The memory of a cell

Before the emergence of modern tools of molecular biology, cell biology, and pathology, nineteenth-century medical researchers could study cancer only by their own eyes, with the help of microscope at best. Like naturalists, they observed how the tumors looked macroscopically and microscopically, how they behaved in patients, animals, and petri dishes, and how they evolved in patients over time. Researchers looked for the patterns similar to the physiological or pathological phenomena they knew better: bacterial infections, parasites, or developmental aberrations, etc., and then considered the possible explanations.    

Looking at tumor tissue under the early microscopes of his era, the German pathologist Julius Cohnheim, a student of Rudolf Virchow, was struck by the crowdedly packed, proliferative, morphologically primitive cells of a tumor. They looked more like the cells of the developing embryo than like the mature adult tissue from which the tumor had supposedly arisen. In 1875, he formalized the “embryonal rest hypothesis”: adult organs harbor small remnants of embryonic cells left over from development. They normally lie dormant and silent, but that can be reactivated to give rise to cancer under the stimulated conditions such as injury or irritation [1]. The hypothesis was popular in pathology textbooks through the late nineteenth and early twentieth centuries and then fell out of fashion as genetics took center stage. The discovery of oncogenic viruses and inheritability of certain types of cancer seemed to offer a cleaner explanation of origin of cancer. Cohnheim's intuition would take another century to name precisely what “embryonic remnants” really was, and to do so would require biological tools and models that had not yet been invented.

Nevertheless, Cohnheim’s hypothesis could be described from the opposite direction: some embryonic tissues could demonstrate the abilities or functions of the cancer arisen from the mature tissues derived from them. After almost a century, the discoveries belonged to Nicole Le Douarin, a French developmental biologist. However, for her, it was a problem of embryology rather than oncology.

Since the nineteenth century, biologists had known of the neural crest, a strip of cells that appears briefly along the closing neural tube of an embryo and then disperses, giving rise to an astonishing range of tissues: nerves, ganglia, adrenal cells, facial cartilage and bone, and the pigment cells of skin and eye. At that time, however, no one could reliably follow an individual neural crest cell to see where it went or what it became, because once it dispersed into a tissue, there was no good method to track them.

Le Douarin's solution came not from a stain or a genetic marker but from an accident of avian anatomy. Quail and chick embryos develop on similar schedules and can be surgically grafted into one another, yet their cell nuclei are visibly different: quail cells carry a dense clump of heterochromatin that chick cells lack. Under the microscope, a quail cell would always look like a quail cell, no matter what chick tissue surrounded it. In 1973, Le Douarin used this quirk to build the quail–chick chimera: she transplanted neural crest tissue from a quail embryo into the corresponding site in a chick embryo, let development proceed, and then simply looked to see where the labeled cells had gone. Using this system to trace neural crest, Le Douarin showed that the "second brain" of the intestine: its dense mesh of neurons controlling digestion descended from this same fleeting population beside the neural tube.

In the following years, Le Douarin showed the amazing versality of neural crest, which derives autonomic nervous system along the length of the body axis, cartilage and bone of the vertebrate face and skull. She also identified melanoblasts, the precursor of melanocytes breaking off from neural crest, that migrated through body axes, disseminated, and invaded skin, hair follicles, inner ear, eyes, where they further differentiated to melanocytes. In 1982, Le Douarin gathered the field into a single volume, The Neural Crest. Some began calling the structure a fourth germ layer, instead of a derivative of ectoderm, mesoderm, or endoderm in the embryo. Her chimeras had also shown that developmental fate was never simply an internal script unspooling in isolation; the environment a cell moved through helped decide what it would become.

Le Douarin’s discoveries inspired later generations of melanoma researchers. Neural crest and melanoblasts reminded them so much of melanoma's later, uncanny talents: its willingness to travel, to switch identity under pressure, to colonize foreign organs. The genetic instructions for migration, for adaptability, for alternative identities in melanoma would probably be dated back to the time of melanoblasts and remain in their genomes.

