed VMR and CMR variants, we assessed baseline and VEGFA- or CSF1-dependent STAT5 phosphorylation as a surrogate for c-MPL signalling in human T cells transduced with VMR_SHORT, VMR_INT, VMR_FL or CMR_SHORT, CMR_INT, CMR_FL. We found that all variants were capable of transmitting signal upon VEGFA or CSF1 exposure. For VMR, VMR_FL produced the most efficient STAT5 phosphorylation followed by VMR_INT and VMR_SHORT that were characterized by significantly lower levels of %pSTAT5⁺ cells and lower peak mean fluorescence intensity (MFI) of pSTAT5 expression when compared with VMR_FL or IL15 positive control (Fig. 3a,b, gating strategy in Supplementary Fig. 1). Importantly, no spontaneous pathway activation was detected in any of the three VMR variants, indicating that VMRs are fully ligand inducible. To assess whether VEGFA triggers receptor dimerization or intramolecular reorganization of pre-formed dimers, we characterized the size of VMR_FL in transgenic T cells by western blot in the presence or absence of VEGFA. To facilitate detection of receptor oligomers, samples were treated with the crosslinking reagent BS3 (Methods). In the absence of VEGFA, we found VMR_FL as monomers, while in the presence of VEGFA and BS3, VMR_FL was detectable as ligand-bound oligomers (Fig. 3c, Supplementary Fig. 2 and original blot supplement file). These observations imply that VMR_FL signalling is at least partly controlled by VEGFA-induced dimerization. For CMR, CMR_FL had the strongest constitutive baseline activity in the absence of CSF1, while CMR_INT and CMR_SHORT had significantly lower baseline activity. In addition, all CMR variants were highly inducible in the presence of the ligand CSF1 (Fig. 3d,e, gating strategy in Supplementary Fig. 2).

**Fig. 3: T-SenSER variants signalling responses.**

Overall, the measured signal transductions are consistent with the intended designed properties. Higher coupling in VMRs locks the responder in the inactive monomeric state in the absence of ligand, hence lowering constitutive activity while promoting potent switching and activation upon ligand binding (Fig. 1f). The stronger coupling in VMR_FL results in ligand binding driving a higher proportion of receptors into the active state than in VMR_INT and VMR_SHORT. Owing to lower communication, the c-MPL responder in CMRs often occupies the active state and triggers constitutive activity (Fig. 1g). The lower communication in CMR_FL results in higher basal activity than in CMR_INT and CMR_SHORT while maintaining maximal activity in the presence of CSF1. In addition, within each family of sensors, we observed a linear relationship between the ligand-induced shifts in activity (that is, induced–constitutive) and the calculated coupling values (Fig. 3f,g and Supplementary Table 2). These findings indicate that our calculated coupling metrics are strong determinants of receptor signalling activity.

T-SenSER pathway activation and signalling thresholds

The first step in c-MPL signalling is activation of the JAK/STAT pathway, leading to phosphorylation of STAT5 and STAT3. In addition, c-MPL also activates the PI3K/AKT/mTOR axis as well as the MAPK/ERK1/2 pathway⁴⁹ (Fig. 4a). To gain deeper insight into pathway activation upon VMR_FL, CMR_FL or c-MPL activation in transgenic human T cells by the respective recombinant ligands (VEGFA, CSF1 and thrombopoietin (TPO)), we assessed phosphorylation of STAT5, STAT3, S6 and ERK1/2. We found that the c-MPL endo-domain used in VMR_FL and CMR_FL reliably transmitted signals activating all evaluated components, and that the profile of ligand-dependent pathway activation was comparable between VMR_FL, CMR_FL and c-MPL transgenic T cells (Fig. 4b–f, gating strategy in Supplementary Fig. 3).

**Fig. 4: Characterization of VMR_FL and CMR_FL phospho-signalling and ligand sensitivity.**

Both serum VEGF and CSF1 levels have been intensively studied across tumour histologies and are significantly higher in patients with cancer than in healthy individuals^50,51. Body compartment distribution of VEGFs and CSF1 is altered in cancer, indicating that T-SenSER T cells will likely encounter higher VEGF or CSF1 levels in malignant than in normal tissues, thereby mediating enhanced tumour specificity. To determine the activation threshold and the half maximal effective concentration (EC50) of both VMR_FL and CMR_FL in response to ligand, we quantified pSTAT5 levels in VMR_FL⁺ and CMR_FL⁺ T cells in response to increasing concentrations of VEGFA or CSF1 (Fig. 4g–j, gating strategy in Supplementary Fig. 4). The EC50 for VMR_FL was 144 pg ml⁻¹ (Fig. 4h), while the EC50 for CMR_FL was 315.8 pg ml⁻¹ (Fig. 4j).

To evaluate our strategy in different cancer models, we selected metastatic lung cancer and MM. Metastatic lung cancer is a solid tumour that is difficult to cure despite the introduction of immune checkpoint blockade therapy⁵², and CAR-T or TCR-T cells are still in early development. MM has approved CAR-T cell therapies, but current products and indications do not confer long-lasting remissions¹⁰, thus enhanced targeting moieties or combinations with novel approaches are highly warranted. To model the various VEGFA or CSF1 levels in mouse xenografts and compensate for the lack of a human TME as a source of human VEGFA and CSF1, we engineered A549.GFP-ffLuc cells with VEGFA and MM.1S.GFP-ffLuc cells with CSF1 overexpression (OE). In vitro, A549.GFP-ffLuc.VEGFA-wild-type (WT) cells produced low levels of VEGFA, contrary to A549.GFP-ffLuc.VEGFA-OE cells, which secreted VEGFA levels above the VMR_FL EC50 threshold (Supplementary Fig. 5a). Next, A549 cells were engrafted intravenously in immunocompromised NOD-SCID-γ-chain^−/− (NSG) mice and lung tissue levels of VEGFA were determined after 3 days (Fig. 4k,l). VEGFA levels in the lung remained below the EC50 for VMR_FL activation in A549.VEGFA-WT-engrafted mice (VEGFA_low model), while a median of 1.775 ng VEGFA per g total protein (range 1.725–3.061 ng g⁻¹, n = 5) was reached in lung tissues of A549.VEGFA-OE-engrafted mice, significantly above the EC50 for VMR_FL activation (VEGF_high model) (Fig. 4l). VEGFA tissue levels in the VEGF_high model were on average 126-fold lower than the levels reported in patients with lung cancer (median 224 ng VEGFA per g total protein, range 30–1,870, n = 71)⁵³. The VEGF_high model is therefore appropriate to evaluate CAR.VMR_FL⁺ T cell function in vivo but may underestimate VMR_FL potency owing to lower tissue levels in the animal model compared with patient tissues. In MM, MM.1S.GFP-ffLuc.B2MKO.CSF1-WT cells did not produce detectable CSF1 in vitro. However, high levels of human CSF1 were detected from engineered MM.1S.GFP-ffLuc.B2MKO.CSF1-OE cells, above the CMR_FL EC50 levels (Supplementary Fig. 5b). To quantify CSF1 levels in vivo, we engrafted both types of MM.1S cell intravenously in NSG mice and analysed BM lysates (Fig. 4m,n). In BM of mice engrafted with MM.1S.GFP-ffLuc.B2MKO.CSF1-OE, we detected CSF1 levels above the CMR_FL EC50 with a median of 3.998 ng ml⁻¹ (range 2.360–13.878 ng ml⁻¹, n = 5), while no CSF1 was detected in BM of MM.1S.GFP-ffLuc.B2MKO.CSF1-WT-engrafted mice (Fig. 4n). We termed our two models CSF1_high and CSF1_neg models respectively and used these to characterize CMR_FL + CAR-T cell function in vivo.

To maximize the impact of the designed T-SenSERs on CAR-T cell functions, we selected VMR_FL, the VMR construct with the highest signalling response to VEGF, and CMR_FL, the CMR construct combining the strong response to CSF1 with the highest basal activity to favour also constitutive homeostasis and enhanced effector function in the absence of cytokines. To evaluate T-SenSER activity in vivo in relation to the levels of ligand present in the TME, we established animal models with different levels of VEGFA or CSF1 that reflect the clinical situation of patients with lung cancer and myeloma.

VMR_FL effects in human CAR-T cells targeting lung cancer

VMR_FL was efficiently co-transduced in activated human T cells with conventional second-generation (28ζ, BBζ) or non-signalling control (Δ) CARs targeting the antigen ephrin A2 (EphA2). CAR expression levels were comparable between CAR and CAR.VMR_FL transduced cells, and VEGFA-induced STAT5 phosphorylation was comparable between VMR_FL and CAR.VMR_FL T cells⁵⁴ (Fig. 5a,b, gating strategy in Supplementary Fig. 6). The 4H5 single chain variable fragment recognizes a conformational epitope of EphA2 that is exposed on a wide variety of malignant but not on normal cells, including A549 lung cancer cells⁵⁵ (Supplementary Fig. 7). The impact of VMR_FL activation on tumour killing and T cell expansion was assessed in sequential co-cultures where T cells were repetitively challenged with fresh tumour cells ± VEGFA (Fig. 5c,d). Tumour killing and T cell expansion were quantified after each challenge. Full T cell activation with target cell killing and sustained T cell expansion occurred in the presence of both tumour cells and VEGF (Fig. 5e–g). Cytotoxicity was sustained in vitro even in T cells transduced with CAR alone (Fig. 5e), but T cell expansion was enhanced in the presence of VEGFA and VMR_FL signalling (Fig. 5f,g). A slight but not significant enhancement of both killing and T cell expansion with CAR.VMR_FL T cells was observed without exogenous VEGFA addition. This is probably due to low levels of VEGFA production by A549-WT cells used in the assay (Fig. 5e–g and Supplementary Fig. 5a). VMR_FL activity in the presence of VEGFA did not alter cytokine nor cytolytic granule production when compared with CAR alone and did not impact the T cell subset composition and differentiation status of T cells (Supplementary Figs. 8 and 9).

**Fig. 5: Effects of VEGFA-dependent VMR_FL activation on EphA2-CAR-T cell cytotoxicity, expansion and transcriptional profile targeting lung cancer.**

To assess the potential for long-term persistence of engineered T cells in the absence of EphA2⁺ tumour, we performed a 4-week homeostatic maintenance experiment. VMR_FL activation by VEGFA alone provided T cell homeostasis and survival. In line with our previous observations on transgenic c-MPL signalling in human T cells⁴⁰, overall expansion levels were significantly increased with VEGFA compared with media alone but remained below those of IL-2 control (Fig. 5h–j, gating strategy in Supplementary Fig. 10).

Lastly, we hypothesized that VMR_FL signalling is complementary to EphA2-CAR BBζ signalling since the pathways activated by the endo-domains are largely distinct. Thus, we assessed differential gene expression analysing global immune response signatures in BBζ.VMR_FL T cells after one in vitro tumour challenge ± VEGFA, and also compared BBζ with BBζ.VMR_FL T cells in the presence of VEGFA. We identified several highly differentially expressed genes that are associated with enhanced T cell co-stimulation (for example, CD80, TNFRSF8 and HLA class II molecules) or enhanced effector function (for example, GNLY) in cells with VMR_FL stimulation. VMR_FL activation also led to a reduction in expression of genes associated with T cell exhaustion (for example, CTLA4, LAG3 and TIGIT), or factors associated with immune suppression (for example, reduced transcription of NT5E, TGFB1, increased transcription of ADA) (Fig. 5k,l).

