"Trojan Microbe": Bacteria Hide Oncolytic Virus from Immune System and Launch It Directly into Tumors

Oncolytic viruses can kill cancer cells, but are often powerless against… our immunity: neutralizing antibodies intercept viruses in the blood, preventing them from reaching the tumor. A team from Columbia Engineering has proposed a clever workaround: hide the virus inside a bacterium that itself finds and populates the tumor. In Nature Biomedical Engineering, they presented the CAPPSID platform - "Coordinated Activity of Prokaryote and Picornavirus for Safe Intracellular Delivery". The bacterium Salmonella typhimurium produces RNA of the oncolytic virus Senecavirus A (SVA) and releases it inside the tumor cell, from where the virus takes off and spreads, remaining invisible to circulating antibodies. In immunocompetent mice, such a "hitch" suppressed tumor growth and worked even with existing antiviral immunity.

Background of the study

Oncolytic viruses have long been considered "self-replicating drugs": they select cancer cells, replicate inside them, and trigger an immune response against the tumor. But the approach has a persistent systemic barrier - delivery. When administered intravenously, viruses are quickly intercepted by neutralizing antibodies and elements of the innate immune system, some of the particles "stick" in the liver and spleen, and only a small proportion reaches a dense, poorly perfused tumor. Therefore, many clinical protocols are forced to limit themselves to intratumoral injections, which narrows the range of indications and makes it difficult to treat multiple foci.

In parallel with viruses, another branch of “live” antitumor agents developed – engineered bacteria. Weakened strains of Salmonella, E. coli, Listeria, etc. demonstrate tumorotropism: they readily populate hypoxic tumor zones and can serve as carriers for local delivery of cytotoxins, cytokines, or genetic cassettes. But bacterial therapy acts locally and is limited by the scale of colonization: it is difficult to reach cells outside the “bacterial nests,” and safety and controllability are always under close control of regulators.

Against this background, the idea of combining the strengths of both worlds seems logical. Previously, attempts were made to “shield” viruses with polymers, hide them in carrier cells (for example, mesenchymal stem cells), use exosomes - all these approaches partially bypass antibodies, but complicate production and control. Bacteria are able to independently find a tumor and deliver the “cargo” deep into the tissue; if they are taught to launch the virus directly inside the tumor cell, it is possible to bypass the systemic immune “anti-air umbrella” and simultaneously expand the affected area beyond the colony due to further viral spread.

The key to translation is safety control. A naked oncolytic virus in a bacterium could theoretically “go wild.” That’s why modern platforms build multi-level fuses: viral RNA is synthesized and released only in the tumor cell, and full assembly of virions is made dependent on the “key” — a specific protease or other factor that only the bacterium supplies. As a result, the virus remains a “blind passenger” until it reaches the target; the immune system does not see it in the bloodstream; it is launched in a targeted manner, and the probability of uncontrolled dispersal is reduced. This is the strategy that the new work develops, demonstrating that a “courier bacterium” can reliably deliver an oncolytic picornovirus to a tumor and turn it on where it is really needed.

How it works

• Bacteria-spotter. Engineering S. typhimurium naturally reaches for the tumor and is able to penetrate cancer cells. Inside, it transcribes viral RNA (including the full-length SVA genome) using specified promoters.
• Autolytic "trigger". The bacterium is programmed to lyse in the cytoplasm of the tumor cell and simultaneously releases viral RNA and an auxiliary enzyme. The virus begins a replication cycle and infects neighboring cells.
• Security control. The virus is further modified: to assemble mature virions, it requires a protease "key" (for example, TEV protease), which is supplied only by the bacterium. This limits uncontrolled spread.
• "Shield" from antibodies. While the viral RNA is "packed" in the bacteria, neutralizing antibodies in the blood do not see it, which helps delivery to the tumor.

What the experiments showed

• In culture: CAPPSID triggered full-fledged SVA infection and virus dissemination among cells not infected with the bacterium (including on H446 neuroendocrine lung cancer lines).
• In mice, intratumoral and intravenous administration of CAPPSID inhibited tumor growth and allowed robust viral replication; in some models, subcutaneous SCLC tumors were completely eradicated.
• Immune “noise immunity”: the system worked even in the presence of neutralizing antibodies to SVA: the bacteria delivered the genome to the tumor, and the virus was launched “behind the line of defense.”
• Control of spread: The virus's conditional dependence on a bacterial protease allowed it to limit the number of infection cycles outside the original cell - an additional layer of safety control.

Why this is important (and how it differs from conventional approaches)

Classic oncolytic viruses suffer from two problems: antibodies intercept them in the blood, and systemic spread carries risks of toxicity. Engineered bacteria, on the contrary, love tumors, but act locally and have a hard time “reaching” the periphery of the neoplasm. CAPPSID combines the strengths of both worlds:

• delivery via bacteria → higher chance of reaching the tumor, bypassing antibodies;
• virus inside → infects neighboring cells and expands its area of action beyond the bacterial colony;
• A built-in "fuse" in the form of a virus requiring a bacterial protease reduces the risk of uncontrolled dissemination.

Technical details

• In Salmonella, the SPI-1/SPI-2 pathogenicity island promoters were recruited to precisely activate transcription of viral RNA and lysis proteins (HlyE, φX174 E) at the right time and place.
• They tested both replicons (self-amplifying but non-spreading RNA) and full-length SVA, which was more effective at expanding the lesion through reinfection.
• TEV protease was used as an “external key” to the assembly of virions: without it, the virus “does not mature.”

Limitations and questions for future reference

• For now, this is preclinical: cells, immunocompetent mice, a limited set of tumor models; orthotopic models and GLP toxicology are ahead.
• A thorough assessment of the safety of bacteria during systemic administration and the resistance of the “fuse” to the mutational escape of the virus is needed (the authors are already laying down the choice of incision sites that reduce the chance of reversions).
• A real clinic will require strains with proven safety (e.g. human attenuated Salmonella derivatives) and a well-thought-out combination with immunotherapy.

What could this mean tomorrow?

• New 'living drugs' for solid tumors where delivery is the main bottleneck.
• Viral target personalization: SVA demonstrates tropism for neuroendocrine tumors; theoretically, the platform could be repurposed for other oncolytic picornaviruses or replicons.
• Reduction of viral particle consumption and risk of systemic side effects due to local launch at the site of infection.

Conclusion

Engineers have turned the bacteria into a “living capsid” that hides the virus from antibodies, delivers it to the tumor, and provides the key to safely launching it inside. In mice, this curbs tumor growth and bypasses antiviral immunity — the next step is to confirm the platform’s safety and customizability on the way to clinical trials.

Source: Singer ZS, Pabón J., Huang H., et al. Engineered bacteria launch and control an oncolytic virus. Nature Biomedical Engineering (online 15 August 2025). doi: 10.1038/s41551-025-01476-8.