It was a late night in 1984, and graduate student Xiaoning Wang was all alone in the lab. At First Military Medical University (now Southern Medical University) in Guangzhou, China, he was working at the lab’s only camera-equipped microscope, under a wooden hood constructed by his advisor, using a hair dryer to keep a cell culture of murine sarcoma cells at 37°C—mouse (and human) body temperature. He’d been doing the same thing nearly every night for the past month, hoping to capture on film what he had witnessed just a few weeks earlier—a tumor cell engulfing a live natural killer (NK) cell, a lymphocyte of the innate immune system and a major player in tumor suppression. After many sleepless nights, he had started to worry that perhaps he’d be unable to repeat the results.
“I watched day by day, night by night,” he recalls. He often saw the two cell types interact, but “most of the time, the lymphocyte killed the tumor cell.” But that fateful night it happened again. “I watched the whole process,” he says, “one cell entering another cell.” Controlling his excitement, Wang snapped pictures repeatedly as the two cell types came together, attached—and then the NK cell moved completely within the tumor cell. After another nerve-racking few hours of developing the film, Wang surveyed his pictures with pride. “I see success. I got the whole process from the beginning to the end. I think, I can graduate!”
Not having easy access to scientific literature in post-revolution China, Wang didn’t realize that he wasn’t the first to observe these so-called cell-in-cell structures (though he was one of the first to photograph the engulfment process). In fact, sporadic reports of live cells inside tumor cells date back to the beginning of the 20th century. With an internalized cell housed in a large vacuole, the strange cellular event was typically recognized by the crescent shape of the host-cell nucleus, squished along the cell’s perimeter by the intracellular structure. Over the last 100 years, lymphocytes, neutrophils, NK cells, and even other tumor cells have been found living temporarily within the cells of a variety of human cancers, including those of the skin, lung, breast, brain, pancreas, kidney, and blood. But until recently, little research has been devoted to how the cells got there and what consequences their engulfment has for tumor growth and survival. “For a long time, the cell-in-cell phenomenon had been overlooked in research,” Wang says.
Over the last 100 years, lymphocytes, neutrophils, NK cells, and even other tumor cells have been found living temporarily within the cells of a variety of human cancers.In the last few years, however, a handful of labs have begun to detail the process, documenting cell-in-cell structures in as many as 12 percent of tumor cells in vivo and 30 percent or more in vitro, depending on cancer type. In 2006, a group in Italy described metastatic melanoma cells that internalized T cells.[1. L. Lugini et al., “Cannibalism of live lymphocytes by human metastatic but not primary melanoma cells,” Cancer Research, 66:3629-38, 2006.] The following year, another group analyzed breast tumor cells ingesting other tumor cells.[2. M. Overholtzer et al., “A nonapoptotic cell death process, entosis, that occurs by cell-in-cell invasion,” Cell, 131:966-79, 2007.] And in 2009, Wang, now with his own lab complete with a proper hood, confirmed and added to his results from the 1980s, demonstrating that a variety of tumor cell lines harbor NK cells.[3. S. Wang et al., “Internalization of NK cells into tumor cells requires ezrin and leads to programmed cell-in-cell death,” Cell Research, 19:1350-62, 2009.]
The names given to the process include cell cannibalism, entosis, and emperipolesis. The molecular details of the engulfment seem to vary depending on cell type—as do the fate of internalized cells and, most likely, the consequences of the phenomenon, which have been proposed to range from tumor promotion to tumor suppression. But there is one thing most researchers tend to believe: cell engulfment is biologically important in tumor pathogenesis, and may hold potential for both the diagnosis and the treatment of cancer.
“Cannibalism is a very important discovery,” says Stefano Fais, an oncologist at the Istituto Superiore di Sanità in Italy, who led the study on T-cell-eating melanoma cells. “Cannibalism is a feature common probably to all malignant cancers.”
Who’s eating whom?In the summer of 2006, a small group of researchers gathered in Guangzhou, China, to discuss recent developments in stem cell research. One evening, after hosting a social event alongside the picturesque Pearl River for the meeting attendees, Wang—by then the dean of the School of Bioscience and Bioengineering at South China University of Technology in Guangzhou—decided to take the opportunity to share his work from grad school (which had been published in English in 1987, along with more detailed electron micrograph images of the cell engulfment activity[4. X. Wang and L. Wenjian, “Mechanisms of natural killer cell-mediated tumor cell cytolysis at a single cell level,” Journal of Medical Colleges of PLA, 2:107-17, 1987.]). In the dim light of a local pub that overlooked the river and the city, Wang showed the researchers the photos. At this time, Wang says, few people had ever heard of such cell-in-cell structures, and “they were very excited about this phenomenon.”
The images caught the eye of one researcher in particular, cell biologist Xuebiao Yao of the University of Science and Technology of China in Hefei. “I was surprised,” Yao recalls. “Seeing Dr. Wang’s early electronic micrographs reminded me of the technology used for hybridoma cell making, chemically putting two different kinds of cells together”—but Wang had not chemically induced this phenomenon. “I [wanted] to investigate whether this cell-in-cell process occurs in regular physiology or pathophysiology in [the] human body,” Yao says.