These results suggest that, as intended, VMR_FL activation delivered signals for CAR-T cell expansion and persistence, and involved transcriptional changes associated with co-stimulation, effector function and reduced exhaustion in combination with a BBζ CAR.

CMR_FL effects in human CAR-T cells targeting MM

CMR_FL was co-expressed efficiently in activated human T cells with A proliferation-inducing ligand (APRIL)-based CARs targeting two antigens expressed on MM: B cell maturation antigen (BCMA) and transmembrane activator and CAML interactor (TACI). Monomers of APRIL (m) were used as ligand-binding domains and CARs were expressed in conventional first- (mζ) or second-generation (m28ζ, mBBζ) format as previously described⁵⁶ (Fig. 6a,b and Supplementary Fig. 11a–c). CAR cell surface expression levels were slightly lower in CAR.CMR_FL compared with CAR-T cells, and STAT5 phosphorylation levels were slightly lower in CAR.CMR_FL compared with CMR_FL T cells (Supplementary Fig. 11d,e). The impact of CMR_FL activation on tumour killing and T cell expansion was assessed in sequential co-cultures (Fig. 6c,d) with two different MM cell lines expressing different target antigen levels (NCI-H929 (BCMA⁺⁺TACI⁻) and MM.1S (BCMA⁺TACI⁺)) (Supplementary Fig. 12). CMR_FL expression provided a significant advantage for sequential killing of NCI-H929 cells only in combination with the mBBζ CAR, while killing of MM.1S cells was enhanced with all three evaluated mAPRIL-based CARs (mζ, m28ζ and mBBζ). The CMR_FL constitutive baseline activity was sufficient to enhance killing, which was not further improved with the addition of CSF1 at the tested effector to target (E:T) ratio (Fig. 6e). CMR_FL also boosted T cell expansion in vitro that was significantly higher in all conditions except with the m28ζ CAR targeting NCI-H929 (Fig. 6f,g). CMR_FL activity did not alter cytokine nor granzyme production when compared with CAR alone (Supplementary Fig. 13). During sequential co-culture, the CD4/CD8 ratio changed with an enrichment in CD8⁺ T cells (Supplementary Fig. 14). The effector/memory differentiation status was mostly dictated by the type of CAR and not significantly impacted by CMR_FL activity (Supplementary Fig. 15). To assess the impact of CMR_FL activity on T cell exhaustion/activation phenotype, we assessed expression of LAG3, CTLA4, TIGIT, PD1 and TIM3 at baseline and after 5 tumour challenges and compared % positive populations and marker MFI on CAR⁺ and CAR.CMR_FL⁺ T cells. We found a trend to decreased expression of LAG3, CTLA4 and PD1 on mBBζ CAR.CMR_FL⁺CD4⁺ T cells, and increased TIM3 on mBBζ CAR.CMR_FL⁺ CD8⁺ T cells (Supplementary Fig. 16). Lastly, in the 4-week homeostatic maintenance experiment in the absence of tumour, we found that the CMR_FL constitutive baseline activity in T cells was sufficient to mediate significant low-level expansion and survival. Addition of CSF1 significantly enhanced expansion and survival of CMR_FL⁺ T cells, comparable to the levels of non-transduced (NT) controls supplemented with IL2. Adding IL2 to CMR_FL⁺ T cells significantly augmented the effects of the baseline constitutive activity, indicating that the combination of c-MPL signalling and native common γ-chain cytokine signals mediated by IL2 is complementary (Fig. 6h–j and Supplementary Fig. 17).

**Fig. 6: Effects of constitutive baseline and CSF1-dependent CMR_FL activation on mAPRIL CAR-T cell cytotoxicity and expansion targeting MM.**

These results demonstrate that the constitutive baseline activity of CMR_FL enhanced the homeostatic expansion capacity of engineered T cells and increased the in vitro sequential killing and expansion capacity of mAPRIL CAR-T cells in situations of high tumour loads.

In vivo potency of VMR_FL ⁺ CAR-T cells in lung cancer

We next evaluated the in vivo impact and VEGF dependence of VMR_FL function in EphA2-BBζ CAR-T cells. In the systemic VEGF_low model, NSG mice engrafted with A549.GFP-ffLuc.VEGFA-WT cells were treated with a single limiting dose of 1 × 10⁵ T cells. Bioluminescence imaging (BLI) revealed partial response to BBζ CAR-T cell therapy, and as expected, no impact of VMR_FL addition was detected (Extended Data Fig. 6). In the systemic VEGF_high model, mice were engrafted with A549.GFP-ffLuc.VEGFA-OE cells (Fig. 7a). Mice treated with Δ or BBζ CAR-T cells had rapidly progressive disease and reached the experimental endpoint within 10 days. By contrast, mice treated with BBζ.VMR_FL⁺ T cells mounted a potent anti-tumour response (Fig. 7b,c), associated with a significant reduction in VEGFA serum levels (Fig. 7d). Most importantly, the overall survival of mice treated with BBζ.VMR_FL⁺ T cells was significantly enhanced compared with BBζ CAR-T cell-treated mice (Fig. 7e).

**Fig. 7: VEGFA-dependent in vivo anti-tumour function of EphA2-BBζ.VMR_FL T cells in lung cancer.**

Next, we assessed whether a ligand-independent constitutively active c-MPL receptor mediated similarly potent effects in vivo as VMR_FL in VEGFA-rich tumours. We generated a retroviral vector to express the c-MPL_W515L mutant (Extended Data Fig. 7a) previously described as a constitutively active c-MPL variant in patients with myeloproliferative neoplasms^57,58. We first confirmed that c-MPL_W515L can be expressed as transgene in T cells and mediated spontaneous STAT5 phosphorylation (Extended Data Fig. 7b, gating strategy in Supplementary Fig. 18a–c). Then, we co-expressed c-MPL_W515L with the EphA2-BBζ CAR in T cells (Extended Data Fig. 7c, gating strategy in Supplementary Fig. 18d,e) to evaluate the impact of c-MPL_W515L on in vivo anti-tumour function of T cells in the VEGFA_high systemic lung cancer model, using a limiting T cell dose of 1 × 10⁵ cells per mouse (Extended Data Fig. 7d). BBζ.VMR_FL⁺ T cells mediated a significantly stronger anti-tumour response than BBζ.c-MPL_W515L⁺ or BBζ⁺ T cells. The activity of BBζ.c-MPL_W515L⁺ T cells was comparable to T cells expressing the BBζ CAR alone, indicating that c-MPL signalling in T cells is most beneficial when activated with a ligand-dependent tumour-specific T-SenSER such as VMR_FL (Extended Data Fig. 7e–g).

To evaluate whether VMR_FL also enhanced BBζ CAR-T cell function in a subcutaneous solid tumour model, we engrafted A549.GFP-ffLuc.VEGFA-OE cells subcutaneously in the flanks of NSG mice. After stable tumour engraftment, mice were treated with a limiting dose of 5 × 10⁵ T cells intravenously and anti-tumour activity was assessed by BLI. We found again a significantly better anti-tumour response in mice treated with BBζ.VMR_FL⁺ T cells with clearance of their tumours within 2 weeks, compared with BBζ CAR-T cells or untreated controls (Fig. 7f–i), confirming the results obtained with the systemic VEGF_high model.

Thus, EphA2-BBζ CAR-T cells equipped with the VMR_FL T-SenSER provided potent VEGFA-dependent in vivo anti-tumour activity in both systemic and subcutaneous lung cancer xenograft models and favoured tumour eradication.

In vivo potency of CMR_FL ⁺ CAR-T cells in MM

Lastly, we assessed both the low constitutive and CSF1-dependent CMR_FL-mediated enhancement of CAR-T cells targeting MM. In the CSF1_neg model, MM.1S.GFP-ffLuc.B2MKO.CSF1-WT-engrafted NSG mice were treated with a single dose of 5 × 10⁶ T cells using the mAPRIL CAR as a model system (Extended Data Fig. 8a, gating strategy in Supplementary Fig. 19a,b). Significant but transient anti-tumour activity was observed in the mBBζ but also in the mBBζ.CMR_FL treatment groups, with no benefit observed by the co-expression of CMR_FL (Extended Data Fig. 8b–e). Thus, unlike the in vitro results, the constitutive baseline activity of CMR_FL was not sufficient to enhance mAPRIL CAR-T cell potency in vivo. To test whether endogenous tissue levels of human CSF1 are sufficient to mediate enhanced anti-tumour activity to CAR.CMR_FL T cells, we used NSG-Quad mice that express transgenic human CSF1 in tissues as recipients (CSF1_endo model)⁵⁹. We systemically engrafted MM.1S.GFP-ffLuc.B2MKO.CSF1-WT cells, treated mice with fully human heavy-chain-only BCMA directed FHVH33-BBζ or FHVH33-BBζ.CMR_FL T cells, and followed tumour growth by BLI⁶⁰ (Fig. 8a, gating strategy in Supplementary Fig. 20). We found more potent anti-tumour responses with FHVH33-BBζ.CMR_FL than with FHVH33-BBζ T cells at limiting dose. Tumour progression was significantly delayed, and survival of mice prolonged in the FHVH33-BBζ.CMR_FL treatment group (Fig. 8b–e). Finally, in the CSF1_high model, co-expression of CMR_FL along with a CAR also significantly improved outcomes of mice, either when combined with the mBBζ or the FHVH33-BBζ CAR (Fig. 8f–l, gating strategy in Supplementary Fig. 21). We found significantly enhanced tumour control (Fig. 8h–j) and improved survival of mice (Fig. 8k,l) at limiting T cell doses.

**Fig. 8: CSF1-dependent in vivo anti-tumour function of APRIL-mBBζ.CMR_FL and FHVH33-BBζ.CMR_FL T cells in MM.**

Thus, our results demonstrate that the CMR_FL T-SenSER significantly enhanced anti-tumour activity of CAR-T cells against MM in a CSF1-dependent manner. Enhanced T cell potency was achieved both in response to physiological tissue levels of human CSF1 in NSG-Quad mice and in an engineered cell line model with CSF1 overexpression by tumour cells.

Discussion

We developed a computational method for the bottom-up assembly and design of receptors with programmable signalling activity. We applied our method to create T-SenSERs with predictable signalling responses to soluble factors enriched in TMEs and validated our assembly protocol through large-scale MD simulations. T-SenSER’s constitutive and ligand-induced signalling were computationally tuned to enhance the context-dependent potency of CAR-T cells. We demonstrate that both VEGFA and CSF1, enriched in TMEs of various tumour types, can be exploited as chemical cues for activating synthetic signalling through VMR_FL or CMR_FL expressed in CAR-T cells, respectively, and overcome the lack of endogenous co-stimulation and cytokine signalling in TMEs. Both VMR_FL and CMR_FL increased the potency of limiting doses of CAR-T cells in ligand-rich lung cancer and MM xenograft models. Our approach has the potential to significantly improve next generation engineered T cell therapies and provides a framework for the design of entirely novel cellular therapeutics for oncology, autoimmune and regenerative medicine applications.

