The precision and efficacy of cell therapies in oncology have raised new hopes for patients battling refractory and relapsed liquid cancers. Chimeric antigen receptor (CAR)-T cell therapy, relying on the engineering of autologous T cells to target surface antigens of haematopoietic malignancies, bridged synthetic biology with medicine (Zugasti et al., 2025). There are currently six FDA approved CAR-T cell therapies against haematological cancers. All of them represent the second out of five CAR generations, characterised by an intracellular co-stimulatory domain at the tail of a CAR. Third-generation CARs incorporate multiple co-stimulatory domains, fourth-generation ('TRUCKs') add inducible cytokine expression to modulate the tumor environment, and fifth-generation CARs include additional signaling motifs such as JAK/STAT for enhanced activation and control (Korell et al., 2022).
However, about 90% of human cancers are solid tumors. Solid cancers create a characteristic tumor microenvironment (TME) that consists of distinct physical features like hypoxia and acidity, mechanical features like vascularity, metabolic gradients, inflammation, and immune modulation (Y. Wang et al., 2025). Tumors mitigate native immune defense actively, by upregulating checkpoint ligands (PD-L1), secreting immunosuppressive factors (TGF-β, IL-10), and recruiting regulatory T-cells that inhibit their anti-tumoral counterparts. Additionally, passive immune evasion mechanisms hinder immune reaction by minimizing neoantigen display, and limiting the T-cell infiltration into dense, vascular extracellular matrix (ECM) (B. Wu et al., 2024).

Even though extensive effort has been invested in unlocking CAR-T cell therapy for solid tumors, major hurdles prevent their treatment with currently approved measures. These encompass immunosuppressive TME, antigen loss and heterogeneity, hostile tumor metabolism, limited persistence, and laborious personalized manufacturing (Uslu & June, 2025). Another important area for improvement in CAR T-cell therapy is controlling their excessive activation, which can trigger life-threatening cytokine release syndrome (CRS) in a significant proportion of patients (Lu et al., 2024).

Inspired by the efficacy of CAR T cell therapy in haematological cancers, yet aware of its limitations, we identified an opportunity to separate the immune effector function from native immune mechanisms to enhance their accuracy and efficacy for solid tumors. We used a bottom-up approach to engineer therapeutic cells, insensitive to cancer immunosuppression, while allowing precise control of therapeutic response.

To keep pace with the latest advancements in cell therapy, we have incorporated the most extensively researched and clinically most promising features into our system:

Reversible and Safe Response

Background

Current approved CAR T-cell therapies lack safety switches, and uncontrolled immune activation can lead to cytokine release syndrome (CRS), a life-threatening condition characterized by an exaggerated immune response. This response results in the excessive release of cytokines such as interferon-γ, granulocyte-macrophage colony-stimulating factor, proinflammatory interleukins, and tumor necrosis factor-α (B. Liu et al., 2024). CRS rate ranges from 42-100% and is the most common side effect in CAR T patients, while immune effector cell-associated neurotoxicity syndrome (ICANS) constitutes another major threat (X. Xiao et al., 2021).

---

Our Implementation:

• Improved safety of PHOENICS cells - We implemented an inhibitory input receptor that senses ligands indicative of healthy tissue to prevent immune overreaction.
• Dephosphorylation of the substrate - The inhibitory input receptor is connected to a phosphatase, which reverses substrate phosphorylation
• Dynamic response reversal - Dephosphorylation restrains the immune-like effector functions when the cell is located in healthy tissue
• Kill switch - We plan to implement the inducible iCasp9 kill switch for complete termination of PHOENICS cell activity when they pose a danger to the patient.

Click to read more

State of Research:

Currently available mechanisms of switching therapeutic cells’ response fall into two classes: reversible switches and irreversible ‘suicide’ system. Reversible switches are either small-molecule-gated or adapter-controlled and allow dose-tunable, temporary suppression of cells’ activity. The function is restored when the adapter is withdrawn. Irreversible safety switches based on iCasp9 or HSV-TK permanently eliminate CAR T-cells via induced apoptosis. iCasp9-based switches have progressed to phase I/II clinical testing, and some two adapter switches are in phase one (Lu et al., 2024; Minagawa et al., 2019). By implementing these safety layers into our PHOENICS cells, we are aligning our project with the latest research and developments in cell therapy.

