Every cancer researcher recognizes a familiar pattern: a transformative therapy delivers remarkable results for a subset of patients, yet fails most others. Immune checkpoint inhibitors (ICIs) have redefined oncology, but the majority of patients remain trapped inside cold, immunosuppressive tumor microenvironments (TMEs) that render treatment ineffective. Our joint team from Tsinghua University, the Chinese Academy of Medical Sciences, and Ningbo Combireg Pharmaceutical Technology Co. Ltd set out to decode this treatment resistance and uncovered an unexpected molecular culprit: the innate immune sensors NOD1 and NOD2, which act as a hidden switch that turns the TME against cancer immunotherapy, chemotherapy, and adoptive T cell therapy.
The unexpected culprits
NOD1 and NOD2 are best known as sentinels against bacterial infection, recognizing fragments of bacterial cell walls and triggering inflammation. Our team uncovered them as the drivers of adaptive resistance to chemotherapy, radiotherapy, and immune checkpoint blockade. This discovery rested on nearly twenty years of continuous laboratory work.
Our design of NOD-modified taxane conjugates began in 2000. At that time, only NOD agonists were considered promising antitumor and antimetastatic agents, and our early efforts were directed toward developing such agonists. Over several years, we iteratively refined the molecular scaffolds to fine-tune their interactions with the NOD pathway, hypothesizing that these conjugates would not only exhibit tumor cytotoxicity but also inhibit tumor metastasis by activating the NOD1/2 signaling pathway. However, our subsequent studies revealed that it was NOD antagonism, rather than agonism, that sensitized tumors to chemotherapy. This unexpected finding helped shift the paradigm toward NOD1/2 signaling antagonism. In 2017, we reported the first non-cleavable paclitaxel–MDP derivative conjugate that functioned as a NOD2 antagonist, which sensitized tumors to paclitaxel therapy and significantly prevented metastasis. Shortly thereafter, we developed Salutaxel (CBRG001), a docetaxel–MDP analogue conjugate that acts as a multifunctional prodrug, inhibiting both tumor growth and metastasis through NOD1/2 antagonism. This conjugate has since progressed to clinical trials (NCT-TR20211264); phase Ib has been completed, and phase II is currently in preparation. Together with our earlier work, these findings established that NOD antagonists can facilitate the elimination of certain cancers, a conceptual shift that has opened new therapeutic avenues.
Another key milestone was reached in 2022, when we validated the role of NOD1 in promoting colorectal cancer metastasis via macrophage inflammatory responses. We demonstrated that tumor-generated extracellular vesicles containing CDC42 could activate NOD1 in macrophages. This long-term commitment ensured that we were well-prepared to evaluate the pathway within the context of immunotherapy.
The pivotal moment arrived during a routine experiment comparing wild-type and NOD1/2 double-knockout mice. When both groups were treated with anti-PD-L1, the knockout mice responded dramatically – tumors shrank, while wild-type tumors remained largely unaffected. An entire year was then devoted to eliminating confounding variables: co-housing experiments ruled out gut microbiome interference, and RIPK2 small-molecule inhibition precisely replicated the knockout phenotype, confirming NOD signaling as the root cause. At that point, we realized that two decades of incremental research had led us to an important resistance regulator.
From observation to mechanism
Determining the relevant cellular population was the most challenging aspect of the study. Single-cell RNA-sequencing data revealed substantial shifts in CD8⁺ T cells and macrophages following NOD1/2 deletion. To investigate causation, we constructed a reductionist system: bone-marrow-derived macrophages were co-cultured with CD8⁺ T cells in the presence of NOD1/2 agonists and RIPK2 inhibitor.
The results were clear. Activated macrophages became potently immunosuppressive, inhibiting T cell proliferation and interferon-γ production. This suppression depended entirely on NOD1/2 signaling–knockout macrophages failed to exert any inhibitory effect. The critical mediator was PD-L1 on macrophages in the TME, which was upregulated through the NOD1/2-RIPK2–NF-κB axis. This created a protective shield on the macrophage surface, effectively turning these cells into barriers against T cell attack. The chromatin immunoprecipitation data provided the definitive confirmation: p65 (NF-κB) bound to the PD-L1 promoter exclusively following NOD1/2 activation.
The clinical bridge
Having established the mechanism in mice, we next evaluated whether it operates in patients. We focused on microsatellite-stable (MSS) colorectal cancer, a subtype representing approximately 95% of colorectal cancer patients that is notoriously resistant to ICIs. Using patient-derived organoids and autologous PBMCs, we reconstructed the human TME in vitro.
The addition of a RIPK2 inhibitor together with anti-PD-L1 produced a striking effect: previously quiescent T cells were reactivated, infiltrating the organoids and eliminating cancer cells. Bioinformatic analysis across multiple cancer types, colorectal, liver, and breast, confirmed that NOD1/2 signatures are consistently upregulated in tumor-associated macrophages following treatment, indicating that this may represent a general adaptive resistance program rather than a tumor-specific phenomenon.
Lessons learned and future directions
Reflecting on this journey, patience emerges as one of the most valuable lessons. Each step, from the initial chemotherapy conjugates to discovering the broader role of NOD1/2, required time and meticulous experimentation.
An equally important lesson was the need to follow the data, even when it contradicts established dogma. Conventionally, NOD1/2 were viewed as pro-inflammatory mediators; yet, in this context, they drove immunosuppression. This apparent contradiction was resolved when we recognized that context is critical. Within a therapy-stressed TME where immunosuppression is fully established in cancer patients, the activation of these receptors upregulates PD-L1 on macrophages. This serves as a supplementary immunosuppressive mechanism that impairs effector CD8⁺ T cell activation and infiltration, complementing the classical PD-1/PD-L1 immune checkpoint.
Our ongoing work focuses on whether combining NOD1/2 antagonists with existing therapies can overcome resistance in other immunotherapy-refractory tumors. Given that our team has already advanced a first-in-class NOD1/2-antagonistic taxane conjugate into clinical trials (CBRG001, phase Ib, NCT‑TR20211264), the translational path is shorter than we had anticipated.
Science is often incremental and demanding. Occasionally, however, a result stands out in a way that offers a compelling direction for understanding a biological mechanism with possible therapeutic implications. For our team, the day those NOD1/2 knockout mice responded to anti-PD-L1 was such a moment. We hope this work marks the beginning of a new chapter in overcoming therapy resistance, one in which macrophages are no longer silent enablers of tumor progression, but targets we have finally learned to control.