Past decades have witnessed the development of computational methods for creating novel protein domain structures and binding interactions^61,62,63,64. However, protein allosteric functions such as signal transduction relying on long-range structural changes and dynamic communication between protein domains have largely been neglected and remain challenging to design. By leveraging fast protein structure and dynamics calculations, our computational platform provides a practical and efficient solution to the optimization of protein association and long-range mechanical coupling that govern signal transduction in single-pass multi-domain membrane receptors. On the basis of general biophysical principles, the approach is not limited to a particular scaffold architecture or molecular mechanism and can be applied to design a wide range of biosensor functions. De novo globular domain structures and ligand-binding domains can now be reliably designed using deep-learning-based approaches^{62,65,66,67,68,69} and inserted as building blocks into our assembly protocol to couple any desired cues to arbitrary cell signalling. As such, our strategy should have a significant and far-reaching impact in basic and synthetic biology.

Splitting the T cell activation signals into separately expressed transgenic receptors allows for context-dependent tuning of T cell activation and anti-tumour activity and simultaneously enhances the tumour specificity of engineered T cells^70,71,72. TME-derived IL4 or transforming growth factor β (TGF-β) have been used to elicit an immune stimulatory signal through empirically assembled chimeric receptors that leverage endo-domains from classical co-stimulatory (4-1BB) or cytokine receptors (IL7Rα)^73,74,75,76. An engineered autocrine feedback loop based on GM-CSF, produced from T cells upon antigen recognition by the CAR, with a signalling output linked to IL18, has been explored as a strategy to enhance anti-tumour responses of CAR-T cells⁷⁷. Constitutively active cytokine receptors have also been explored in combination with CARs, but those do not provide context-dependent functions^78,79. Despite sustained efforts to develop chimeric receptors exploiting TME-associated input signals to enhance CAR-T cell therapy outcome, only a few chimeras responding to soluble factors with demonstrated advantage in preclinical studies have been reported so far^{18,73,74,75,76}. In contrast to these traditional empirical domain swapping chimeric designs, our approach highlights the benefits of in silico pre-screening, creating non-intuitive receptor scaffolds that would otherwise never be identified. Our approach also avoids extensive experimental screens, focusing experimental validation on selecting the best combinations of ligand binding, constitutive activity and dynamic signalling responses. Given the numerous factors affecting multi-domain signalling receptor structure and function, computational approaches like ours can accelerate the engineering of receptors with customized functions, addressing a critical challenge in synthetic biology.

On the basis of rational design principles of signal transduction, our technology can engineer synthetic chimeric receptor structures with predictable and desired signalling output and sets the stage for the broader and more efficient development of biosensors with novel input–output functions. We validated our approach by building two distinct classes of chimeric signalling receptors that have only the TM and CT regions in common. In fact, the EC parts of VEGFR2 and CSF1R share very little sequence and structure homology and have no binding motifs in common. For each of these two classes, we have created several variants with rationally tuned constitutive and ligand-induced activities by programming the two most important features of receptor engineering, that is, the combination of EC domains and the type of inter-domain linkers. Therefore, our study constitutes a strong proof of concept of the rational bottom-up assembly and design of chimeric receptors and supports the generalizability of the engineering approach. Since multiple methods can now confidently model a large variety of TM sequences and structures^{65,80,81,82,83,84,85}, our approach could also be applied to the design of chimera involving alternative TM and CT sequences. Lastly, our approach is not restricted to specific natural protein domains but could, in principle, assemble fully synthetic components, provided that their structure can be reliably predicted.

The potential field of application of T-SenSERs is large. Here we performed experimental characterization when T-SenSERs are co-expressed with CARs in T cells. However, we envision future applications of VMR or CMR in other therapeutic T cell products, such as CAR-T cells with other endo-domains, TCR-T cells, tumour-infiltrating lymphocytes or virus-specific T cells, where significant room for improvement exists, and TME-specific enhancements of the anti-tumour response could be beneficial. Beyond αβ-T cells as cellular therapy platform, other cell types should also be investigated, for example, γδ-T cells, natural killer (NK) cells or cytokine-induced killer cells, that are amenable to off-the-shelf allogeneic cellular therapy development. The versatility of our computational platform will also allow the efficient adaptation of T-SenSER signalling domains to other cellular contexts, and to the exploitation of other disease-specific inputs when envisioning applications in the autoimmune disease or regenerative medicine fields.

In conclusion, we have developed an in silico method for the assembly and design of synthetic receptors with programmable signalling activity for therapeutic applications. We demonstrate that these novel synthetic receptors can be tuned to enhance CAR-T cell functions by providing weak constitutive and/or strong ligand-specific co-stimulatory and cytokine signals upon sensing soluble factors enriched TMEs. We fully characterized two selected T-SenSERs in two completely independent experimental systems in combination with various CARs targeting both solid tumours and haematologic malignancies. In principle, T-SenSERs can be expressed in any engineered cell therapy for cancer or other diseases, in various immune cell types, activating pathways that favour long-term persistence and function of immune effector cells that directly eliminate or support the elimination of cancer cells. In addition to soluble factors, our technology can also be broadly applied to the sensing of chemical or mechanical cues such as cell surface ligands, EC matrix components or metabolites. Overall, the ability to engineer and control signal transduction should impact basic and translational cell biology, synthetic biology and biomedicine.

Methods

Computational methods

A detailed description of all the computational methods and protocols for receptor modelling and design is available in the Supplementary Information file.

Cell lines

MM.1S (ATCC CRL-2974) and HEK 293T/17 (ATCC CRL-11268) cell lines were purchased from the American Type Culture Collection (ATCC) and maintained in complete DMEM or IMDM medium (Hyclone, Thermo Scientific, Gibco) with 10% fetal bovine serum (FBS) (Hyclone or Gibco), 1% GlutaMAX (Gibco) and 1% penicillin/streptomycin (Gibco). NCI-H929 cells were purchased from the German Cell Culture Collection (DSMZ, number ACC 163). The PG-13 retroviral producer cell line for the generation of retroviral particles encoding green fluorescent protein (GFP) and firefly luciferase (ffLuc) (GFP-ffLuc), as well as A549 and A549.GFP-ffLuc cells, was kindly provided by S. Gottschalk, Baylor College of Medicine⁸⁶. Purity of A549.GFP-ffLuc cells was verified by flow cytometry. A549.GFP-FFLuc VEGF-A₁₆₅ knock-out (VEGF-KO) cells were generated using Clustered Regularly Interspaced Short Palindromic Repeats/CRISPR associated protein 9 (CRISPR/Cas9)⁸⁷. Guide RNAs were designed with the CRISPRscan algorithm (CRISPRscan.com)⁸⁸, and four guide RNA sequences were selected to generate the KO cell line. Guide RNAs were synthesized by in vitro transcription (NEB, HiScribe T7 High Yield RNA Synthesis Kit, E2040S). A total of 1 μg of four VEGF-A₁₆₅ single-guide RNAs (sgRNA, 5′-GCCCCTGATGCGATGCGG-3′, 5′-GAGCCGTGGTCCGCGCGG-3′, 5′-CGGGCTCATGGACGGGTG-3′ and 5′-GATGTTGGACTCCTCAGT-3′) were mixed with 1 μg recombinant Cas9 protein (CP01, PNA Bio) at room temperature and used to electroporate 0.25 × 10⁶ A549 cells (2 pulses of 1,200 V for 30 ms, Neon Transfection System, Invitrogen). Electroporated cells were expanded in antibiotic-free DMEM media containing 10% FBS. Single-cell clones were generated by limiting dilution. Each clone was analysed for VEGF-A₁₆₅ secretion by enzyme-linked immunosorbent assay (ELISA) (R&D Systems, DY293B), and a clone with the lowest VEGF-A₁₆₅ production was selected. A549.GFP-ffLuc cells with overexpression of human VEGF-A₁₆₅ (VEGFA) (A549.GFP-ffLuc.VEGFA-OE) were generated by retroviral transduction with a vector encoding for human VEGFA and truncated human CD271 (ΔCD271) for detection and selection of transgenic cells. Transduced A549 cells were stained for CD271-PE (clone C40-1457, BD Biosciences) and CD271-positive cells were sorted by fluorescence activated cell sorting (FACS) (SH800, Sony Biotechnology) to over 98% purity. MM.1S and NCI-H929 cells were retrovirally transduced with GFP-ffLuc retroviral supernatant and FACS sorted (FACSAria IIu, BD Biosciences) for >98% purity. For the generation of the β2-microglobulin knock-out (B2M-KO) MM.1S cell line, cells were transduced with a lentiviral vector (lentiCRISPR v2, #52961 Addgene) coding for Cas9 protein, a single-guide RNA targeting B2M (5′-GGCCACGGAGCGAGACAUCU-3′) and a puromycin resistance gene. Transduced cells were selectively expanded in puromycin (ant-pr, InvivoGen) containing media, stained with HLA-A,B,C-PE (clone W6/32, Biolegend) and FACS sorted for HLA-A,B,C negativity, to a purity of over 98%. MM.1S.GFP-ffLuc.B2MKO cells with overexpression of human CSF1 (MM.1S.GFP-ffLuc.B2MKO.CSF1-OE) were generated by retroviral transduction with a vector encoding for human CSF1 and ΔCD271 for detection and selection of the transgene. Transduced cells were positively selected with an EasySep CD271 selection kit (Stemcell Technologies) to over 98% purity.

Peripheral blood mononuclear cells

Buffy coats of de-identified healthy volunteer blood donors were purchased from the Gulf Coast Regional Blood Center or at the Center of Interregional Blood Transfusion SRK Bern. Additional peripheral blood mononuclear cells (PBMCs) from those same donors are not available.