Detection of Soluble Protein Ligands

Background

Immunotherapies often target cancer-specific surface ligands to engage immune cells or directly kill the cancer cells. However, TME is characterized by a distinct repertoire of secreted proteins. Soluble ligand gradients can be universal for many cancer entities, including TGF-β, IL-10, IL-8 or cancer specific, like GDF15 for pancreatic cancer or LY6K for many cervical tumors (Luo et al., 2016, p. 201; Melero et al., 2025; J. Yang et al., 2025, p. 202).

---

Our Implementation:

• Soluble protein sensing via GPCRs - We engineered the PHOENICS circuit to target soluble protein ligands within the TME and have shown a selective gene expression and secretion of effector proteins controlled by VEGF sensing.
• Soluble protein sensing via MESA-receptors - We demonstrated PHOENICS' ability to process soluble TNF-alpha concentrations to regulate gene expression via MESA-receptors.
• De novo design of protein binding domains - To integrate various soluble protein cues of the TME, we created and screened a library of cancer-relevant de novo binders for possible stimulatory and inhibitory ligands, using our dry lab model SPARC. With that, the PHOENICS circuit can be expanded to sense any soluble protein-ligands characteristic of a patient's TME and express or secrete therapeutic effectors of choice.

Click to read more

State of Research

Several phase I and II clinical trials sought to leverage the characteristic cues of the TME by specifically targeting the tumor secretome in solid tumors. These approaches often involve blocking immunosuppressive effector proteins through small molecules, monoclonal antibodies, or antibody-drug conjugates (J. Yang et al., 2025). The influence of soluble ligands in the TME on the immune response was also acknowledged in the field of cell therapy. An example, currently in phase I clinical trials for myeloma are TGF-β induced NK cells equipped with engineered dominant-negative TGF-β receptor II. Though they bind soluble TGF-β, it does not induce immunosuppressive signaling (B. Yang et al., 2013). Inverted cytokine receptors are currently engineered to sense immunosuppressive IL-4 cytokines, but convert the downstream signal to that of the immunostimulatory IL-7 receptor, propagating pro-T-cell signaling (Y. Wang et al., 2019). Moreover, CAR T cells have also been adjusted to sense soluble ligands, but platforms whose primary logic relies on endogenous soluble-factor sensing have not yet entered clinical trials (Chang et al., 2018). Building upon these precedents of engineering immune cells to respond to TME-derived soluble factors, we have designed PHOENICS cells to specifically activate in TME.

Detection of Environmental Small Molecules

Background

Aberrant small molecule repertoires in TME, such as high lactate concentrations, is one of many strategies of cancer to evade immune response (X. Jin et al., 2025; Peralta et al., 2024). Some molecules, like succinate, are simply byproducts of aberrant metabolism, but others, like extracellular lactate or ATP, not only accumulate due to metabolic changes but also actively signal to promote tumor progression, including processes like angiogenesis (X. Jin et al., 2025; Tannahill et al., 2013; Vultaggio-Poma et al., 2020).

---

Our Implementation:

• Small-molecule sensing via synthetic Receptors - Both, our MESA-receptors and synthetic GPCRs are able to sense small molecules. By retargeting them to lactate, ATP or other small-molecules upregulated in the TME, we can tap into a wide variety of cancer-relevant ligands to precisely activate our system.
• Proof of concept with SalB - We optimized an ultrasensitive small molecule input, achieving a reproducible 151-fold change in phosphorylation-dependent reporter expression.
• ATP sensing receptor - Based on our proof of concept, we engineered a receptor that translates environmental ATP levels into substrate phosphorylation levels, and response.

Click to learn more

State of Research:

Although small molecule sensing has not yet been applied for tumor targeting, there are existing tools for coupling effector response to the environmental concentration of small molecules. Zhang et. al recently reengineered CAR receptor architecture to selectively bind antigens in TME-specific high ATP concentrations (Q. Zhang et al., 2024). Another approach to harness small molecules to control CAR activation is through an ON-switch. Some proof-of-concept systems have been developed to gate either extra- or intracellular transduction of stimulatory signals based on small molecule dependence (Park et al., 2021; C.-Y. Wu et al., 2015). By integrating the ability to sense not only proteins but also to process small-molecule cues, we equip the PHOENICS toolbox with exceptional flexibility by design.

