We designed and tested a new class of immunotherapies - MiTEs - that combine myeloid checkpoint inhibition with NK and T cell activation in the TME. Over the course of the project, we developed the full experimental pipeline, identified lead molecules, and demonstrated their strong anti-tumor potential.
1. Targeting TREM2 to reprogram myeloid cells
Using our humanized TREM2 mouse model, we established in-vitro screening systems based on bone-marrow-derived tumor-associated macrophages (TAMs). Flow cytometry and single-cell RNA sequencing (scRNA-seq) showed that monocytes lacking TREM2 do not become immunosuppressive macrophages. We generated hundreds of anti-TREM2 antibodies, identified lead candidates with blocking activity equal to genetic deletion of TREM2, and confirmed that our top antibody (E3C7) replicated the beneficial profiles of TREM2-deficient cells.
Using our Zman-Seq time-stamping technology, we mapped in vivo how monocytes differentiate into TAMs in glioblastoma. Blocking TREM2 disrupted this process, preventing the development of highly immunosuppressive macrophages and shifting them toward a pro-inflammatory state. However, anti-TREM2 alone did not achieve strong tumor control, highlighting the need for combination approaches.
2. Selecting and engineering potent cytokines
To boost anti-tumor immunity, we screened clinically relevant cytokines (including IL-2 “superkine” (IL2SK), IL-12, GM-CSF, and IL-15 “sushi”) in inducible tumor cell line models. IL2SK showed the strongest individual activity, and combinations - particularly IL2SK with IL-12 - were highly synergistic.
We created an anti-TREM2–IL2SK fusion protein, which strongly activated immune cells in vitro but caused severe systemic toxicity in vivo. To solve this, we used single-cell multiomics and machine learning to identify a TAM-specific protease expressed alongside TREM2 in tumors but absent from healthy tissues. We then engineered MiTE prototypes with cytokines “masked” by a cleavable blocking moiety that activates only in the presence of this protease.
3. Safe, tumor-specific activation
Our tumor-activated cytokines remained inert in vitro until exposed to the TAM-specific protease, at which point they matched the potency of unmasked cytokines. In mice, these protease-activated MiTEs showed no toxicity - even at high repeated doses - and no weight loss or systemic cytokine spikes.
4. In vivo efficacy of lead MiTEs
Our lead molecule, MiTE-144, completely eliminated MC38 tumors and outperformed anti-TREM2, anti-PD-1, or their combination. Single-cell analysis revealed a massive increase in cytotoxic and proliferating NK and T cells, with NK cells showing the strongest expansion and activation, even in conditions where T cells typically fail (e.g. tumors lacking MHC-I). This NK response is critical for targeting PD-1–resistant tumors.
Outcomes
We have established a robust, data-driven pipeline for designing safe and effective MiTEs that combine myeloid checkpoint inhibition with NK and T cell activation. Our work demonstrates:
• Proof-of-concept that TREM2 blockade reprograms suppressive macrophages.
• Identification of optimal cytokines and synergistic combinations.
• Development of tumor-specific cytokine activation technology that avoids systemic toxicity.
• Discovery of a lead MiTE molecule with complete tumor-eradicating potential in vivo.
These results position our TREM2-focused MiTEs as a promising therapeutic strategy for immune checkpoint blockade–resistant solid tumors, with a clear path toward clinical translation.