Generation of retroviral vectors and supernatant

To generate a retroviral vector for human VEGF-A₁₆₅ overexpression, human VEGF-A₁₆₅ cDNA (HG11066-G, Sino Biological) was cloned into the SFG retroviral vector backbone followed by an IRES sequence and a ΔCD271 selectable marker gene (In-Fusion HD Cloning Kit, Clontech). To generate the VEGFR2-MPL receptor (VMR) constructs, the protein sequence of the human VEGFR2 was obtained from the Uniprot database (P35968.1, https://www.uniprot.org/uniprotkb/P35968/entry). The natural signal peptide (SP) sequence was predicted using the Signal-IP4.1 server (http://www.cbs.dtu.dk/services/SignalP/) and replaced with an SP sequence derived from a human immunoglobulin as previously described³⁹. A myc-tag sequence (EQKLISEEDL) was added at the VEGFR2 protein N-terminus for detection and selection. Codon optimization for expression in human cells and gene synthesis were performed by Geneart (ThermoFisher). Three different VMR versions were generated by PCR cloning (In-Fusion HD Cloning Kit, Clontech) using the VEGFR2 and c-MPL plasmids as templates⁴⁰. The different chimeras were inserted into the SFG retroviral vector backbone. CAR constructs targeting the ephrin type A receptor 2 (EphA2) antigen were generously provided by S. Gottschalk, Baylor College of Medicine, and their generation and functional validation were previously described⁸⁹. Transient retroviral supernatant for all constructs was prepared by transfection of 293T cells as described³⁹. To generate a retroviral vector for human CSF1 overexpression, human CSF1 cDNA (#86797, Addgene) was cloned into the SFG retroviral vector backbone followed by an IRES sequence and a ΔCD271 selectable marker gene (In-Fusion HD Cloning Kit, Clontech). To generate the CSF1R-MPL receptor (CMR) constructs, the EC domain of the human CSF1R (HG10161-M, Sino Biological) was ligated with the TM and IC domain of c-MPL sequence and cloned into the SFG retroviral vector backbone followed by an IRES sequence and a ΔCD271 selectable marker gene (In-Fusion HD Cloning Kit, Clontech). The natural SP sequence was replaced with an SP sequence derived from a human immunoglobulin as described for VMR. Three different CMR versions were generated by PCR cloning (In-Fusion HD Cloning Kit, Takara Bio) using the CSF1R and c-MPL plasmids as templates. The different chimeras were inserted into the SFG retroviral vector backbone. CAR constructs used were the fully human heavy-chain only (FHVH33) targeting BCMA or APRIL-based CAR (APRIL-CAR) dual-targeting BCMA and TACI^56,60. APRIL-CARs target myeloma antigens BCMA and TACI and are designed based on a truncated monomeric APRIL unit (m) and varying signalling domains, including ζ (mζ), 4-1BBζ (mBBζ) and CD28ζ (m28ζ). Combinations of CMR and different APRIL-CARs (mζ.CMR, mBBζ.CMR and m28ζ.CMR) or FHVH33-CAR (FHVH33.BBζ.CMR) were cloned into the SFG retroviral vector as a polycistronic construct separated by a 2A sequence. The full-length c-MPL construct was previously described⁴⁰. To generate the constitutive c-MPL_W515L variant, a single W515L point mutation was introduced in the full-length c-MPL via PCR amplification. Transient retroviral supernatant for all constructs was prepared by transfection of 293T cells as described.

Generation of transgenic T cells

PBMCs were collected from healthy donor buffy coats using density gradient centrifugation by Lymphoprep (Accurate Chemical and Scientific Corporation or Serumwerk). Total PBMCs were activated in non-tissue culture-treated 24-well plates (Corning) coated with OKT3 1 µg ml⁻¹ (purified from hybridoma CRL-8001, ATCC, or purchased from Biolegend) and anti-CD28 antibody 1 µg ml⁻¹ (BD Biosciences or Biolegend), and IL2 (100 U ml⁻¹) or IL7 and IL15 (10 ng ml⁻¹ each, Miltenyi Biotec) for 3 days, and transduced on RetroNectin (Takara Bio) coated non-tissue culture-treated 24-well plates. After 48–72 h, T cells were collected and further expanded in T cell media (1:1 mixture of RPMI 1640 and Click’s media, Hyclone, or RPMI 1640 alone) supplemented with 10% FBS, 1% penicillin–streptomycin, 1% GlutaMAX, and IL2 (50 U ml⁻¹) or IL7 and IL15 (10 ng ml⁻¹ each). To generate T cells transduced with both CAR and VMR vectors, activated PBMCs were first transduced with CAR supernatant for 24 h, followed by a separate transduction with VMR supernatant for an additional 24–48 h. After the second transduction, T cells were collected and expanded for 7–10 days in fresh T cell media with IL2 (50 U ml⁻¹) or IL7 and IL15 (10 ng ml⁻¹ each) and were fed every 3–4 days with fresh media and cytokines. To generate T cells transduced with both CAR and c-MPL_W515L vectors, activated PBMCs were transduced with a 1:1 mix of CAR and c-MPL_W515L supernatant for 72 h. After transduction, T cells were collected and expanded for 7–10 days in fresh T cell media with IL7 and IL15 (10 ng ml⁻¹ each) and were fed every 3–4 days with fresh media and cytokines. If required, transduced T cells were positively selected with an EasySep CD271 selection kit (Stemcell Technologies), an EasySep Biotin selection kit after staining with anti-c-myc-biotin antibody (clone 9E10.3, Invitrogen) or an EasySep PE selection kit after staining with CD110-PE (clone 1.6.1, BD Biosciences).

Immunophenotyping

To assess transduction efficiencies, T cells were stained with VEGFR2 (CD309)-PE (R&D Systems), CD19-APC (clone H1B19) and 7AAD (BD Biosciences) or CD115-PE (clone 9-4D2-1E4, Biolegend), CD271-APC (clone C40-1457, BD Biosciences, or clone ME20.4, Biolegend), CD110-PE (clone 1.6.1, BD Biosciences) and DAPI (Biolegend). mAPRIL CAR cell surface staining was performed with a primary biotinylated antibody against APRIL (clone 53E11, Biolegend, validated for ELISA but not for FACS) followed by Streptavidin-APC (Biolegend). To assess T cell subset distribution and differentiation phenotypes, we used 7-AAD (BD Biosciences), CD4-Krome Orange (clone 13B8.2, Beckman Coulter), CD8-Pacific Blue (clone B9.11, Beckman Coulter), CD45RA-APC (clone HI100 BD, Biosciences), CD45RO-PE (clone UCHL1, BD Biosciences), CD62L-ECD (clone DREG56, Beckman Coulter) and CCR7-V450 (clone 150503, BD Biosciences) or DAPI (Biolegend), CD4-PE-Cy7 (clone OKT4, Biolegend), CD8-BV605 (clone RPA-T8, Biolegend), CD45RA-PerCPCy5 (clone HI100, Biolegend), CD45RO-APC-Cy7 (clone UCHL1, Biolegend), CD62L-BV650 (clone DREG-56, Biolegend) and CCR7-FITC (clone 150503, Biolegend). To assess T cell exhaustion, cells were stained with Zombie UV Fixable Viability Dye (Biolegend), CD3-BV711 (clone SK7, Biolegend), CD4-PE-Cy7 (clone OKT4, Biolegend), CD8-BV605 (clone RPA-T8, Biolegend), CD271-BV786 (clone C40-1457, BD Biosciences), CD45RO-APC-Cy7 (clone UCHL1, Biolegend), CD62L-BV650 (clone DREG-56, Biolegend), LAG3-APC (clone 7H2C65, Biolegend), PD-1-AF700 (EH12.2H7, Biolegend), TIM3-PE (clone A18087E, Biolegend), TIGIT-BV421 (clone A15153G, Biolegend) and CTLA4-PerCP-Cy5.5 (clone L3D10, Biolegend). For analysis of phospho-proteins at baseline or in response to VEGFA, CSF1 or TPO, T cells were rested in cytokine-free medium overnight and then stimulated with 25 ng ml⁻¹ VEGFA (or indicated concentration), 10 ng ml⁻¹ CSF1 (or indicated concentration), 25 ng ml⁻¹ TPO (or indicated concentration), or IL15 10 ng ml⁻¹ as a positive control, for 30 min to 1 h for pSTAT3, pSTAT5 and pS6 and 5 min for pERK. Cells were fixed with Cytofix (BD Biosciences), permeabilized with Perm Buffer III (BD Biosciences), washed thoroughly and stained with STAT5-PE or -AF647 (pY694) (clone 47/Stat5, BD Biosciences), STAT3-PE (pY705) (clone 4/P-STAT3, BD Biosciences), S6-PE (Ser235/236) (clone D57.2.2E, Cell Signaling) and ERK1/2-PE (Thr202/Tyr204) (clone 6B8B69, Biolegend).

Data acquisition was performed on a Beckman Coulter Gallios using Kaluza software version 2.1, or on a FACSCanto or SORP-LSRII with BD FACSDiva version 8.0.1 software. Data analysis was performed with FlowJo software version 10.8.2 (Tree Star).

Sequential co-culture assay

A549.GFP-ffLuc cells were co-cultured with T cells in 6 replicates at an E:T ratio of 1:1 (75,000 cells each in a 48-well plate) in the presence or absence of 25 ng ml⁻¹ VEGFA. Transgenic T cells were enriched by CD19-positive selection with beads (Miltenyi, 130-050-301), and the number of T cells used in functional assays was based on the total number of CAR transgenic cells in both CAR and CAR.VMR conditions. For cytokine analysis, culture supernatants were collected 24 h after initial plating or rechallenge of the co-culture. Tumour cells and T cells in co-culture were quantified every 3–4 days by flow cytometry. T cells were stained with CD3-APC (BD Biosciences), tumour cells were identified by GFP expression, and 7AAD-negative live cells were quantified with counting beads (CountBright Beads, Life Technologies). Residual T cells were challenged with fresh A549.GFP-ffLuc tumour cells (75,000 cells per well) in each replicate well when >90% of tumour cells were killed at the analysis timepoint; otherwise, the killing was considered as incomplete and T cells were not rechallenged. NCI-H929.GFP-ffLuc and MM.1S.GFP-ffLuc cells were co-cultured with T cells in 6 replicates at an E:T ratio of 1:10 (50,000 T cells and 500,000 target cells per well in a 48-well plate) in the presence or absence of 10 ng ml⁻¹ CSF1. The number of T cells used in functional assays was adjusted for the number of CAR transgenic cells in both CAR and CAR.CMR conditions. For cytokine analysis, culture supernatants were collected 24 h after initial plating of the co-culture. Tumour cells and T cells in co-culture were quantified every 3–4 days by flow cytometry. T cells were stained with CD3-BV711 (clone SK7, Biolegend), tumour cells were identified by GFP expression, and DAPI-negative live cells were quantified with counting beads (CountBright Beads, Life Technologies). Residual T cells were challenged with fresh target cells (500,000 cells per well) in each replicate well when >60% of tumour cells were killed at the analysis timepoint; otherwise, the killing was considered as incomplete and T cells were not rechallenged.

Multiplex cytokine detection

Co-culture supernatants were analysed for cytokines and cytotoxic granules with the MILLIPLEX human CD8⁺ T cell magnetic bead panel (EMD Millipore) and the Luminex 200 instrument (Luminex) or with the or CAR-T cell combo 1 kit (Mesoscale Discovery) and analysed on a MESO QuickPlex SQ 120 (Mesoscale Discovery). Data were analysed on the Mesoscale Discovery Workbench 4.0 and graphed with GraphPad Prism 8.1.2 or higher.

Detection of VEGFA or CSF1

Human VEGFA levels were analysed in culture supernatant by ELISA (Human VEGF Duoset ELISA Kit DY293B, R&D Systems) or MSD Mesoscale (U-Plex human VEGFA assay kit), and in mouse serum and lung tumour tissue lysates by MSD Mesoscale. Human CSF1 levels were analysed in culture supernatant by MSD Mesoscale (U-Plex human M-CSF assay kit), and in mouse bone marrow lysates by MSD Mesoscale. To determine VEGFA or CSF1 production in culture, A549 or MM.1S cells were plated in 48-well plates at a density of 2 × 10⁵ cells per well per ml. Culture supernatants were collected 6 h, 24 h and 48 h later to quantify VEGFA or CSF1 production. Mouse serum samples were analysed at a 1:5 dilution. Protein concentration of 50 mg tissue lysate was determined with the Pierce BCA Protein Assay Kit (ThermoFisher Scientific). Data were analysed on the Mesoscale Discovery Workbench 4.0 and graphed with GraphPad Prism.

