ABOVE: Bacteria (pink) cozy up to dividing colorectal cancer cells (blue) in this false-color scanning electron micrograph.

In the 1966 movie Fantastic Voyage, a team of scientists is shrunk to fit into a tiny submarine so that they can navigate their colleague’s vasculature and rid him of a deadly blood clot in his brain. This classic film is one of many such imaginative biological journeys that have made it to the big screen over the past several decades. At the same time, scientists have been working to make a similar vision a reality: tiny robots roaming the human body to detect and treat disease. 

Although systems with nanomotors and onboard computation for autonomous navigation remain fodder for fiction, researchers have designed and built a multitude of micro- and nanoscale systems for diagnostic and therapeutic applications, especially in the context of cancer, that...

Although systems with nanomotors and onboard computation for autonomous navigation remain fodder for fiction, researchers have designed and built a multitude of micro- and nanoscale systems for diagnostic and therapeutic applications, especially in the context of cancer, that could be considered early prototypes of nanorobots. Since 1995, more than 50 nanopharmaceuticals, basically some sort of nanoscale device incorporating a drug, have been approved by the US Food and Drug Administration. If a drug of this class possesses one or more robotic characteristics, such as sensing, onboard computation, navigation, or a way to power itself, scientists may call it a nanorobot. It could be a nanovehicle that carries a drug, navigates to or preferentially aggregates at a tumor site, and opens up to release a drug only upon a certain trigger. The first approved nanopharmaceutical was DOXIL, a liposomal nanoshell carrying the chemotherapeutic drug doxorubicin, which nonselectively kills cells and is commonly used to treat a range of cancers. The intravenously administered nanoshells preferentially accumulate in tumors, thanks to a leaky vasculature and inadequate drainage by the lymphatic system. There, the nanoparticles slowly release the drug over time. In that sense, basic forms of nanorobots are already in clinical use. 

Precise navigation to tumor sites remains a holy grail of nanorobot research and development.

Scientists can manipulate the shape, size, and composition of nanoparticles to improve tumor targeting, and newer systems employ strategies that specifically recognize cancer cells. Still, precise navigation to tumor sites remains a holy grail of nanorobot research and development. A 2016 meta-analysis assessing the efficiency of nanodelivery vehicles tested in animal studies in the previous 10 years revealed that a median of fewer than 1 percent of the injected nanovehicles actually reached the tumor site, and that this could be only marginally improved with active targeting mechanisms, such as surface decoration with specific antibodies or peptides for tumor-specific receptor binding. 

How can we make these nanobots better at steering themselves to tumor sites? Wireless energy transmission remains a huge challenge, and batteries are not yet efficient at the nanometer scale. Researchers have used external forces such as ultrasound or magnetic fields to promote the homing of nanomedicines to tumor tissues, but the fluid dynamics of the circulatory system work against nanoshuttles, whose surface-to-volume ratio is 1 billion times that of objects on the scale of meters. This causes surface and drag forces to become more dominant: to the nanoparticle, it might feel like moving through honey when navigating the aqueous environment of the vasculature. 

But as it so often does, nature might just have a solution: bacteria. The microscopic organisms swim autonomously through fluids, driven by molecular motors that spin their cilia or flagella in a corkscrew-like fashion—a very effective propulsion mechanism at this scale that has inspired many nanoroboticists that try to mimic this functionality. Researchers have fabricated helical, magnetic swimmers that can be spun forward by a rotating magnetic field, for example. But bacteria, especially in the context of treating cancer, are more than just role models for efficient swimming; some are actually themselves therapeutic. In addition, microbes can sense biochemical cues and adjust their trajectories accordingly, similar to the envisioned on-board computation.

The idea of using bacteria to treat cancer is not new. One of the earliest reports on bacteria as a cancer therapy comes from the immunotherapy pioneer William Coley, who in the late 19th century recognized that some cancer patients also suffering from skin infections were more likely to get better. He began injecting bacterial toxins, heat-inactivated microbes, or even live cultures of Streptococcus bacteria into his patients with inoperable bone and soft-tissue cancers, often leading to remissions. It was a bold approach, given the risk of uncontrollable infections from these bacterial formulations before the widespread availability of antibiotics. Largely because of that danger, and the promise of the nascent concepts of radiation and chemotherapy, the clinical use of bacteria as therapeutic agents for cancer went undeveloped. Today, this revolutionary idea has been experiencing a renaissance.

Thanks to the convergence of fields from biology and chemistry to materials science, engineering, and computer sciences, new avenues for the development of bacterial therapies for cancer are opening up. The toolkits made available thanks to reduced costs of both sequencing and synthesis of DNA, along with synthetic-biology approaches for custom genetic design of bacterial-like behaviors, are paving the way for the emerging fields of micro- and nanorobotics. 