Simultaneous Dual-Ligand Recognition

Background

The recognition of two ligands simultaneously is mostly combined with the logic gates of the effector function of a cell. Main types of logic gates include AND, OR, and NOT gates, and they are frequently combined in more sophisticated decision computation to ensure the precision of the output of a system (Savanur et al., 2021).

---

Our Implementation:

• Multi-ligand integration in PHOENICS cells - Our PHOENICS circuit computes two opposing inputs and only activates when the stimulatory signal outweighs the inhibitory one. We demonstrated TNFα (stimulatory) and rapalog (inhibitory) to show that gene expression activation is stopped once the inhibitory cue is sensed.
• AND-gate using PAGER - Our Synthetic nanobody-fused PAGER GPCRs are a small-molecule-protein AND-gate, enabling more complex cellular processing of extracellular ligands.
• Expanded MESA extracellular domain library - We integrate Bindcraft-generated binders into extracellular domains design, allowing the user to adjust the response activation thresholds based on select TME cues.

Click to learn more

State of Research:

The potential of simple gates in cell therapies is constantly explored in preclinical studies, but only few sophisticated gate combinations have progressed to clinical trials. SynNotch, an universal receptor architecture for various logic gates implementation in cells, is in phase 1 for irreversibly induced CAR expression against glioblastoma (Okada, 2025; Shirzadian et al., 2025). Another trial involving multiple solid cancer entities, implements the NIMPLY gated CAR receptors, that sense heterozygous gene loss, and trigger the response when HLA surface protein is absent (A2 Biotherapeutics Inc., 2025; Tokatlian et al., 2022). Lastly, a Phase 1 CAR-NK therapy integrates OR with NOT logic by engaging leukemia targets but switching off when sensing HSC markers to protect hematopoietic stem cells (Garrison et al., 2021). PHOENICS serves as a framework for cellular processing of extracellular ligand cues with logic operations. With it we lay the foundation for cell therapies with different levels of complexity.

Rapid Effector-Secretion for Immune Activation and Apoptosis Induction

Background

Interleukins mediate communication in innate and adaptive immunity. In immunotherapy, suppressive interleukins are blocked, while stimulatory ones are delivered to activate immune responses at the tumor site (Briukhovetska et al., 2021). IL-12 is a powerful cytokine that stimulates anti-tumor immunity. However, its systemic use causes severe toxicity, making locally confined delivery necessary for safe cancer therapies (Nguyen et al., 2020). TNF-related apoptosis-inducing ligand (TRAIL) is a cytokine used in cancer immunotherapy. TRAIL binds to receptors specifically expressed on cancer cells, triggering their apoptosis while sparing healthy cells. Continuous tumor evolution often leads to TRAIL resistance, limiting its effectiveness. This resistance can be overcome through the combined secretion of kinase inhibitors that prevent the development of TRAIL resistance, restoring its tumor-killing capabilities (Montinaro & Walczak, 2023).

---

Our Implementation:

• Phosphorylation-controlled release - In PHOENICS effector proteins can be retained in the endoplasmic reticulum and released in a burst upon activation.
• Localized therapeutic delivery - Precise TME-mediated activation of PHOENICS cells ensures robust secretion of tumor-relevant IL-12 or TRAIL of tumor tissue while minimizing off-target effects.
• Adaptable secretion platform - Our proof-of-concept for a robust secretion platform, allows to adapt the PHOENICS cells to secrete a wide range of therapeutically-relevant proteins.