Western blotting

For detection of pERK1/2, 5 × 10⁶ VMR_FL⁺ or CMR_FL⁺ T cells were rested in cytokine-free medium overnight and then stimulated with 25 ng ml⁻¹ VEGFA or 10 ng ml⁻¹ CSF1 for 5 min or 10 min. Cells were subsequently lysed in RIPA buffer supplemented with Halt phosphate/protease inhibitors. Cell lysate was boiled at 97 °C for 10 min in Bolt LDS sample buffer and reducing agent (ThermoFisher Scientific). Lysates were separated with Bolt 4–12% Bis-Tris Plus gels (Invitrogen) and then transferred on iBlot 2 PVDF Mini Stacks nitrocellulose membranes (Invitrogen) using an iBlot transfer system (ThermoFisher Scientific). After protein transfer, nitrocellulose membranes were incubated for 1 h in 5% nonfat milk in TBST (TBS, pH 7.4 + 0.1% Tween 20) at room temperature. After blocking, nitrocellulose membranes were incubated overnight with antibodies against total ERK1/2 (clone 137F5, Cell Signaling) apERK1/2 (clone 197G2, Cell Signaling) at 4 °C and incubated for 2 h with HRP-conjugated anti-rabbit IgG (ThermoFisher Scientific). After substrate addition (SuperSignal West Femto Maximum Sensitivity Substrate, ThermoFisher Scientific), images were acquired with a western blot imager (Fusion, Vilber Lourmat).

For analysis of VMR oligomerization, 5 × 10⁶ VMR_FL⁺ T cells were rested in cytokine-free medium overnight and then stimulated with 25 ng ml⁻¹ VEGFA for 20 min. Cells were either lysed directly in RIPA buffer supplemented with Halt phosphate/protease inhibitors (ThermoFisher Scientific) or incubated with 3 mM, 1.5 mM or 0.75 mM bis(sulfosuccinimidyl)suberate (BS3) before lysis to induce crosslinking of surface proteins⁹⁰. Cell lysate was boiled at 97 °C for 10 min in Bolt LDS sample buffer and reducing agent (ThermoFisher Scientific). Lysates were separated with Bolt 4–12% Bis-Tris Plus gels (Invitrogen) and then transferred on iBlot 2 PVDF Mini Stacks nitrocellulose membranes (Invitrogen) using an iBlot transfer system (ThermoFisher Scientific). After protein transfer, nitrocellulose membranes were incubated for 1 h in 5% nonfat milk in TBST (TBS, pH 7.4 + 0.1% Tween 20) at room temperature. After blocking, nitrocellulose membranes were incubated overnight with anti-c-myc-biotin antibody (clone 9E10.3, Invitrogen) or anti-GAPDH (clone 0411, Santa Cruz Biotechnology) at 4 °C and incubated for 2 h with m-IgGκ BP-HRP (Santa Cruz Biotechnology). After substrate addition (SuperSignal West Femto Maximum Sensitivity Substrate, ThermoFisher Scientific), images were acquired with a western blot imager (Fusion, Vilber Lourmat).

Gene expression by NanoString nCounter

Activated PBMCs from 5 healthy donors were transduced with EphA2-BBζ CAR alone or double transduced with EphA2-BBζ-CAR and VMR_FL. NT T cells from each donor were expanded as controls. Transgenic T cells were purified by FACS or magnetic bead sorting. EphA2-BBζ.VMR_FL⁺ T cells were stained with VEGFR2-PE and CD19-APC, and double-positive T cells were FACS sorted (SH800, Sony Biotechnology) to >94% purity. EphA2-BBζ CAR⁺ T cells were selected with CD19 magnetic microbeads (Miltenyi Biotec) to >95% purity. Sorted T cells were co-cultured with A549.GFP-ffLuc.VEGFA KO cells (E:T ratio 1:1) in 24-well plates in the presence or absence of VEGFA (25 ng ml⁻¹). After 3–4 days, live T cells and tumour cells were quantified by FACS. In cultures with complete tumour cell killing, live T cells were purified by Dead Cell Removal Kit (Miltenyi Biotec), while T cells from cultures with residual tumour cells (>1%) were purified with a mixture of CD4 and CD8 magnetic beads and positive selection on LS columns (Miltenyi Biotec). From each sample, RNA was purified with the RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. Gene expression from a total of 36 RNA samples, from 9 experimental groups and 4 individual donors each were analysed with the Immunology V2 Panel (NanoString). RNA quality control, NanoString nCounter sample processing and data analysis were performed at the Genomic Technology Facility at the Center for Integrative Genomics at the University of Lausanne, Switzerland. Data normalization was performed in nSolver Analysis Software 4.0, using total counts per lane as normalization factor. Data were log₂ transformed and hierarchical clustering and principal component plots were generated in R (version 3.4.4). Differential gene expression analysis was performed using the R Bioconductor package limma, using the limma trend approach. For pairwise comparisons of experimental conditions, moderated t-tests with paired samples were used, taking into account the sample donor. For each comparison, a separate linear model was fitted to the subset of data belonging to the two conditions. P values were adjusted for multiple testing correction by the Benjamini–Hochberg method, separately for each comparison.

Mouse xenograft experiments

All animal studies were conducted in accordance with a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of Baylor College of Medicine or according to Swiss federal regulations and approved by the veterinary authority of Canton Vaud (Authorization VD3390). Female NOD-SCID-γc^−/− (NSG) mice (4–6 weeks old) were purchased from the Jackson Laboratory (JAX:005557) and housed at the Baylor College of Medicine Animal Facility or obtained from the shared animal facility of the University of Lausanne (UNIL). NOD-SCID-γc^−/− Tg(CMV-IL3,CSF2,KITLG)1Eav Tg(CSF1)3Sz/J (NSG-Quad) mice were purchased from the Jackson Laboratory (JAX:028657)⁵⁹, housed, bred, maintained and obtained from the shared animal facility of UNIL. Female mice were 6–10 weeks old at the start of the experiments and mice were randomized into comparable groups according to their body weight and level of tumour engraftment on the day of adoptive T cell transfer. Details for each type of experiment are depicted in the associated figures and described in the figure legends as well as the Supplementary Information file.

Statistics

Data were summarized using descriptive statistics. For continuous variables, comparisons were made by t-test. EC50 values were calculated upon log transformation of VEGFA or CSF1 concentrations and normalization of dose response, followed by a nonlinear fit. Area under the curve (AUC) comparisons were analysed with unpaired t-test (two-sided) and Welch’s correction when appropriate. Survival of mice was analysed by the Kaplan–Meier method and significance assessed with log-rank test. Analyses were performed with GraphPad Prism version 8.1.2 or higher. Significance levels reported in figures use the following code: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; NS, not significant. Actual P values for the key comparisons are indicated in the respective figure legends.

Other software

Schematic drawings were done with Inkscape version 1.4, bioRender (institutional license to EPFL School of Life Sciences), PowerPoint or Adobe Illustrator 2022. Molecular visualization was done with Visual Molecular Dynamics (VMD) v1.9.3.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Code availability

The modelling, design and Dimeric MultiDomain Biosensor Builder software developed in this study, together with a detailed Readme for running the simulations, are available in the following GitHub repository⁹¹: https://github.com/barth-lab/Dimeric_MultiDomain_Biosensor_Builder.