While Le Douarin was mapping migration through time, a parallel idea was forming about geography. In 1978, Ray Schofield proposed the concept of the stem-cell niche: the notion that "stemness" was not a property a cell simply carried within itself, but something sustained by its location—a specific neighborhood of signals, contacts, and neighboring cells. Remove a stem cell from that neighborhood, and its behavior might change entirely; place it elsewhere, and it might differentiate, or die, or become something else altogether. Identity, in this view, was not a possession but an ongoing negotiation between a cell and its surroundings.

Schofield's niche completed something Le Douarin's chimeras had started. The chimera experiments had shown that an embryonic environment could steer a migrating cell toward a particular fate. The niche proposed that adult tissue kept enforcing that fate long after development ended, through a kind of continuous local discipline. Together, the two ideas carried an unspoken warning that neither man was writing about cancer at all when he arrived at it: if a microenvironment is what keeps old developmental potential locked away, then damage to that microenvironment—injury, inflammation, the chaos of a growing tumor—might be enough to release capacities the adult body had spent decades trying to forget. That warning would resurface first in the biology of the hair follicle, and then, unmistakably, in melanoma.

References

• Cohnheim, J. (1875). Congenitales, quergestreiftes Muskelsarkom der Nieren. Virchows Archiv für pathologische Anatomie und Physiologie und für klinische Medicin, 65(1), 64–69.
• Le Douarin, N. M. (1973). A biological cell labeling technique and its use in expermental embryology. Developmental Biology, 30(1), 217–222.
• Le Douarin, N. M., & Teillet, M.-A. (1973). The migration of neural crest cells to the wall of the digestive tract in avian embryo. Journal of Embryology and Experimental Morphology, 30(1), 31–48.
• Le Douarin, N. M., & Teillet, M.-A. (1974). Experimental analysis of the migration and differentiation of neuroblasts of the autonomic nervous system and of neurectodermal mesenchymal derivatives, using a biological cell marking technique. Developmental Biology, 41(1), 162–184.
• Le Lièvre, C. S., & Le Douarin, N. M. (1975). Mesenchymal derivatives of the neural crest: analysis of chimaeric quail and chick embryos. Journal of Embryology and Experimental Morphology, 34(1), 125–154.
• Le Douarin, N. M. (1982). The Neural Crest. Cambridge University Press.
• Schofield, R. (1978). The relationship between the spleen colony-forming cell and the haemopoietic stem cell. Blood Cells, 4(1–2), 7–25.

Seed and soil

In the autumn of 1889, a London surgeon named Stephen Paget sat alone with a stack of autopsy records and a growing sense in how cancer determined its next move. He had collected the postmortem reports from 735 women who had died of breast cancer. As he worked through them, a pattern emerged that defied the contemporary concept. The liver was almost always involved; the ovaries were disproportionately so; certain bones were frequently impacted. But the spleen — large, vascular, perpetually awash in blood — was nearly always clean. Rudolf Virchow, the towering figure of nineteenth-century pathology, had argued that cancer spread by a simple mechanical logic: cells broke free from the primary tumor, entered the circulation, and lodged wherever the current deposited them. If that were true, the spleen made no sense, and the distribution in his observation could not be explained.

Paget reached for a botanical metaphor, as Victorians were inclined to do. "When a plant goes to seed," he wrote in The Lancet, "its seeds are carried in all directions; but they can only live and grow if they fall on congenial soil." The breaking-off tumor cell was the seed, and the distant organ was the soil. And metastasis, Paget proposed, was not a matter of chance or anatomy but of compatibility. It is a specific, almost ecological relationship between a travelling cell and the tissue it hoped to inhabit. It was a beautiful idea but also controversial. For the following decades, the debate between Paget's biological hypothesis and the anatomical hypothesis was waged almost entirely with statistics drawn from autopsies. What the field needed was an experiment. The argument needed to wait for almost a century to be resolved.

In the 1960s, a young Israeli, Isaiah J. Fidler, came to the United States to study in school of veterinary medicine and then received a Ph.D. When starting his own laboratory at NCI-Frederick, Maryland, in 1970s, he focused his research on metastasis. NCI-Frederick was no ordinary campus. It was an autonomous hub for translational research, home to one of the largest mouse-study facilities in the country, and the place where landmark chemotherapy drugs, including paclitaxel, had been developed. For a scientist of Fidler's style and background — restless, rigorous, always asking for physiological and pathological relevance — it was exactly the right place.