A History of Nanoengineering and Bacterial Therapy

Engineered nanorobots that can roam inside the body to detect and treat tumors have been a vision for the past half century, and the idea of using bacteria to fight cancer is even older than that. Researchers have come to understand that some bacteria innately possess some traits of a nanorobot: they can autonomously seek out tumors and have readily toxic payloads that can kill cancer cells. Combining bacteria with classical approaches in robotics and engineering for external control and guidance, researchers may now be turning the once-fictional idea of a cancer-fighting nanorobot into reality—and the robot is alive.

The scientist staff

Bacteria with anti-cancer payloads

Bacillus Calmette-Guérin (BCG), an attenuated bacterium typically used as a vaccine strain for tuberculosis, has been repurposed for the last several decades to locally treat bladder cancer. The concept behind this approach, similar to that postulated by Coley, is that the administration of bacteria stimulates the patient’s immune system to fight off the cancer. 

Even better, though unbeknownst to Coley, many bacteria (though, for unknown reasons, not BCG) also have the potential to selectively grow within solid tumors, in the bladder and elsewhere; reduced immune surveillance in the tumor’s hypoxic and acidic environment provides anaerobic bacteria with a safe haven to grow and thrive. While inside tumors, some bacteria produce toxins and compete with cancer cells for nutrients. Ultimately, the accumulation of bacteria within the tumor induces immune-cell infiltration, which can then lead to anti-cancer responses. Still, despite having tested many naturally occurring and laboratory-made bacterial strains in animal models of cancer, and having conducted human trials testing bacteria to treat cancer, researchers have observed little efficacy beyond the benefits that continue to be seen in bladder cancer patients. 

As a result, the field has shifted to genetically engineering bacteria to serve as ferries for recombinant payloads. The selective targeting and subsequent growth of bacteria in tumors, along with local delivery of therapeutics facilitated by the microbes themselves, could minimize the collateral damage to healthy cells that is common with systemic cancer therapies. Several groups have engineered bacteria to produce a wide variety of cargo, including anticancer toxins, cytokines, and apoptosis-inducing factors. The production of potentially toxic therapeutic cargo necessitates further control over the bacteria, in case they land in locations they shouldn’t. Thus, researchers are now moving toward engineering next-generation bacterial systems to sense a physiological cue and respond by producing a therapeutic at the local disease site. 

Salmonella typhimurium

To aid in this goal, over the last two decades the field of synthetic biology has developed a repertoire of genetic circuits to control microbial behaviors. These circuits consist of positive and negative feedback motifs to modulate dynamic cellular functions, acting as toggle switches, oscillators, counters, biosensors, and recorders—tools that researchers have used to design cancer-fighting microbes. 

One example of genetic control over cancer-fighting bacteria is the synchronized lysis circuit developed in 2016 by Jeff Hasty’s group at the University of California, San Diego, in collaboration with Sangeeta Bhatia’s laboratory at MIT, where both of us did our postgraduate training. (T.D. was a coauthor on this 2016 study.) In this circuit, bacteria localize to tumors and grow to a critical density, then synchronously rupture to release therapeutic compounds that the microbes had been producing. This approach, which takes advantage of natural bacterial quorum sensing, improves upon several features of previously developed bacterial therapies, most of which constitutively produce drugs, meaning they might make and release the therapeutics in unintended areas of the body. Because bacteria only reach critical density within tumors, they will only self-destruct and release their therapeutic payload there. This leads to pruning of the microbial population, preventing uncontrolled growth of bacteria in the tumor or elsewhere. In a colorectal liver metastasis mouse model, this system resulted in a twofold increase in survival when paired with chemotherapy, as compared with chemotherapy or bacteria alone.

Several groups have further developed this approach. In 2019, for example, one of us (T.D.), along with Columbia University microbiologist and immunologist Nicholas Arpaia and colleagues, created bacteria that produced molecules known to block immune checkpoints, such as CD47 or PD-L1, which ordinarily put the brakes on immune cells and thereby decrease anti-tumor activity. As a result of blocking these pathways in tumors, bacteria were able to prime T cells and to facilitate the clearance of cancer in a lymphoma mouse model. Most surprisingly, untreated tumors within treated animals also shrank, suggesting that local priming could trigger distant and durable antitumor immunity. 

The approach of using bacteria as a cancer therapy is starting to attract the attention of the biotech industry. One company, BioMed Valley Discoveries, has been testing injections of the spores of Clostridium novyi-NT, an obligate anaerobe that can only grow in hypoxic conditions and is genetically attenuated so that a lethal toxin is not produced, in several clinical trials. In rats, dogs, and the first human patient, the treatment showed “precise, robust, and reproducible antitumor responses,” according to a 2014 report.