References

Labanieh, L. & Mackall, C. L. CAR immune cells: design principles, resistance and the next generation. Nature 614, 635–648 (2023).
Article CAS PubMed Google Scholar
Lim, W. A. & June, C. H. The principles of engineering immune cells to treat cancer. Cell 168, 724–740 (2017).
Article CAS PubMed PubMed Central Google Scholar
Mulvey, A., Trueb, L., Coukos, G. & Arber, C. Novel strategies to manage CAR-T cell toxicity. Nat. Rev. Drug Discov. 24, 379–397 (2025).
Article CAS PubMed Google Scholar
Neelapu, S. S. et al. Axicabtagene ciloleucel CAR T-cell therapy in refractory large B-cell lymphoma. N. Engl. J. Med. 377, 2531–2544 (2017).
Article CAS PubMed PubMed Central Google Scholar
Maude, S. L. et al. Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia. N. Engl. J. Med. 378, 439–448 (2018).
Article CAS PubMed PubMed Central Google Scholar
Schuster, S. J. et al. Tisagenlecleucel in adult relapsed or refractory diffuse large B-cell lymphoma. N. Engl. J. Med. 380, 45–56 (2019).
Article CAS PubMed Google Scholar
Abramson, J. S. et al. Lisocabtagene maraleucel for patients with relapsed or refractory large B-cell lymphomas (TRANSCEND NHL 001): a multicentre seamless design study. Lancet 396, 839–852 (2020).
Article PubMed Google Scholar
Wang, M. et al. KTE-X19 CAR T-cell therapy in relapsed or refractory mantle-cell lymphoma. N. Engl. J. Med. 382, 1331–1342 (2020).
Article CAS PubMed PubMed Central Google Scholar
Raje, N. et al. Anti-BCMA CAR T-cell therapy bb2121 in relapsed or refractory multiple myeloma. N. Engl. J. Med. 380, 1726–1737 (2019).
Article CAS PubMed PubMed Central Google Scholar
Munshi, N. C. et al. Idecabtagene vicleucel in relapsed and refractory multiple myeloma. N. Engl. J. Med. 384, 705–716 (2021).
Article CAS PubMed Google Scholar
Berdeja, J. G. et al. Ciltacabtagene autoleucel, a B-cell maturation antigen-directed chimeric antigen receptor T-cell therapy in patients with relapsed or refractory multiple myeloma (CARTITUDE-1): a phase 1b/2 open-label study. Lancet 398, 314–324 (2021).
Article CAS PubMed Google Scholar
Rafiq, S., Hackett, C. S. & Brentjens, R. J. Engineering strategies to overcome the current roadblocks in CAR T cell therapy. Nat. Rev. Clin. Oncol. 17, 147–167 (2020).
Article PubMed Google Scholar
Wagner, J., Wickman, E., DeRenzo, C. & Gottschalk, S. CAR T cell therapy for solid tumors: bright future or dark reality? Mol. Ther. 28, 2320–2339 (2020).
Article CAS PubMed PubMed Central Google Scholar
D’Angelo, S. P. et al. Afamitresgene autoleucel for advanced synovial sarcoma and myxoid round cell liposarcoma (SPEARHEAD-1): an international, open-label, phase 2 trial. Lancet 403, 1460–1471 (2024).
Article PubMed PubMed Central Google Scholar
Chen, L. & Flies, D. B. Molecular mechanisms of T cell co-stimulation and co-inhibition. Nat. Rev. Immunol. 13, 227–242 (2013).
Article PubMed PubMed Central Google Scholar
Speiser, D. E., Ho, P. C. & Verdeil, G. Regulatory circuits of T cell function in cancer. Nat. Rev. Immunol. 16, 599–611 (2016).
Article CAS PubMed Google Scholar
Manhas, J., Edelstein, H. I., Leonard, J. N. & Morsut, L. The evolution of synthetic receptor systems. Nat. Chem. Biol. 18, 244–255 (2022).
Article CAS PubMed PubMed Central Google Scholar
Scheller, L., Strittmatter, T., Fuchs, D., Bojar, D. & Fussenegger, M. Generalized extracellular molecule sensor platform for programming cellular behavior. Nat. Chem. Biol. 14, 723–729 (2018).
Article CAS PubMed Google Scholar
Divine, R. et al. Designed proteins assemble antibodies into modular nanocages. Science 372, eabd9994 (2021).
Article CAS PubMed PubMed Central Google Scholar
Quijano-Rubio, A. et al. De novo design of modular and tunable protein biosensors. Nature 591, 482–487 (2021).
Article CAS PubMed PubMed Central Google Scholar
Giordano-Attianese, G. et al. A computationally designed chimeric antigen receptor provides a small-molecule safety switch for T-cell therapy. Nat. Biotechnol. 38, 426–432 (2020).
Article CAS PubMed Google Scholar
Ben-Sasson, A. J. et al. Design of biologically active binary protein 2D materials. Nature 589, 468–473 (2021).
Article CAS PubMed PubMed Central Google Scholar
Cao, L. et al. De novo design of picomolar SARS-CoV-2 miniprotein inhibitors. Science 370, 426–431 (2020).
Article CAS PubMed PubMed Central Google Scholar
Vorobieva, A. A. et al. De novo design of transmembrane beta barrels. Science 371, eabc8182 (2021).
Article PubMed PubMed Central Google Scholar
Mravic, M. et al. Packing of apolar side chains enables accurate design of highly stable membrane proteins. Science 363, 1418–1423 (2019).
Article CAS PubMed PubMed Central Google Scholar
Hyde, C. A. et al. Targeting extracellular domains D4 and D7 of vascular endothelial growth factor receptor 2 reveals allosteric receptor regulatory sites. Mol. Cell. Biol. 32, 3802–3813 (2012).
Article CAS PubMed PubMed Central Google Scholar
Leppanen, V. M. et al. Structural and mechanistic insights into VEGF receptor 3 ligand binding and activation. Proc. Natl Acad. Sci. USA 110, 12960–12965 (2013).
Article CAS PubMed PubMed Central Google Scholar
Moraga, I. et al. Tuning cytokine receptor signaling by re-orienting dimer geometry with surrogate ligands. Cell 160, 1196–1208 (2015).
Article CAS PubMed PubMed Central Google Scholar
Wilmes, S. et al. Mechanism of homodimeric cytokine receptor activation and dysregulation by oncogenic mutations. Science 367, 643–652 (2020).
Article CAS PubMed PubMed Central Google Scholar
Walsh, S. T. Structural insights into the common gamma-chain family of cytokines and receptors from the interleukin-7 pathway. Immunol. Rev. 250, 303–316 (2012).
Article PubMed PubMed Central Google Scholar
Cai, T., Lenoir Capello, R., Pi, X., Wu, H. & Chou, J. J. Structural basis of gamma chain family receptor sharing at the membrane level. Science 381, 569–576 (2023).
Article CAS PubMed Google Scholar
Du, G. et al. Autoinhibitory structure of preligand association state implicates a new strategy to attain effective DR5 receptor activation. Cell Res. 33, 131–146 (2023).
Article CAS PubMed PubMed Central Google Scholar
Endres, N. F. et al. Conformational coupling across the plasma membrane in activation of the EGF receptor. Cell 152, 543–556 (2013).
Article CAS PubMed PubMed Central Google Scholar
Freed, D. M., Alvarado, D. & Lemmon, M. A. Ligand regulation of a constitutively dimeric EGF receptor. Nat. Commun. 6, 7380 (2015).
Article PubMed Google Scholar
Chen, K. M., Keri, D. & Barth, P. Computational design of G protein-coupled receptor allosteric signal transductions. Nat. Chem. Biol. 16, 77–86 (2020).
Article CAS PubMed Google Scholar
Yin, J. et al. Structure of a D2 dopamine receptor-G-protein complex in a lipid membrane. Nature 584, 125–129 (2020).
Article CAS PubMed PubMed Central Google Scholar
Ellis, L. M. & Hicklin, D. J. VEGF-targeted therapy: mechanisms of anti-tumour activity. Nat. Rev. Cancer 8, 579–591 (2008).
Article CAS PubMed Google Scholar
Lin, W. et al. Function of CSF1 and IL34 in macrophage homeostasis, inflammation, and cancer. Front. Immunol. 10, 2019 (2019).
Article CAS PubMed PubMed Central Google Scholar
Arber, C. et al. Survivin-specific T cell receptor targets tumor but not T cells. J. Clin. Invest. 125, 157–168 (2015).
Article PubMed Google Scholar
Nishimura, C. D. et al. c-MPL provides tumor-targeted T-cell receptor-transgenic T cells with costimulation and cytokine signals. Blood 130, 2739–2749 (2017).
Article CAS PubMed PubMed Central Google Scholar
Yang, Y., Xie, P., Opatowsky, Y. & Schlessinger, J. Direct contacts between extracellular membrane-proximal domains are required for VEGF receptor activation and cell signaling. Proc. Natl Acad. Sci. USA 107, 1906–1911 (2010).
Article CAS PubMed PubMed Central Google Scholar
Felix, J. et al. Structure and assembly mechanism of the signaling complex mediated by human CSF-1. Structure 23, 1621–1631 (2015).
Article CAS PubMed Google Scholar
Ferrao, R. D., Wallweber, H. J. & Lupardus, P. J. Receptor-mediated dimerization of JAK2 FERM domains is required for JAK2 activation. eLife 7, e38089 (2018).
Article PubMed PubMed Central Google Scholar
Defour, J. P. et al. Tryptophan at the transmembrane-cytosolic junction modulates thrombopoietin receptor dimerization and activation. Proc. Natl Acad. Sci. USA 110, 2540–2545 (2013).
Article CAS PubMed PubMed Central Google Scholar
Arkhipov, A. et al. Architecture and membrane interactions of the EGF receptor. Cell 152, 557–569 (2013).
Article CAS PubMed PubMed Central Google Scholar
Defour, J. P. et al. Constitutive activation and oncogenicity are mediated by loss of helical structure at the cytosolic boundary of thrombopoietin receptor mutant dimers. eLife 12, e81521 (2023).
Article CAS PubMed PubMed Central Google Scholar
Westerfield, J. M. & Barrera, F. N. Membrane receptor activation mechanisms and transmembrane peptide tools to elucidate them. J. Biol. Chem. 295, 1792–1814 (2020).
Article CAS PubMed Google Scholar
Arkhipov, A., Shan, Y., Kim, E. T. & Shaw, D. E. Membrane interaction of bound ligands contributes to the negative binding cooperativity of the EGF receptor. PLoS Comput. Biol. 10, e1003742 (2014).
Article PubMed PubMed Central Google Scholar
Vainchenker, W. & Constantinescu, S. N. JAK/STAT signaling in hematological malignancies. Oncogene 32, 2601–2613 (2013).
Article CAS PubMed Google Scholar
Kut, C., Mac Gabhann, F. & Popel, A. S. Where is VEGF in the body? A meta-analysis of VEGF distribution in cancer. Br. J. Cancer 97, 978–985 (2007).
Article CAS PubMed PubMed Central Google Scholar
Laoui, D., Van Overmeire, E., De Baetselier, P., Van Ginderachter, J. A. & Raes, G. Functional relationship between tumor-associated macrophages and macrophage colony-stimulating factor as contributors to cancer progression. Front. Immunol. 5, 489 (2014).
Article PubMed PubMed Central Google Scholar
Reck, M., Remon, J. & Hellmann, M. D. First-line immunotherapy for non-small-cell lung cancer. J. Clin. Oncol. 40, 586–597 (2022).
Article CAS PubMed Google Scholar
Iwasaki, A., Kuwahara, M., Yoshinaga, Y. & Shirakusa, T. Basic fibroblast growth factor (bFGF) and vascular endothelial growth factor (VEGF) levels, as prognostic indicators in NSCLC. Eur. J. Cardiothorac. Surg. 25, 443–448 (2004).
Article PubMed Google Scholar
Chow, K. K. et al. T cells redirected to EphA2 for the immunotherapy of glioblastoma. Mol. Ther. 21, 629–637 (2013).
Article CAS PubMed Google Scholar
Coffman, K. T. et al. Differential EphA2 epitope display on normal versus malignant cells. Cancer Res. 63, 7907–7912 (2003).
CAS PubMed Google Scholar
Camviel, N. et al. Both APRIL and antibody-fragment-based CAR T cells for myeloma induce BCMA downmodulation by trogocytosis and internalization. J. Immunother. Cancer 10, e005091 (2022).
Article PubMed PubMed Central Google Scholar
Pikman, Y. et al. MPLW515L is a novel somatic activating mutation in myelofibrosis with myeloid metaplasia. PLoS Med. 3, e270 (2006).
Article PubMed PubMed Central Google Scholar
Pardanani, A. D. et al. MPL515 mutations in myeloproliferative and other myeloid disorders: a study of 1182 patients. Blood 108, 3472–3476 (2006).
Article CAS PubMed Google Scholar
Svoboda, D. S. et al. Human iPSC-derived microglia assume a primary microglia-like state after transplantation into the neonatal mouse brain. Proc. Natl Acad. Sci. USA 116, 25293–25303 (2019).
Article CAS PubMed PubMed Central Google Scholar
Lam, N. et al. Anti-BCMA chimeric antigen receptors with fully human heavy-chain-only antigen recognition domains. Nat. Commun. 11, 283 (2020).
Article CAS PubMed PubMed Central Google Scholar
Leaver-Fay, A. et al. ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods Enzymol. 487, 545–574 (2011).
Article CAS PubMed PubMed Central Google Scholar
Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science 378, 49–56 (2022).
Article CAS PubMed PubMed Central Google Scholar
Watson, J. L. et al. De novo design of protein structure and function with RFdiffusion. Nature 620, 1089–1100 (2023).
Article CAS PubMed PubMed Central Google Scholar
Goudy, O. J., Nallathambi, A., Kinjo, T., Randolph, N. & Kuhlman, B. In silico evolution of protein binders with deep learning models for structure prediction and sequence design. Preprint at bioRxiv https://doi.org/10.1101/2023.05.03.539278 (2023).
Baek, M. et al. Accurate prediction of protein structures and interactions using a three-track neural network. Science 373, 871–876 (2021).
Article CAS PubMed PubMed Central Google Scholar
Tunyasuvunakool, K. et al. Highly accurate protein structure prediction for the human proteome. Nature 596, 590–596 (2021).
Article CAS PubMed PubMed Central Google Scholar
Praetorius, F. et al. Design of stimulus-responsive two-state hinge proteins. Science 381, 754–760 (2023).
Article CAS PubMed PubMed Central Google Scholar
Liu, Q. & Perez, A. Assessing a computational pipeline to identify binding motifs to the alpha2beta1 integrin. Front. Chem. 11, 1107400 (2023).
Article CAS PubMed PubMed Central Google Scholar
Yeh, A. H. et al. De novo design of luciferases using deep learning. Nature 614, 774–780 (2023).
Article CAS PubMed PubMed Central Google Scholar
Kloss, C. C., Condomines, M., Cartellieri, M., Bachmann, M. & Sadelain, M. Combinatorial antigen recognition with balanced signaling promotes selective tumor eradication by engineered T cells. Nat. Biotechnol. 31, 71–75 (2013).
Article CAS PubMed Google Scholar
Zhao, Z. et al. Structural design of engineered costimulation determines tumor rejection kinetics and persistence of CAR T cells. Cancer Cell 28, 415–428 (2015).
Article CAS PubMed PubMed Central Google Scholar
Haubner, S. et al. Cooperative CAR targeting to selectively eliminate AML and minimize escape. Cancer Cell https://doi.org/10.1016/j.ccell.2023.09.010 (2023).
Wilkie, S. et al. Selective expansion of chimeric antigen receptor-targeted T-cells with potent effector function using interleukin-4. J. Biol. Chem. 285, 25538–25544 (2010).
Article CAS PubMed PubMed Central Google Scholar
Leen, A. M. et al. Reversal of tumor immune inhibition using a chimeric cytokine receptor. Mol. Ther. 22, 1211–1220 (2014).
Article CAS PubMed PubMed Central Google Scholar
Sukumaran, S. et al. Enhancing the potency and specificity of engineered T cells for cancer treatment. Cancer Discov. 8, 972–987 (2018).
Article CAS PubMed PubMed Central Google Scholar
Chang, Z. L. et al. Rewiring T-cell responses to soluble factors with chimeric antigen receptors. Nat. Chem. Biol. 14, 317–324 (2018).
Article CAS PubMed PubMed Central Google Scholar
Lange, S. et al. A chimeric GM-CSF/IL18 receptor to sustain CAR T-cell function. Cancer Discov. 11, 1661–1671 (2021).
Article CAS PubMed PubMed Central Google Scholar
Shum, T. et al. Constitutive signaling from an engineered IL7 receptor promotes durable tumor elimination by tumor-redirected T cells. Cancer Discov. 7, 1238–1247 (2017).
Article CAS PubMed PubMed Central Google Scholar
Bell, M. et al. Modular chimeric cytokine receptors with leucine zippers enhance the antitumour activity of CAR T cells via JAK/STAT signalling. Nat. Biomed. Eng. 8, 380–396 (2024).
Article CAS PubMed Google Scholar
Nugent, T. & Jones, D. T. Predicting transmembrane helix packing arrangements using residue contacts and a force-directed algorithm. PLoS Comput. Biol. 6, e1000714 (2010).
Article PubMed PubMed Central Google Scholar
Hegedus, T., Geisler, M., Lukacs, G. L. & Farkas, B. Ins and outs of AlphaFold2 transmembrane protein structure predictions. Cell. Mol. Life Sci. 79, 73 (2022).
Article CAS PubMed PubMed Central Google Scholar
Lin, P., Yan, Y., Tao, H. & Huang, S. Y. Deep transfer learning for inter-chain contact predictions of transmembrane protein complexes. Nat. Commun. 14, 4935 (2023).
Article CAS PubMed PubMed Central Google Scholar
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Article CAS PubMed PubMed Central Google Scholar
Lin, Z. et al. Evolutionary-scale prediction of atomic-level protein structure with a language model. Science 379, 1123–1130 (2023).
Article CAS PubMed Google Scholar
Bernhofer, M. & Rost, B. TMbed: transmembrane proteins predicted through language model embeddings. BMC Bioinform. 23, 326 (2022).
Article CAS Google Scholar
Kakarla, S. et al. Antitumor effects of chimeric receptor engineered human T cells directed to tumor stroma. Mol. Ther. 21, 1611–1620 (2013).
Article CAS PubMed PubMed Central Google Scholar
Gundry, M. C. et al. Highly efficient genome editing of murine and human hematopoietic progenitor cells by CRISPR/Cas9. Cell Rep. 17, 1453–1461 (2016).
Article CAS PubMed PubMed Central Google Scholar
Moreno-Mateos, M. A. et al. CRISPRscan: designing highly efficient sgRNAs for CRISPR-Cas9 targeting in vivo. Nat. Methods 12, 982–988 (2015).
Article CAS PubMed PubMed Central Google Scholar
Yi, Z., Prinzing, B. L., Cao, F., Gottschalk, S. & Krenciute, G. Optimizing EphA2-CAR T cells for the adoptive immunotherapy of glioma. Mol. Ther. Methods Clin. Dev. 9, 70–80 (2018).
Article CAS PubMed PubMed Central Google Scholar
Turk, H. F. & Chapkin, R. S. Analysis of epidermal growth factor receptor dimerization by BS(3) cross-linking. Methods Mol. Biol. 1233, 25–34 (2015).
Article PubMed PubMed Central Google Scholar
Rath, J. A. et al. Designed allosteric biosensors for engineered T cell therapy against cancer. GitHub https://github.com/barth-lab/Dimeric_MultiDomain_Biosensor_Builder (2025).