Click to learn more

Gene and cell therapy approaches, including intratumoral delivery of IL-12-encoding plasmids and engineering of adoptive T cells (TRUCKs) to secrete IL-12 upon antigen recognition, have to successfully enhance antitumor immune responses in early clinical trials. These therapies leverage IL-12's ability to stimulate immune activation specifically within the TME, avoiding the severe side effects reported upon systemic IL-12 administration (Briukhovetska et al., 2021).
TRAIL remains another potent pro-apoptotic agent, showing clinical effectiveness when delivered with CDK9 kinase inhibitors. One clinical trial for TRAIL release by ACTs in metastatic melanoma has reached phase II but was terminated later (Rosenberg, 2015). Another trial for ACTs involves tumor-homing MSCs, expressed as lung cancer therapy (University College, London, 2021). Current development focuses on combination therapies and engineered cells secreting TRAIL to overcome resistance and improve tumor selectivity (Snajdauf et al., 2021). The PHOENICS toolbox features two powerful effector moieties that are currently the focus of extensive research for their use in cell therapy applications. Its modular architecture enables the seamless integration of additional effectors, which can be customized for specific cancer types.

We envision that the specificity of PHOENICS cells will make them a safe and effective solution for metastatic and solid tumors. The modularity of the circuit and the interchangeability of its components will allow scientists and clinicians to adapt our platform as a tunable cell therapy. Using our dry lab model SPARC, additional components can easily be integrated and activation thresholds can be specifically tailored to cancer types or patient needs. Ultimately, we foresee that the universal design of the circuit will provide a foundation for experiments in different cellular chassis, enabling the identification of the optimal chassis for specific tumor moieties. This approach could pave the way toward an off-the-shelf therapy against common tumor entities with a defined, standardized secretome while also enabling personalized treatments specifically tailored to patients' TME profiles.

Pancreatic Cancer

Pancreatic cancer is an exceedingly aggressive solid tumor that originates in the digestive tract. It is characterized by a high risk of local invasion and metastasis. For patients with pancreatic cancer, the preferred treatment is early surgical restriction. Nevertheless, the tumor is usually diagnosed in advanced stages, when surgical restriction is only an option in 20% of the time, and the overall prognosis is poor (Zheng et al., 2024). Advanced unretractable tumors are most frequently treated with a neoadjuvant chemotherapy regimen, such as FOLFIRINOX, consisting of the drugs 5-fluorouracil, leucovorin, irinotecan, and oxaliplatin. Although, this harsh treatment only marginally increases the short 5-year survival rate, from 5% to 8%. For patients with metastatic pancreatic cancer, palliative chemotherapy still often remains the only option (Czaplicka et al., 2024; Zheng et al., 2024).

Standard therapeutic interventions including, radiation therapy, and chemotherapy are associated with severe adverse effects and a low overall survival rate, particularly in cases of pancreatic cancer (Zheng et al., 2024). Immune checkpoint inhibitors (ICIs) such as anti-programmed death receptor 1 (PD-1) antibodies, anti-programmed cell death ligand (PDL-1) antibodies, and anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) antibodies have been approved for the treatment of various tumors, including pancreatic cancer (Czaplicka et al., 2024). In combination with chemotherapy, the PD-1 antibody nivolumab leads to an increased 1-year survival rate of 35% for pancreatic cancer patients. A downside is that ICIs that target PD-1 can only be applied to the minority of patients whose tumors express PD-1. Therefore, the heterogeneity of cancer limits the efficacy of the limited therapeutic options (Q. Yang et al., 2024).

Regarding the application in pancreatic cancer, we consulted Prof. Dr. Rienk Offringa, an expert in cell therapies for gastrointestinal tumors at DKFZ, who advised us to target ATP as a small-molecule input and GDF15 as a soluble protein ligand.Click to see interview

PHOENICS circuit design

We propose the PHOENICS circuit design to accurately compute the concentrations of soluble ligands within the TME of pancreatic cancer. The therapeutic response is triggered by integrating signals from cancer-specific stimulatory ligands and inhibitory ligands, thereby protecting healthy tissue from off-target effects. High cancer specificity and minimized tissue damage can be achieved by the PHOENICS cells, computing the ratio between upregulated ligands such as GDF15, OGN, and LY6K, and those downregulated, like SPOCK2.

Positive Ligands

GDF15

Click to learn more

Growth Differentiation Factor 15, a member of the TGF-β family, is normally expressed at low levels in healthy tissues but becomes strongly upregulated under conditions of cellular stress, tissue damage, and inflammation. Upregulated GDF15 promotes proliferation and metastasis in pancreatic cancer (Zhu et al., 2024). It is released into the ECM (Guo & Zhao, 2024) which makes a perfect potential stimulatory input for pancreatic cancer

OGN

Click to learn more

Osteoglycin is a proteoglycan of the small leucine-rich proteoglycan (SLRP) family, which is found in the ECM. OGN can act as a tumor suppressor or promoter, with its function varying in different cancer types. For Pancreatic Cancer, overexpression compared to healthy tissue was found during all stages of the malignancy (Qin et al., 2022).

