The therapy, described in a study published in Frontiers in Pharmacology, is built on Mycobacterium smegmatis, a fast-growing mycobacterial species that is inherently immunostimulatory but does not produce endotoxin. The researchers, led by Leilei Zhou, Zhaoye Qian, and Jipeng Wang of the Affiliated Huai’an No. 1 People’s Hospital of Nanjing Medical University, engineered the bacterium to secrete a bifunctional fusion protein: a single-chain variable fragment that blocks the PD-L1 immune checkpoint, fused to an interleukin-15 (IL-15) immunocytokine that includes the IL-15 receptor alpha sushi domain. The construct, dubbed Ms-PI15, is designed to combine checkpoint blockade with sustained local cytokine signaling that promotes the proliferation and activation of natural killer cells and CD8-positive T cells.

To test the approach, the team established a syngeneic model of malignant pleural effusion by injecting luciferase-expressing Lewis lung carcinoma cells directly into the left pleural cavity of C57BL/6 mice. Within days, the cells established pleural tumors, and by day 15 tumor cells were detectable in the pleural fluid, confirming that the model faithfully reproduced the human condition. Beginning on day five after tumor inoculation, mice received intrapleural injections of either phosphate-buffered saline, an empty-vector control bacterium, or Ms-PI15, repeated on days 10 and 15.

A dose-optimization experiment comparing 10 million and 100 million colony-forming units showed a clear dose-dependent effect. Mice receiving the higher dose displayed significantly reduced bioluminescent signal from the chest region on serial imaging, along with markedly smaller pleural tumor masses and reduced effusion volumes at the endpoint on day 21. Body weight remained stable across all groups throughout the observation period, an early indication that the treatment was well tolerated.

In the head-to-head efficacy experiment, the differences were striking. Tumor burden, quantified as total radiance efficiency over the chest, was significantly lower in the Ms-PI15 group from day 13 onward compared with both the saline and empty-vector controls. By day 21, tumor mass and effusion volume were both dramatically reduced, with p-values below 0.001 for tumor weight and below 0.0001 for effusion volume. Critically, the empty-vector bacterium performed no better than saline on any efficacy endpoint, demonstrating that the antitumor activity was specifically attributable to the PD-L1/IL-15 fusion protein rather than to the nonspecific immunostimulatory properties of the bacterial carrier itself.

Survival data reinforced the therapeutic signal. Median survival was 29 days in the saline group, 37 days in the empty-vector group, and 47 days in the Ms-PI15 group—a substantial extension that translated tumor suppression into a meaningful survival benefit in this aggressive model.

Understanding where the engineered bacteria go after injection was central to evaluating both mechanism and safety. The team labeled Ms-PI15 with the near-infrared fluorescent dye DiR and tracked its distribution over two weeks. Fluorescence accumulated predominantly in the pleural tumor region, peaking at 72 hours after administration and declining thereafter. Ex vivo imaging confirmed persistent signal at the tumor site and detectable fluorescence in the thoracic draining lymph nodes, suggesting that the bacteria, or antigens derived from them, migrate to or are transported by immune cells into the draining lymphatic system. Colony-forming unit counts from homogenized tissues corroborated the imaging: viable bacteria were abundant in tumor tissue and detectable in the draining nodes, but colonization of the heart, liver, spleen, lung, kidney, and blood was negligible or below the detection limit at every timepoint examined.

The immunological remodeling induced by Ms-PI15 was comprehensive. Flow cytometric analysis of the pleural tumor revealed a significant increase in the proportion of CD80-positive, CD86-positive mature myeloid dendritic cells among CD11c-positive cells, indicating enhanced antigen-presenting capacity. Macrophage polarization shifted markedly: the fraction of CD206-positive, M2-like immunosuppressive tumor-associated macrophages fell, while the proportion of CD86-positive, M1-like antitumor macrophages rose. Because tumor-associated macrophages are among the most abundant immune populations in malignant pleural effusion and actively promote tumor progression through pro-angiogenic signaling and suppression of cytotoxic lymphocytes, this repolarization represents a key mechanism of action.

Cytotoxic effector cells surged in parallel. The percentage of CD8-positive T cells in the tumor increased significantly, as did the fraction of NK1.1-positive, CD3-negative natural killer cells, and immunohistochemical staining of tumor sections independently confirmed the denser CD8-positive infiltration. The immune effects extended beyond the pleural cavity: CD8-positive T cells were also elevated in the spleen, and mature dendritic cells increased in the thoracic draining lymph nodes, pointing to systemic immune activation that the authors suggest may help guard against metastatic dissemination. Together, these changes describe a conversion of the immunosuppressive, cold pleural microenvironment into an immunologically active, hot one.

Biosafety findings were equally encouraging within the limits of the model. Histopathological examination of the heart, liver, spleen, lung, and kidney revealed normal architecture with no inflammation, necrosis, or other treatment-related abnormalities in any group. Serum biochemistry showed no significant differences in alanine aminotransferase, aspartate aminotransferase, or alkaline phosphatase, markers of hepatic function, nor in creatinine or urea, markers of renal function. The authors caution, however, that several questions remain before clinical translation: the Lewis lung carcinoma model represents a single tumor histology, the study used only male mice, some mechanistic experiments involved small cohorts, and long-term durability and immunological memory were not assessed. Larger animal studies and eventual clinical trials will be needed to determine whether intrapleural Ms-PI15 can reproduce these results in human patients with malignant pleural effusion.

Subject of Research: Intrapleural engineered Mycobacterium smegmatis immunotherapy for malignant pleural effusion in a mouse model

Article Title: Engineered Mycobacterium smegmatis-based intrapleural immunotherapy remodels the pleural immune microenvironment for malignant pleural effusion treatment

Article References: Zhou, L., Qian, Z., Yao, R., Chen, C., Pan, P., Guo, Z., Xu, Z., Mei, Y., & Wang, J. (2026). Engineered Mycobacterium smegmatis-based intrapleural immunotherapy remodels the pleural immune microenvironment for malignant pleural effusion treatment. Frontiers in Pharmacology, 17, Article 1972278. https://doi.org/10.3389/fphar.2026.1972278

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1972278

Keywords: malignant pleural effusion, Mycobacterium smegmatis, in situ vaccine, PD-L1 blockade, interleukin-15, tumor microenvironment, dendritic cell maturation, macrophage polarization, CD8 T cells, NK cells, biodistribution, preclinical study