Download references

Acknowledgements

We are very grateful to S. Gottschalk, Baylor College of Medicine and St. Jude Children’s Research Hospital, for providing the previously published EphA2 CAR constructs and the A549.GFP-FFluc cell line. We thank R. Sharma (Barth lab) for initiating the computational framework, and M. Varrin (Arber lab) for technical assistance and lab management. We also thank D. Meraviglia, L. Polak, P. Reichenbach and R. Vuillefroy de Silly for technical advice, as well as G. Giordano Attianese, M. Bell and all Arber and Barth lab members for helpful discussions and comments. J.A.R. was supported by a Swiss Government Excellence Scholarship for Foreign Scholars and the Emma Muschamp Foundation. P.B. is supported by a Swiss National Science Foundation grant (SNSF grant 31003A_182263), Swiss Cancer Research (KFS-4687-02-2019), funds from EPFL and the Ludwig Institute for Cancer Research. C.A. received funding for this study from Swiss Cancer Research KFS-4542-08-2018-R, Stiftung für Krebsbekämpfung, The Leukemia & Lymphoma Society (now Blood Cancer United) Translational Research Program (LLS-TRP 6676-24) and the University of Lausanne.

Author information

These authors contributed equally: Jan A. Rath, Lucas S. P. Rudden, Nazila Nouraee.
These authors jointly supervised this work: Patrick Barth, Caroline Arber.

Authors and Affiliations

Department of Oncology UNIL-CHUV, University Hospital Lausanne (CHUV) and University of Lausanne (UNIL), Lausanne, Switzerland
Jan A. Rath, Tiffany X. Y. Que, Christine Von Gunten, Cynthia Perez, Flora Birch, Yashashvi Bhugowon & Caroline Arber
Ludwig Institute for Cancer Research Lausanne, Lausanne, Switzerland
Jan A. Rath, Tiffany X. Y. Que, Christine Von Gunten, Cynthia Perez, Flora Birch, Yashashvi Bhugowon, Patrick Barth & Caroline Arber
Institute of Bioengineering, School of Life Sciences, Ecole Polytechnique Fédérale Lausanne, Lausanne, Switzerland
Lucas S. P. Rudden, Andreas Fueglistaler, Aisima Chatzi Souleiman & Patrick Barth
Center for Cell and Gene Therapy, Baylor College of Medicine, Houston, TX, USA
Nazila Nouraee & Caroline Arber
Swiss Cancer Center Léman, Lausanne, Switzerland
Patrick Barth & Caroline Arber
Service of Immuno-Oncology, Department of Oncology UNIL-CHUV, University Hospital Lausanne, Lausanne, Switzerland
Caroline Arber
Service of Hematology and Central Laboratory of Hematology, Departments of Oncology UNIL-CHUV and Laboratory Medicine and Pathology, University Hospital Lausanne, Lausanne, Switzerland
Caroline Arber
AGORA Cancer Research Center, Lausanne, Switzerland
Caroline Arber

Authors

Jan A. Rath
View author publications
Search author on:PubMed Google Scholar
Lucas S. P. Rudden
View author publications
Search author on:PubMed Google Scholar
Nazila Nouraee
View author publications
Search author on:PubMed Google Scholar
Tiffany X. Y. Que
View author publications
Search author on:PubMed Google Scholar
Christine Von Gunten
View author publications
Search author on:PubMed Google Scholar
Cynthia Perez
View author publications
Search author on:PubMed Google Scholar
Flora Birch
View author publications
Search author on:PubMed Google Scholar
Yashashvi Bhugowon
View author publications
Search author on:PubMed Google Scholar
Andreas Fueglistaler
View author publications
Search author on:PubMed Google Scholar
Aisima Chatzi Souleiman
View author publications
Search author on:PubMed Google Scholar
Patrick Barth
View author publications
Search author on:PubMed Google Scholar
Caroline Arber
View author publications
Search author on:PubMed Google Scholar

Contributions

J.A.R. and N.N. designed the research, performed the experiments, analysed and interpreted the results, and wrote parts of the paper. L.S.P.R. and A.F. designed the research and developed the de novo assembly method for modelling and design of biosensors. L.S.P.R. analysed and interpreted the results, wrote and released the software, and wrote parts of the paper. A.F. designed the research, developed the de novo assembly method for modelling and design of biosensors, and analysed and interpreted the results. A.C.S. performed the mechanical coupling calculations and analysed the results. J.A.R., T.X.Y.Q., C.V.G., C.P., F.B. and Y.B. performed the experiments and analysed the results. P.B. conceived the computational study, designed the research, supervised the study, analysed and interpreted the results and wrote the paper. C.A. conceived the study, designed the research and supervised the entire study, performed some experiments, analysed and interpreted the results, and wrote the paper. All authors reviewed and approved the final version of the paper.

Corresponding authors

Correspondence to Patrick Barth or Caroline Arber.

Ethics declarations

Competing interests

P.B. holds patents in the field of protein engineering. C.A. holds patents in the field of engineered T cell therapies. C.A. and P.B. hold patents and provisional patent applications in the field of engineered T cell therapies. J.A.R. and L.S.P.R. hold a provisional patent application in the field of engineered T cell therapies. C.A. receives licensing fees and royalties from Immatics (through previous institution Baylor College of Medicine), participated in advisory boards for Kite/Gilead, Janssen and Celgene/BMS, and received sponsored travel from Gilead (through current institution University Hospital Lausanne). The other authors declare no competing interests.

Peer review

Peer review information

Nature Biomedical Engineering thanks Jonathan Bramson, James Chou and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 Impact of extracellular-intracellular (EC-IC) communication and dimerization on designed receptor signaling with inactive biased sensor.