LY6K

Click to learn more

Lymphocyte antigen 6 family member K, member of the LY6 family, was found to be upregulated across multiple solid tumors, including pancreatic and cervical cancer. Elevated LY6K expression in these malignancies is linked to poor clinical outcomes, marking it as both a potential prognostic biomarker and a promising target for therapeutic intervention (Luo et al., 2016).

Negative Ligands

SPOCK2

Click to learn more

SPOCK2 is an ECM proteoglycan found to be significantly downregulated in pancreatic cancer cell lines secretome, compared to healthy pancreatic tissue. Its decreased expression is associated with enhanced tumor cell proliferation and migration, while higher SPOCK2 levels correlate with better patient prognosis, making it a promising inhibitory input ligand for PHOENICS cells against pancreatic cancer (Aghamaliyev et al., 2023).

Our dry lab team generated binders for all of the ligands mentioned above. To demonstrate the relevance of the synthetic binder generation, we functionally validated the GDF15 binder in the wet lab (link to GDF15).

Outlook in Cell Therapies

Adoptive cell therapies (ACT) are a rapidly advancing approach in cancer treatment, offering new hope for patients with challenging, heterogeneous solid tumors like pancreatic cancer. For instance, the use of CAR-T cells has recently reached phase III clinical trials. Patients with advanced pancreatic cancer overexpressing mesothelin were treated with anti-mesothelin 7*19 CAR-T cells. For the majority, the therapy resulted in a near-complete tumor eradication within 240 days. These results show the potential of ACT and the importance of further research in this field, for the treatment of advanced solid tumors (Pang et al., 2021).

Regardless of the recent advances, all ACTs are associated with a range of adverse effects and limitations. A critical limitation pertains to interpatient and intratumoral tumor heterogeneity, owing to the highly variable and dynamic antigen landscape of solid tumors, which severely restricts the efficacy and broad applicability of ACT. Additional barriers include the poor tumor infiltration capability of immune cells or the persistence of the therapeutic agents within the tumor microenvironment. Another major concern is on-target off-tumor toxicity, which occurs when the targeted antigens are not strictly tumor-specific. This can lead to collateral damage to healthy tissue. Finally, cytokine release syndrome (CRS) represents a potentially life-threatening systemic inflammatory response. CRS is caused by excessive immune activation and can result in high fever, hypotension, multi–organ dysfunctions, and in severe cases, death (Y. Jin et al., 2021).

Cervical Cancer

Cervical cancer is a common and potentially lethal malignancy of the female reproductive tract, ranking as the fourth most frequent cancer in women worldwide. Almost all cases (~95%) are attributable to persistent infection with high-risk human papillomavirus (HPV), particularly genotypes 16 and 18, which drive oncogenesis through viral oncoproteins E6 and E7 that inactivate the tumor suppressors p53 and pRb. Despite being largely preventable through HPV vaccination and screening programs, cervical cancer continues to represent a major global health burden, particularly in low- and middle-income countries where access to preventive strategies is limited due to socioeconomic factors (Caruso et al., 2024).

The prognosis of cervical cancer is closely tied to the stage at diagnosis. Early-stage disease is mostly curable with radical hysterectomy, a surgical removal of the whole uterus and cervix. For locally advanced disease, chemoradiation with cisplatin and brachytherapy remains the standard of care. However, patients with recurrent or metastatic cervical cancer face a poor prognosis, with a 5-year survival rate of only ~19%. These cases have historically relied on chemotherapy, sometimes combined with bevacizumab immunotherapy, but responses have been modest (Caruso et al., 2024; Tewari, 2025). Recent advances in immunotherapy have reshaped the treatment of cervical cancer. In the metastatic setting, the PD-1 inhibitor pembrolizumab combined with chemotherapy ± bevacizumab is FDA-approved, improving survival but adding risks of anemia, neutropenia, and other severe side effects. In locally advanced disease, pembrolizumab with chemoradiation is now FDA-approved, providing a progression-free treatment, regardless of PD-L1 status. Other checkpoint strategies, including or bispecific antibodies are in clinical trials. New systemic options include antibody-drug conjugates (ADCs). The tissue factor-targeted ADC tisotumab vedotin received full FDA approval in 2024 after demonstrating a survival benefit, but it also carries notable ocular toxicities and peripheral neuropathy. In addition, HER2-directed therapy (trastuzumab deruxtecan) received accelerated approval for the small subset of HER2-positive cervical cancers (Caruso et al., 2024; Tewari, 2025; Yu et al., 2023).