Another company, Synlogic, is developing intratumorally injected bacteria designed to produce a STING (STImulator of INterferon Genes) agonist and act as an innate immune activator. The bacteria are sensed and engulfed by antigen-presenting cells that have infiltrated the tumor, and within those immune cells they activate the STING pathway, resulting in interferon release and tumor-specific T cell responses. A Phase 1 clinical trial is underway to evaluate this therapy for the treatment of refractory solid tumors, and trials for use in combination with a checkpoint inhibitor are planned. 

The results of these and other trials will serve to guide further innovations in safety and efficacy for engineered bacterial cancer therapies. For instance, these studies will shed light not only on therapeutic efficacy, but on bacterial colonization levels and distribution in patient tumors, shedding or off-target colonization, and stability of genetic modifications over time—factors that have only been studied at a detailed level in mouse models. Once a proof-of-principle is established in humans, there will be a big push to determine the optimal bacterial strain, payload, circuitry, and appropriate clinical settings in which to use these types of therapies.  

Building Bacteria to Fight Cancer

Synthetic biologists are applying new strategies in genetic engineering to encode traits and smart circuits in bacteria for more effective in vivo monitoring and drug delivery. At the same time, engineers are developing instruments for external control and guidance of bacteria with the aim of enhancing their ability to find and access tumors. Here are a few examples.


Bacterial bombs

Jeff Hasty of the University of California, San Diego, in collaboration with Sangeeta Bhatia of MIT (and T.D. in Bhatia’s lab), engineered an attenuated Salmonella enterica bacterial strain to synchronously release cancer therapeutics when the population reaches a critical density, allowing periodic drug delivery in mouse tumors. The effect is based on quorum lysis, meaning when a critical bacteria cell density is sensed by the population, they lyse and release the drug, while surviving bacteria keep proliferating until the critical threshold is reached again to repeat the cycle.

Encoded nanostructures for imaging

Mikhail Shapiro of the University of California, Berkeley, and colleagues encoded gas-filled nanostructures in microorganisms, including bacteria and archaea. These structures, when produced by the microbes, serve as contrast agents for ultrasound imaging, allowing researchers to visualize where they go in the body—critical for cancer diagnostics as well as to monitor treatment status by allowing researchers to visualize bacterial accumulation in tumors over time. The group recently demonstrated multiplexing of this approach by encoding a distinct reporter in each of two bacteria, E. coli and Salmonella, to localize and distinguish the microbe in the guts and tumors of mice.

Magnetically assisted navigation

Sylvain Martel of Polytechnique Montréal and colleagues attached drug-containing nanoliposomes onto a magnetotatic bacterial strain called MC-1 that was injected in close proximity to tumors in mice. These bacteria naturally biomineralize magnetic nanoparticles inside their membranes, allowing the researchers to use magnetic fields to guide the bacteria to—and into—tumors, where they can deliver therapeutics or serve as imaging contrast agents. 

Shining light on tumors

Di-Wei Zheng and colleagues at Wuhan University in China used light to enhance the metabolic activities of E. coli by attaching to the bacteria’s surfaces semiconductor nanomaterials that under light irradiation produce photoelectrons. These triggered a reaction with the bacteria’s endogenous nitrate molecules, increasing the formation and secretion of a cytotoxic form of nitric oxide by 37-fold. In a mouse model, the treatment led to an 80 percent reduction in tumor growth. 

See full infographic: WEB | PDF

Remote control guiding of bacteria to tumors

While researchers are succeeding in engineering bacteria to carry or produce anticancer compounds, fewer than 1 percent of those microbes will reach tumors on their own. Since most tumors are not accessible by direct injection, clinicians need to be able to effectively navigate bacterial therapies to tumor sites, where the microbes should reliably and controllably release the toxic drugs they encode. 

This is where synthetic biology has been influenced by the principles of microrobotics. For example, E. coli bacteria can be engineered with genes from marine microorganisms to sense and make use of light energy. In 2018, the University of Edinburgh’s Jochen Arlt and coworkers showed that such photosynthetic strains of motile E. coli could be guided through spatially patterned light fields. In response to patterns of light exposure, the bacteria moved to certain locations; tracking their position informed the next light input to guide them forward along a predefined path—a process that’s known as closed loop control, a fundamental part of robotics. 

New genetic toolkits are paving the way for the emerging fields of micro- and nanorobotics. 

In the same year, Xian-Zheng Zhang and colleagues at Wuhan University in China used light to locally trigger a 37-fold increase in bacterial cytotoxin production by attaching to the bacteria’s membranes nanomaterials that, upon light exposure, release photo-electrons that promote the toxin’s synthesis. In a mouse model of breast cancer, these anaerobic bacteria were found to accumulate in the hypoxic microenvironment of the tumors, and the subsequent light-boosted cytotoxin production resulted in around 80 percent inhibition of tumor growth. This is an example of how the integration of synthetic material into live bacteria can allow remote control of certain actions or functionality, another feature borrowed from classic robotics.