(a) Schematic representation of the two primary structural mechanisms driving single-pass receptor activation: Pre-Formed Dimer (PFD) and Monomer-Dimer Equilibrium (MDE). In the PFD mechanism, the sensor and responder exist in a pre-formed dimer and transition between two dimeric conformations. In contrast, the MDE mechanism involves the sensor and responder switching from monomeric to dimeric conformations. Both mechanisms ultimately converge on the same activation state. (b-c) Biosensor design scenarios involving a sensor and responder, each intrinsically biased toward either the inactive or active state, and their resulting activation behavior. (b) Inactive-biased sensor and active-biased responder. Left. In the low communication mode, the biosensor’s behavior is primarily governed by the intrinsic properties of the sensor and responder domains. The responder tends to occupy the active state, leading to high basal activity even in the absence of ligand. Meanwhile, the sensor favors an inactive conformation, resulting in low apparent ligand binding affinity and a higher EC50. Due to suboptimal signal transmission between the sensor and responder, the sensitivity to ligand binding, as reflected in the hill slope, is also low. Our selected CMRs were designed according to this scenario. Right. In the high communication mode, the sensor substantially influences the responder’s behavior, shifting it toward the inactive state. Conversely, the responder pushes the sensor toward the active state, which enhances ligand binding affinity and lowers the EC50. The hill slope increases due to stronger signal transmission between the sensor and responder. Our VMRs correspond to this high communication scenario. (c) Inactive-biased sensor and inactive-biased responder. Left: In the low communication mode, basal activity remains minimal as the responder favors its inactive state. Ligand-saturating conditions fail to elicit a maximal response due to weak coupling between the sensor and responder. Right: In high communication mode, basal activity remains minimal due to the strong inactive state bias of both the sensor and responder. However, at saturating ligand conditions, maximal activity is achieved. The EC50 is higher compared to (b) and (c) because ligand binding competes with the inactive bias in both the sensor and responder. (a-c) Generated with bioRender.

Extended Data Fig. 2 Impact of extracellular-intracellular (EC-IC) communication and dimerization on designed receptor signaling with active biased sensor.

(a) Schematic representation of the two primary structural mechanisms driving single-pass receptor activation: Pre-Formed Dimer (PFD) and Monomer-Dimer Equilibrium (MDE). In the PFD mechanism, the sensor and responder exist in a pre-formed dimer and transition between two dimeric conformations. In contrast, the MDE mechanism involves the sensor and responder switching from monomeric to dimeric conformations. Both mechanisms ultimately converge on the same activation state. (b-c) Biosensor design scenarios involving a sensor and responder, each intrinsically biased toward either the inactive or active state, and their resulting activation behavior. (b) Active-biased sensor and inactive-biased responder. Left: The responder predominantly adopts its inactive state, resulting in low basal activity. The coupling between the sensor and responder is insufficient to fully switch the responder to the active state under ligand-saturating conditions, leading to a low hill slope and reduced maximal activity. However, the EC50 is low because the sensor in its active state has a high affinity for the ligand. Right: Sufficient coupling between the sensor and responder allows the responder to fully switch to the active state upon ligand binding, driving full biosensor activation. This increases sensitivity (hill slope) while maintaining a low EC50. Basal activity is also enhanced due to the shift in equilibrium caused by the active-biased, ligand-unbound sensor. (c) Active-biased sensor and active-biased responder. Left: The responder predominantly occupies the active state, resulting in potent basal activity. Ligand binding further shifts the remaining inactive receptors into the active state. Right: In high communication mode, basal activity is further enhanced due to a shift in equilibrium, driven by the active-biased, ligand-unbound sensor. Ligand binding activates the remaining inactive receptors, maximizing overall activity. The EC50 remains low in both communication modes, owing to the bias toward high-affinity, active states for ligand binding. (a-c) Generated with bioRender.

Extended Data Fig. 3 Impact of tuned communication on the VMR and CMR design scenarios.

Possible mechanism of activation for VMR (top) and CMR (bottom) based on the level of communication (coupling) between sensor and responder. Left: without ligand. In the low communication mode, the sensor only weakly interacts with the responder. Hence, the responder can readily occupy its preferred active conformation and trigger high levels of basal activity. In the high communication mode, the sensor strongly influences and shifts the responder towards the inactive conformation, turning off the basal activity. Right: with ligand. In the low communication mode, a saturating ligand concentration pushes all remaining inactive receptors into the active state, producing the maximum signal. In the high communication mode, the sensor’s state is intimately tied to the responder, therefore if all sensors switch to the active state on ligand binding, so too do all responders, producing a potent signaling response. Parts of the figure were generated with bioRender.

Extended Data Fig. 4 Stepwise assembly modeling of VMR_FL.

Step 1: input monomeric or dimeric domain structures to the assembly protocol and starting chimeric scaffold. Steps 2–4: assembly of additional domains (blue) into the existing scaffold (purple) given distance constraints. Domains 1 to 7 correspond to the extracellular domains of VEGFR2 (blue to cyan at last stage), while TM and CP correspond to the transmembrane and cytoplasmic regions of the c-MPL receptor, respectively (gray to purple). Generated with bioRender.

Extended Data Fig. 5 Domain composition of the different VMR and CMR variants.

(a) Each domain for all chimeras is colored by their respective structural source. Blue refers to a homology model, purple to a crystal structure from the PDB, red/green the respective ligands for VMR and CMR, black a modeled linker by the Rosetta assembly protocol, and yellow absent structural data from our in silico designs. The dimerization domains, also the crossing points of the assembly protocol, are highlighted via the shaded block, while the predicted helical linker connecting the TM of c-MPL and EC of CSF1R is shown in navy blue. (b) Topology description of the various designed VMR and CMR chimera generated by the assembly protocol. The keys define each type of domain assembled into the chimera. (c) Breakdown of number of unique conformations built by the assembly protocol and the number of corresponding ProteinMPNN sequences generated for the specific linker length. There is no datapoint for MPNNFL_nohelix as it shares the same DSSP prediction as CMR_FL. (d) Measured DSSP helicity/coil nature of linkers sampled by ProteinMPNN at increasing linker length. With increasing linker length, we report the average helical (blue line) and coil (orange line) probability returned by S4PRED across the linker, averaged across all ProteinMPNN-generated sequences. The 9 chosen CMR chimera (where CMR_FL and CMR_FL_nohelix possess the same sequence), are also provided via the circles. (a-d) Parts of the figure were generated with bioRender.

Extended Data Fig. 6 Equivalent in vivo anti-tumor function of EphA2-BBζ and EphA2-BBζ.VMR_FL T cells in the VEGFA_low mouse model of disseminated lung cancer.

(a) Schematic of the VEGFA_low mouse model evaluating the in vivo anti-tumor function of engineered T cells. (b) Summary of bioluminescent imaging (BLI) total flux [p/s], lines representing values from individual mice, n=5 mice/group. 1 representative of 2 independent experiments. (c) Area under the curve analysis of total flux shown in panel b, from day -7 to day 24, mean±SEM, unpaired t-test with Welch’s correction. Key comparisons: ∆ vs ∆.VMR, p=0.0082; ∆ vs BBζ, p<0.0001; ∆ vs BBζ.VMR_FL, p<0.0001; BBζ vs BBζ.VMR_FL, p=0.1156. (d) Survival of mice. n=5 mice/group, Kaplan Meier analysis and log-rank (Mantel Cox) test. Key comparison: BBζ vs BBζ.VMR_FL, p>0.9999.

Extended Data Fig. 7 Signaling and in vivo anti-tumor function of EphA2-BBζ CAR-T cells expressing constitutively active c-MPL_W515L.

(a) Schematic representation of constitutively active c-MPL with the W515L mutation. (b) STAT5 phosphorylation in c-MPL+ or c-MPL_W515L+ T cells with or without stimulation with 25 ng/ml recombinant human thrombopoietin (TPO) for 1h, gated on CD110+ T cells. Single representative donor FACS histograms of n=4 donors evaluated. (c) Schematic representation of the retroviral vector constructs. c-MPL_W515L: Engineered c-MPL expressing the mutation W515L for high constitutive activity. EphA2-BBζ CAR and VMR_FL constructs as introduced previously. (d) Schematic of mouse model used to assess the impact of constitutively active c-MPL_W515L on CAR-T cell function. Mice were infused with 1x10⁵ transgenic T cells (‘transgenic’ is referring to the double positive population for BBζ.VMR_FL and BBζ.c-MPL_W515L conditions, and to the single positive population for VMR_FL and BBζ conditions). The cell product infused in the BBζ.VMR_FL condition contained 1x10⁵ CAR+VMR_FL+, 0.42x10⁵ CAR+VMR_FL-, 0.32x10⁵ CAR-VMR_FL+, and 0.58x10⁵ CAR-VMR_FL- cells. The cell product infused in the CAR.c-MPL_W515L condition contained 1x10⁵ CAR+c-MPL_W515L+, 1.56x10⁵ CAR+c-MPL_W515L-, 0.89x10⁵ CAR-c-MPL_W515L+ and 3.83x10⁵ CAR-c-MPL_W515L- cells. (e) Images of individual mice, color scale ranges from 1x10⁵ to 1x10⁶ p/sec/cm2/sr. (f) Summary of total flux, lines representing average flux and SD. n=5 for no T cells, n=6 for VMR_FL, n=8 for the other groups. Results from 1 experiment. (g) Area under the curve analysis of total flux shown in k, from D0 to D7 (left) and from D0 to D10 (right, end of experiment). mean±SEM, unpaired t-test (two-sided) with Welch’s correction. Key comparisons: (left) BBζ.VMR_FL vs BBζ.c-MPL_W515L, p<0.0001; BBζ vs BBζ.c-MPLW515L, p=0.8266; (right) BBζ.VMR_FL vs BBζ.c-MPL_W515L, p<0.0001.

Extended Data Fig. 8 Equivalent in vivo anti-tumor function of mAPRIL-mBBζ and mAPRIL-mBBζ.CMR_FL T cells in the CSF1_neg mouse model of multiple myeloma.

(a) Schematic of the CSF1_neg mouse model evaluating the in vivo anti-tumor function of engineered T cells using the mAPRIL-mBBζ CAR. (b) BLI images of individual mice, color scale ranges from 1x10⁴ to 1x10⁶ p/sec/cm2/sr. (c) Summary of total flux, lines representing individual mice. n=10 mice/group (except for untreated (n=7) and for NT (n=9)), pooled results of 2 independent experiments. (d) Area under the curve (AUC) analysis of total flux shown in c, from day 0 to day 27 mean±SEM, unpaired t-test (two-sided) with Welch’s correction. Key comparisons: NT vs BBζ, p<0.0001; BBζ vs BBζ.CMR_FL, p=0.2211. (e) Survival of mice. Kaplan Meier analysis, log-rank (Mantel Cox) test. Key comparisons: NT vs BBζ, p=0.0027; NT vs BBζ.CMR_FL, p=0.0027; BBζ vs BBζ.CMR_FL, p=0.1936.

Supplementary information

Supplementary Information

Supplementary Materials and Methods, Supplementary Data references and Supplementary Data figures and legends (total 21 supplementary figures).

Reporting Summary

Peer Review File

Source data

Source Data Figs. 3–8

Statistical source data.

Source Data Fig. 3

Uncropped western blot images of results shown in Fig. 3c.

Source Data Fig. 4

Uncropped western blot images of results shown in Fig. 4f.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Rath, J.A., Rudden, L.S.P., Nouraee, N. et al. Computational design of synthetic receptors with programmable signalling activity for enhanced cancer T cell therapy. Nat. Biomed. Eng (2025). https://doi.org/10.1038/s41551-025-01532-3

Download citation

Received: 26 January 2024
Accepted: 02 September 2025
Published: 28 October 2025
Version of record: 28 October 2025
DOI: https://doi.org/10.1038/s41551-025-01532-3