PHOENICS circuit design

Efficiency and safety of cervical cancer treatment can be achieved by recruiting PhoENICS cells to sense ligands upregulated in the cervical cancer secretome, namely LY6K and Cyr61, and reverse therapeutic response once inhibitory ligands such as OGN and CKRT13 are detected.

Positive Ligands

LY6K

Click to learn more

Except for its role in pancreatic cancer, LY6K is severely overexpressed in cervical cancer secretome and found to regulate its malignant character of cervical cancer (Chen et al., 2016). With a fold change between 4.9 and 5.6 in healthy tissue compared to cancerous tissue (Luo et al., 2016; Robinson et al., 2019), LY6K poses as a promising candidate as stimulatory input.

CYR61

Click to learn more

In more than half of primary cervical cancers, the pro-angiogenic protein CYR61 is significantly upregulated. Overexpression is associated with aggressive tumor features, such as increased depth of invasion and unfavorable prognosis, suggesting a role for CYR61 in driving tumor progression. CYR61 is a known inducer of inflammatory cytokines shaping the TME. Once secreted, it associates with the ECM and cell surfaces, influencing cell adhesion, migration, and angiogenesis (Mayer et al., 2017).

Negative Ligands

DPT

Click to learn more

Dermatopontin is a protein involved in the ECM assembly and interactions, negatively regulates cell adhesion, and is downregulated in cervical cancer (Bettadapura et al., 2025). With a fold change between -5.7 and -5.4, DPT is a possible inhibitory input for PHOENICS cells in cervical cancer (Robinson et al., 2019).

OGN

Click to learn more

Osteoglycin is notably downregulated in squamous cervical cancer, which accounts for approximately 83% of all cervical cancer cases (M. Wang et al., 2024). With a fold change between -6.4 and -7.9 in healthy tissue compared to cancerous tissue, OGN is a promising candidate as an inhibitory input for cervical cancer therapies (Robinson et al., 2019).

CKRT13

Click to learn more

Proteomic analysis and immunohistochemistry confirm significantly lower Keratin 13 (CKRT13) expression in cervical cancer tissues compared to normal cervical mucosa. CKRT13 secretion reflects the disruption of the epithelial cell differentiation and integrity. Its loss marks dedifferentiation of tumor cells and correlates with invasive phenotypes (Lomnytska et al., 2010).

Outlook in Cell Therapies

Cell therapies for cervical cancer are still experimental. Tumor-infiltrating lymphocytes (TILs) therapy has shown durable responses in some patients but is limited by complex manufacturing and IL-2-related toxicities (Yu et al., 2023). Another TILs cell therapy targets both viral and non-viral tumor antigens, helping to overcome immune evasion. CAR-T and TCR-T therapies targeting HPV E6/E7 and other antigens remain in early trials, with challenges including CRS and on-target toxicity. TCR therapies targeting HPV antigens in clinical trials aim to enhance specificity and efficacy while therapeutic vaccines like the Listeria-based ADXS11-001 and DNA vaccines induce HPV-specific immunity.

Although promising, challenges such as immune suppression within the tumor microenvironment, response variability, and manufacturing remain significant (Burmeister et al., 2022).

Glioma

Gliomas are a diverse group of brain tumors originating from glial cells. They can progress to glioblastoma, the most aggressive form, marked by rapid growth, microvascular proliferation, necrosis, and specific metabolic mutations. Glioblastoma presents as the most devastating primary malignancy of the central nervous system, with patients facing a median survival of 12-15 months despite maximal intervention. The disease's infiltrative growth makes complete surgical removal nearly impossible, as tumor cells spread extensively into healthy brain tissue. While neurosurgeons aim for the safest, most complete resection, often aided by fluorescence-guided techniques using 5-aminolevulinic acid, microscopic disease usually remains, leading to recurrence (Singh et al., 2025).