One of his earliest acts there was to establish a cell line. The B16 melanoma had originated from a spontaneously arising tumor in an aged C57BL/6 mouse at the Jackson Laboratory, where it had been maintained by serial transplantation in vivo. Fidler made it a cell line that can be expanded stably in culture dish instead of in mice, so it became a reproducible scientific instrument. His first question came in 1973: are tumor cells interchangeable? If, like the contemporary idea, any cell in a growing mass was as capable of spreading and growing in the metastatic sites as any other. Using the B16 melanoma, he designed an experiment of almost irritating simplicity: inject cells into mice intravenously, wait for lung metastases to appear, harvest those colonies, culture them, and inject them again, cycle after cycle. What he found was that with each passage, the resulting population became more capable of forming metastases, more lethal as growing faster in the new site (Fidler, 1973). This result clearly demonstrated that metastatic site selected better fitting cells in each cycle.  

Nevertheless, how the cells were selected was debatable. It could be explained by two opposite ways. On one hand, the better fitting subclones might pre-exist in the tumor cell population. On the other hand, some subclones may randomly adapt to the environment of the metastatic site, and their adaptability was enhanced in each cycle of transplantation. The person who pushed Fidler toward an answer was his wife, immunologist Margaret Kripke, who had joined him at NCI-Frederick. Could all the cells in a culture dish truly be considered identical? The question rang a bell- how similar is this to Delbrück's question of origin of bacterial resistance to phages!

In 1943, at the dark climax of fascism with no sight of light at the end of the tunnel, Max Delbrück and Salvador Luria were thinking about an important biological question in their quiet office in Tennessee.  Bacteria develop resistance to phages due to genetic mutations. Do such mutations preexist or occur spontaneously after their exposure to the bacteriophages? In other words, is the resistance caused by the Darwinian selection or Lamarckian adaptation? 

At that time, the structure of DNA had yet to be discovered, genes are just concepts, invisible and intangible. Luria suggested that the inconsistency of the frequency of the resistance can be used to distinguish pre-existed or induced mutation. Inspired by Luria’s insight, Delbrück established a statistical model of mutation distribution models to predict the frequency of phage-resistant bacteria. Luria designed experiments based on Delbrück's statistical model.  The results confirmed that gene mutations occurred spontaneously in bacteria before their exposure to phages.  This is the first time in the history of biology that Darwin's theory of evolution by natural selection was proved experimentally. The dynamic duo was awarded the Nobel Prize in 1969. Today this model is well known as the "Luria-Delbrück distribution".

Years later, Fidler recalled that he immediately thought of “go Delbrück”.

Together, Fidler and Kripke generated 10 distinct sublines, each the descendant of a single cell from the B16 parental line. These were injected separately into groups of mice via the tail vein. The logic was straightforward: if metastatic ability were a pre-existing trait, different sublines would behave very differently from one another. If it arose through adaptation, they would converge on the same behavior. The results sided with Darwinism unambiguously: most sublines produced few metastases at all, and a handful generated hundreds of lung colonies. This was the Luria-Delbrück distribution described in the textbook, proving that the metastatic cells had pre-existed in the primary tumor (Fidler & Kripke, 1977). The paper was as concise as just two pages, but its conclusion could not be clearer.

The discoveries of selection by the metastatic site and pre-existing tumor heterogeneity were still not sufficient to prove Paget's hypothesis.  Why did the melanoma colonize the lung and not the kidney? Why the liver and not the spleen? It demanded the evidence of the soil; that is, certain organ environments specially equipped to receive melanoma cells. The anatomical hypothesis had always countered that drainage patterns alone could account for such preferences. 

Fidler, working now with Ian R. Hart, devised a strategy to test the organ environment. They transplanted small fragments of embryonic mouse lung and kidney tissue into the flanks of syngeneic mice (ectopic grafts). They were relocated from their anatomical position, but vascularized and viable. Then they injected B16 melanoma cells into the bloodstream and watched where the tumors grew. The cells colonized the native lungs, as expected; but they also colonized the grafted lung tissues sitting in the flank. In contrast, B16 cells did not appear in the ectopic kidney grafts sitting right beside the lung grafts (Hart & Fidler, 1980).  Paget's century-old conjecture, formed over Victorian autopsy tables, had been confirmed in the mouse flank of a laboratory in Frederick, Maryland. It is the organ microenvironment, its physiological character, biochemical hospitality, and the signals it sent and received, that determined where cancer grew. 