So the two arranged a collaboration to reignite Wang’s graduate work on tumor-cell engulfment. But among many other ongoing projects, the work got pushed to the side until the following year, when a group at Harvard Medical School published a paper detailing the engulfment of breast cancer cells by other breast cancer cells. Led by Michael Overholtzer and Joan Brugge, the group found that approximately one-quarter of breast cancer cells in a suspension culture contained internalized sibling tumor cells. The researchers detailed the engulfment process, which they termed entosis, noting that it required many elements of cytoskeleton regulation and cell-cell junctions, including actin, myosin II, cadherins and Rho signaling. Inhibiting any of these factors suppressed cell engulfment. Unexpectedly, when the researchers inhibited components of the Rho pathway in only a subset of tumor cells, they found that those cells were unable to enter other tumor cells, but they were still able to host internalized cells.2
“The mechanism really took us by surprise,” says Overholtzer, now at Memorial Sloan-Kettering Cancer Center in New York City. Unlike phagocytosis, the process by which macrophages take up dead cells and other extracellular materials for digestion and disposal, the cell that was being internalized seemed to take an active role in its own engulfment. “The data suggested that the cells were not being eaten at all,” he says. “They were invading.”
The researchers proposed that engulfment may be initiated after the cells detach from the extracellular matrix. This is a common occurrence among tumor cells in vivo, Overholtzer says, as cells proliferate from a single epithelial sheet to form a three-dimensional mass. Once they are freed from the forces attaching them in place, the dynamics holding cells together at cell-cell junctions take over. If there is an imbalance in these forces between two cells, with one cell pulling more strongly than the other, engulfment may result.
Indeed, anecdotal reports of cell-in-cell structures in tumors are documented in fluid samples. “We have noted that most of the malignant tumors [have] significant numbers of cell cannibalism, particularly in body fluids where the cells are floating, such as effusion fluids and urine samples, and also in breast carcinomas,” says cytopathologist Pranab Dey of the Postgraduate Institute of Medical Education and Research in India.
Though Overholtzer’s work clearly differs in some experimental details from Wang’s grad-school observations—Overholtzer’s group found that cancer cells engulf other cancer cells, while Wang observed the engulfment of immune cells—the results spurred Wang and Yao into action. They and their colleagues were able to re-document the phenomenon, finding internalized NK cells in up to 10 percent of tumor cells, depending on cancer type. Further characterizing a cell line from an epidermoid, or squamous cell, carcinoma, the researchers identified a role for E-cadherin in forming the connections between the NK and tumor cells prior to engulfment. They also identified a role for ezrin, a protein that links the plasma membrane with the actin cytoskeleton, in the internalization process.3
Like Overholtzer, Wang suspects that the internalized NK cells may play an active role in the process. The tumor cells he studied did not engulf dead lymphocytes, distinguishing the process from phagocytosis and suggesting that the tumor cells may respond to a signal produced by living NK cells. Furthermore, the knockdown of ezrin appeared to affect the fluidity and flexibility of the tumor cells’ plasma membranes, Wang explains, making them more rigid, potentially hindering NK cells from pushing into tumor cells. While siRNA-mediated knockdown of ezrin had no effect on the conjugation of the two cells, it inhibited NK cells from fully entering the tumor cells. Conversely, when the tumor cells were treated with reagents to activate the phosphorylation of ezrin, NK cell engulfment became more efficient.
Just a few years earlier, Fais’s group had similarly documented a role for ezrin in the engulfment of T cells by melanoma cells.1 Those researchers also identified another key player—a protein called caveolin-1, a main component of invaginations in the plasma membrane called caveolae, which research has suggested participate in the uptake of bacteria and viruses. “My idea is that ezrin binds to caveolin-1, thus connecting caveolae to actin and allowing the endocytic process [to proceed],” says Fais, who believes that T cells do not have an active role in engulfment. In the highly acidic tumor microenvironment, he says, “T cell activities are virtually abolished, [yet] T cells are equally cannibalized in both buffered and acidic media.” For this reason, Fais distinguishes this phenomenon as true cell cannibalism, with the host cell actively engulfing its victims, in contrast with the invasive aspect of entosis.
Despite this difference and the suspected role of caveolae in certain endocytic processes, Fais still argues that the cannibalistic activity he observed in melanoma cells is distinct from phagocytosis. His group documented T cells resting close to the cannibalistic melanoma cells before the tumor cells began to invaginate to capture the live lymphocytes. In contrast, phagocytosis is characterized by the formation of cellular extensions that embrace and engulf the external material. “Cannibalism looks to me like a swallowing of the external body,” says Fais, who proposes that it functions to provide nutrition to tumors. “Phagocytosis is to scavenge, not to feed. It’s really different.”