Post-operative management combines radiotherapy with the alkylating agent temozolomide (TMZ). This treatment improves the two-year survival rate from 10.4% to only 26.5%. This minimal improvement of survival rate is dependent on the tumor's molecular profile. Only half of the patients whose tumors exhibit MGMT promoter methylation favorably respond to TMZ treatment, while the rest minimally benefit from systemically toxic chemotherapeutic (Singh et al., 2025). Initially responsive tumors can develop resistance to TMZ or even alter their microenvironment, promoting the evolution into a more aggressive phenotype (Chien et al., 2021).

Multiple approaches to replace the unspecific chemoradiation have been undertaken, mainly targeted against immunomasking PD-1. However, the results of phase III trials immunotherapeutic approaches against glioblastomas have proven profoundly disappointing. PD-1 inhibitors like nivolumab and pembrolizumab, either as monotherapy or combined with standard systemic chemotherapy, have consistently failed to improve survival rates. The tumor establishes an intensely immunosuppressive microenvironment, homing the regulatory T cells, myeloid-derived suppressor cells, and M2-polarized macrophages to actively suppress cytotoxic immune responses. The blood-brain barrier compounds additionally limit both drug penetration and immune cell trafficking. Another obstacle for implementing immunotherapy is glioblastomas' relatively stable genome, which provides few neoantigens.

As an expert in the field of CAR T-cells development, Dr. Mirco Friedrich advised us to use 2-HG as a stimulatory input.

Click to see interview

PHOENICS circuit design

The PhoENICS circuit could harness the complex TME of glioma by activating its effector functions upon sensing a hallmark of glioma - 2-HG. Response can be silenced in an increasing gradient of RELN, an ECM protein abundantly secreted by neurons, but diminished in glioma tissue.

Positive Ligands

2-HG

Click to learn more

2-hydroxyglutarate is a product of mutated isocitrate dehydrogenase (IDH), produced in approximately 80% of lower-grade gliomas and secondary glioblastomas. This metabolite plays a crucial role in shaping the TME by promoting immunosuppressive state in infiltrating myeloid cells. This induces a tolerogenic phenotype in T-cells and suppresses antigen presentation, enabling the cancer to evade the immune response (Friedrich et al., 2021).

Negative Ligands

RELN

Click to learn more

RELN is an ECM protein crucial in neural development, primarily regulating neuronal migration and brain development. RELN downregulation in gliomas correlates with increased tumor malignancy. RELN signaling prevents cancer cell migration, and its loss in glioma cells enhances the migratory behavior, contributing to tumor progression and metastasis (Li et al., 2020).

Outlook in Cell Therapies

Recent clinical trials have explored autologous tumor-infiltrating lymphocytes, engineered to secrete anti-PD-1 antibodies. Another approach that has reached phase I and II utilized CAR-T cells targeting antigens like IL-13Rα2 and EGFRvIII. Only bivalent linked CAR-T cells targeting both IL-13Rα2 and EGFR showed tumor reduction in patients. Although the criteria for overall response rate were not met, this approach highlighted the potential of dual-target strategies for overcoming antigen escape (Y. Liu et al., 2024). Moreover, NK cell-based therapies, including autologous NK cell infusion, have shown safety in early Phase I trials, but the cytotoxic efficacy remains limited. Stem cell-based strategies have also reached phase I/II stages, with trials like the use of mesenchymal (MSCs) or or neural stem cells (NSCs). Their tumor-tropic migration makes them promising vehicles that locally deliver therapeutic enzymes, prodrug-activators, or oncolytic viruses (G. Wang & Wang, 2022). Additionally, the combination of dendritic cell vaccines, such as Deca®-L, with tumor treatment with electrical fields is being explored in Phase III trials, with indications of improved overall survival (Van Gool et al., 2023).

Challenges such as the high variability in tumor response, difficulty in determining progression, and the complex immunological dynamics hindered many trial designs and patient recruitment (G. Wang & Wang, 2022).