In 1983, Fidler and Kripke left NCI-Frederick for the University of Texas MD Anderson Cancer Center, where he became the founding chair of the Department of Cancer Biology. He would spend the rest of his career there, training scientists, extending the framework, and continuing to demonstrate that metastasis was not a mystery, but a problem to be solved. As he wrote in a 2003 review in Nature Reviews Cancer, the metastatic potential of any tumor cell is ultimately determined by its interactions with the homeostatic machinery of the distant organ (Fidler, 2003). The soil was not passive but an active participant.

The descendants of this insight are everywhere in modern oncology. Researchers have since shown that primary tumors send molecular emissaries ahead of themselves — secreted factors and extracellular vesicles that remodel distant organs before a single metastatic cell arrives, preparing the soil for the seed through what are now called pre-metastatic niches (Kaplan et al., 2005). Therapies targeting the tumor microenvironment, rather than the tumor cell alone, are a direct product of this intellectual lineage. The vocabulary has grown more molecular, the mechanisms more intricate — but the essential picture is still the one Paget drew in 1889, and the one Fidler spent his career proving: cancer spreads not by accident but by affinity. “In essence, everything I had done confirmed Paget’s hypothesis,” Dr. Fidler later said.

On May 8, 2020, MD Anderson announced the passing of Dr. Isaiah "Josh" Fidler. His legacies included not only the discoveries in the nature of cancer metastasis, but also his experimental approaches whose impacts on cancer research continue up to these days.  In vivo cycling has become a standard procedure to generate organ-specific metastatic cancer cell sublines, widely used in the study of organotropism of metastases. When researchers consider studying intratumoral heterogeneity, their first thought is to derive sublines from single cells of parental lines.  In the years that followed, B16 cell line and its sublines would become perhaps the most widely used mouse cancer cell lines in the history of biomedical research — a reagent so fundamental that decades later, Jim Allison would use B16 sublines to validate the anti-CTLA-4 immune checkpoint theory that would earn him the Nobel Prize in Physiology or Medicine in 2018.

When Dr. Fidler was asked, what does F in B16F0-10 sublines really stand for? "Fidler." He smirked. To these days, no one could tell whether it is a joke of Fidler style. Nevertheless, he did not have to mark his name on B16 sublines to let us remember his achievements. Cancer researchers have always told their trainees, “Do a Fidler experiment". 

References

1. Paget, S. (1889). The distribution of secondary growths in cancer of the breast. The Lancet, 133(3421), 571–573.
2. Fidler, I. J. (1973). Selection of successive tumour lines for metastasis. Nature New Biology, 242(118), 148–149.
3. Fidler, I. J., & Kripke, M. L. (1977). Metastasis results from preexisting variant cells within a malignant tumor. Science, 197(4306), 893–895.
4. Hart, I. R., & Fidler, I. J. (1980). Role of organ selectivity in the determination of metastatic patterns of the B16 melanoma. Cancer Research, 40(7), 2281–2287.
5. Fidler, I. J. (2003). The pathogenesis of cancer metastasis: the 'seed and soil' hypothesis revisited. Nature Reviews Cancer, 3(6), 453–458.
6. Kaplan, R. N., et al. (2005). VEGFR1-positive haematopoietic bone marrow progenitors initiate the pre-metastatic niche. Nature, 438(7069), 820–827.

The ultimate shapeshifter

Fidler and Kripke's experiments demonstrated metastatic ability belonged to particular subclones in melanoma, so it is a specific skill set instead of general activity. The next question was: what are these skills exactly? In other words, what kind of capacities should melanoma cells to equip so they can metastasize?

By the 1980s, cancer biologists agreed on what metastasis required, if not on how it happened. A malignant cell had to detach from its neighbors, cross the extracellular matrix, penetrate a vessel wall, survive the turbulence of circulation, exit into unfamiliar tissue, and colonize there. This process is very much like how an invasive species appears and takes over a new habitant. How could we study and understand invasiveness of a melanoma cell? Mary Jessica Charles Hendrix wondered.