Hungry tumors or kamikazes?Though the mechanisms of engulfment may differ among cell and cancer types, the end result is nearly indistinguishable—a whole, living cell is housed within a large vacuole inside a tumor cell. Internalized cells usually follow one of three paths. They can continue living, at least temporarily, within the host cell, even dividing within their vacuole homes. Occasionally, they escape from the host cell to once again become a single, individual cell in the extracellular space. By and large, however, death is the most common fate for cells engulfed by tumor cells. Wang and colleagues demonstrated evidence of the apoptotic death of NK cells following their uptake by tumor cells. Nearly 90 percent of the internalized lymphocytes underwent traditional, programmed cell death, as evidenced by the activation of caspase 3. Work by Overholtzer’s group, on the other hand, suggests that cell engulfment between tumor cells represents a different type of cell death altogether—one mediated by lysosomes. The vacuoles housing the internalized tumor cells, his group observed, became acidic and surrounded by lysosomal membranes, indicative of fusion with lysosomes. Furthermore, when the researchers overexpressed B-cell lymphoma 2 (Bcl-2) to inhibit apoptosis, it had little effect on the death of internalized cells. On the other hand, inhibiting lysosomal acidification of the vacuoles could rescue the captured cells, when it was combined with apoptotic inhibitors. Interestingly, when only the lysosomal inhibitors were introduced, more cells appeared to undergo an apoptotic death, suggesting that apoptosis serves as a backup mechanism to the more common lysosomal death of the internalized cells.
Fais suggests it is simply the acidic environment of the tumor-cell vacuoles in metastatic melanoma cells that kills the internalized lymphocytes, though lytic enzymes may help to further digest the cell, he says. He argues that the engulfment and subsequent killing of cells such as lymphocytes is cell cannibalism in the most literal sense—one cell eating another. Once the victim is digested, the tumor cell can theoretically derive nutrients from it, promoting cancer survival and growth.
“We know that nutritional stress is a common feature of tumors,” says Eileen White, a cancer biologist at The Cancer Institute of New Jersey and Rutgers University. “We know they’ll undergo this process of autophagy where they’ll eat themselves. If they have the capability of eating each other or other cells—that would open a whole new door for tumors to sustain themselves.”
As evidence for this hypothesis, Fais showed in vitro that cell cannibalism increased under starvation conditions, and that the ingestion of T cells promoted the survival of melanoma cells. “The T cell is great because it has all these wonderful complex carbohydrates on the surface,” says cancer biologist Thomas Seyfried of Boston College. “They can all be degraded to glucose and other fuels [that tumor cells] could be using.”
But even if cells are deriving nourishment from their cannibalistic activities, it’s likely not the only benefit of the behavior, says immunologist Yufang Shi, who studies apoptosis at the Chinese Academy of Sciences and the Child Health Institute of New Jersey. “For one cell to digest another cell and to get energy . . . this is very uneconomical,” Shi explains. “You have to really make the cell into amino acids and polysaccharides. It’s very hard to use that as energy.” The fact that cell cannibalism increased when the cells were starving may simply be due to the fact that nutrient deprivation can cause cells to become detached from the extracellular matrix, Shi added—an event that Overholtzer’s group suggests could promote cell engulfment as a result of imbalanced cell-cell adhesion forces.
Another possibility is that the engulfed cells are driving the process. Internalized immune cells, for example, may have the potential to suppress tumor growth. During his initial graduate studies in the 1980s and again when he resumed this work more recently, Wang observed that some NK cells internalized by tumor cells can actually kill their host cells from the inside out. “After they enter into the tumor cells, they make the tumor cells erupt,” Wang says. “When [these NK cells] die, they also release a lot of enzymes,” Shi explains. “They are cytotoxic cells, so they can kill by releasing directly into the target cell, like the suicide bombers.”
But whether the internalized NK cells are initiating the engulfment is still unclear. If, on the other hand, the tumor cells are actively consuming the lymphocytes, it could provide a way for cancer to evade attack by the immune system. “I have a suspicion that maybe tumor [cells] in some conditions can kill the NK cells as a way to escape the surveillance of the immune system,” Wang says. This may become particularly important as the cancer metastasizes, Yao adds. “One of the physical challenges for those tumor cells will be how to survive in the new sites. One way is by taking [up] those NK cells and other immune cells to damage the immune response of cancer [patients].”
The bizarre phenomenon may also contribute to the genetic instability of cells, perhaps contributing to the formation of cancer early on. This March, Overholtzer and colleagues published the finding that cell-in-cell structures can act as cleavage barriers that disrupt cell division, leading to changes in ploidy—the number of sets of chromosomes in the cell—which are known to drive tumor progression.[6. M. Krajcovic et al., “A non-genetic route to aneuploidy in human cancers,” Nature Cell Biol, 13: 324-30, 2011.] Conversely, cell engulfment may act to suppress tumor growth, such as when tumor cells eat other tumor cells. “Entosis has a dual nature,” says Overholtzer. “It clearly can kill [tumor] cells, but also, it can disrupt ploidy—one is predicted to be tumor suppressive, one is tumor promoting.”
For now, the question of function remains “a puzzle,” Fais says, and “I don’t have all the pieces.” But with evidence growing for significance of cell engulfment in tumor pathogenesis, researchers are now considering whether the phenomenon could serve to aid in diagnosis or in the development of new cancer treatments. “I think in the next few years this will be a very active field,” Shi says.