While optically triggered navigation and control has enormous potential, light’s limited ability to penetrate tissue hampers the approach. A more widely used form of external energy is ultrasound. It has long had applications in medical diagnostics and monitoring. More recently, gas-filled microbubbles, due to their strong and distinct acoustic response, are used to enhance contrast on ultrasound images of tissues, and special forms of high-powered, focused ultrasound have been applied in therapy to boost the transport of drug-filled nanobubbles by using the acoustic pressure waves as external energy to push them deep into tumor tissues. This approach achieved especially promising results in glioblastoma, because the blood-brain barrier is particularly hard to overcome for drugs. A couple of years ago, researchers used ultrasound to track therapeutic bacteria in vivo. Mikhail Shapiro and colleagues at Caltech genetically engineered bacteria to express what they termed acoustic reporter genes (ARG), which encode the components of hollow structures called gas vesicles that scatter ultrasound waves, generating an echo that enabled them to detect the bacteria’s location deep inside living mice.

Other common sources of external energy that can be safely and remotely applied in the human body are magnetic fields. While magnetic resonance imaging systems have been used clinically for decades, the development of systems for magnetic guidance and control are still fairly new. So far, researchers have applied the approach to guide magnetic catheters for high-precision surgery. The most renowned example is the NIOBE system from St. Louis–based Stereotaxis for the treatment of cardiac arrhythmias. A magnetic catheter tip is precisely steered along abnormal heart tissue, where electrical pulses heat or cool the device to ablate misfiring cells. 

The use of similar magnetic instrumentation to guide bacteria in the context of cancer therapy has been proposed by groups that work with magnetotatic bacteria—marine microbes that naturally synthesize strings of iron oxide nanoparticles wrapped in a lipid shell. This trait has evolved to help them navigate in the water by sensing the Earth’s magnetic field, with these strings working as compass needles inside their unicellular bodies. This was first discovered in the 1970s by Richard Blakemore of Woods Hole Oceanographic Institution in Massachusetts. Roughly 40 years later, Sylvain Martel of Polytechnique Montréal’s NanoRobotics Laboratory and colleagues coupled these magnetotactic bacteria to DOXIL, the liposome-wrapped chemotherapeutic that earned the title of the first approved nanomedicine. Martel’s group, too, took advantage of the fact that anaerobic bacteria tend to home to tumors for their low-oxygen environment, and coupled that natural homing mechanism with an external directing magnetic field, demonstrating increased accumulation and penetration of the therapy in mouse tumors. In another recent study, one of us (S.S.), with researchers at MIT and ETH Zurich, showed in tissue models on a chip that applying rotating magnetic fields could drive swarms of such magnetotactic bacteria to act as little propellers, creating strong flows to push companion nanomedicines out of blood vessels and deeper into tissues.

While the use of such magnetotactic species inside the human body might occur decades in the future, encoding magnetosensation in other, more clinically translatable or already-tested bacterial strains might be an achievable goal in the near term. Several of the proteins involved in the complex biomineralization process that forms the magnetic compounds in magneto-tactic bacteria have been identified, and in a preprint published earlier this year, researchers reported engineering E. coli to form magnetite particles and controlling them by external magnetic fields.

E. coli

Another route to making non-magnetic bacteria controllable by magnetic fields is to simply attach magnetic materials to them. Researchers have taken one or even multiple bacterial strains and bound them to magnetic micro- or nanoparticles. When exposed to an external magnetic field, these magnetic particles will orient with the field, and so will the bacteria, which will then travel in that direction. In 2017, Metin Sitti and colleagues at the Max Planck Institute for Intelligent Systems in Stuttgart, Germany, attached E. coli bacteria to microparticles made of layers of the chemotherapeutic doxorubicin and tiny magnetic nanoparticles. Using cancer cells in a dish, the researchers showed that they could remotely control these drug-carrying bacterial bots with magnets to improve tumor cell targeting compared with just adding drug-loaded microparticles to the cells.

No matter how, genetically engineered bacteria empowered by external energy sources providing triggers, control, and guidance are a fascinating new direction in this field. Fueled by the convergence of synthetic biology, mechanical engineering, and robotics, these new approaches might just bring us one step closer to the fantastic vision of tiny robots that seek and destroy many cancer types.  

Simone Schuerle is an assistant professor at ETH Zurich and a member of the university’s Institute for Translational Medicine. Tal Danino is an assistant professor at Columbia University and a member of the Herbert Irving Comprehensive Cancer Center and the Data Science Institute.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?