In her career, Hendrix was always amazed about how cells and tissues change with the environment. She was trained as a developmental biologist. As a postdoctoral fellow at Harvard Medical School, she developed antibodies as research tools to address important developmental biology questions. She decided to dedicate to melanoma research after moving to the University of Arizona: “The Tucson area had the nation’s highest incidence of metastatic melanoma, and I wanted to set up my lab and try to make a difference in this disease. My passion for cancer research began in Arizona.” Her first action is to dissect the invasiveness of melanoma cells step by step.

Either leaving or entering a tissue, a cell needs to trespass the supportive material called extracellular matrix (ECM) in the tissue structure. A cell needs to attach ECM to migrate, but high density of ECM could also resist the cellular invasion. In 1987, she built a cell culture device (Membrane Invasion Culture System) to measure the efficiency of cancer cells trespass an ECM layer with defined thickness, converting the invasiveness into an experimental variable (Hendrix et al., 1987). Hendrix's group used the system to study the effects of ECM on cellular invasiveness. They found that ECM was not inert scaffolding; instead, it presented adhesion sites, growth factors, and chemical instructions to the migrating cells. The observation echoed Le Douarin's discoveries on neural crest chimeras: the route a migrating cell traveled helped decide what that cell would become.

Hendrix further applied this technology to search for drugs or physiological factors that change the invasiveness of the melanoma cells. In 1990, her team found that retinoic acid could suppress human melanoma invasion through a coordinated shift, including reducing the expression of barrier-dissolving enzymes, mobility-activating receptors, etc (Hendrix et al., 1990). In the following years, Hendrix and colleagues dissected how the cell physically gripped the matrix it crossed. They identified critical parts of the invasion machine, including a cell-surface receptor (an integrin), the ECM component grabbed by the receptor (vitronectin), and the enzyme degrading the ECM to clear the path for invasion (a matrix metalloproteinases, MMP). They also identified the components of cellular skeleton required for the invasive activity of melanoma cells. Interestingly, when they disrupted the components of the skeleton, the cells changed shape, from spreading crawler to steadier cobble look. In contrast, when they forced the melanoma cells to increase the amount of the skeleton components, the cells changed to more mobile shape. Moreover, the integrin was more than a scaffold claw. It can sense signals from the environment to induce the expression of the components of cellular skeleton and the matrix-degrading enzymes.

These discoveries revealed that the invasion machine ran in three-step sequence: (a) tumor cell attachment to ECM, (b) local degradation of the matrix by tumor cell-secreted proteases, and (c) tumor cell locomotion into the matrix modified by proteolysis. Moreover, the alteration in cellular skeleton components can change cellular morphology and invasiveness at the same time, implying that melanoma cells may be able to switch to different statuses if the cells can sense the environment through the receptors like integrins. If this is true, such adaptability can explain why melanoma cells are so tough to kill. As Hendrix’s team kept investigating the invasiveness, the story of adaptability would be unfolded to them in an unexpected way.

In the early 1990s, Robert Folberg was a pathologist in Department of Ophthalmology, University of Iowa. One of his routine tasks was to examine the pathological section of uveal melanomas, cancers of the eye's pigment cells. He had been noticing an interesting structure in those tumors for a while: channels looked like blood vessels but lacked the detailed features of the latter. Or, more precisely, the cells forming the channels were seemingly cancer cells. He identified nine specific vascular patterns in these tumors and showed that the presence of vascular networks was strongly associated with death from metastatic melanoma. The result was published in Ophthalmology in 1993.

The paper did not receive much attention, partly because it was a very specialized topic. Also, in the paper, Folberg did not discuss his suspicion that the channels were formed by cancer cells, because the idea was against the contemporary paradigm. For decades, Judah Folkman's theory of angiogenesis explained how tumors acquired blood supply: a growing cancer secretes signals that recruit the host's own endothelial cells, which then construct new vessels to feed it. Block angiogenesis, in principle one could starve the tumor. Moreover, it seemed hard to imagine that melanoma cells change their functions like turning a switch. However, Folbert knew that he was not the only one who observed and thought that. The observation and idea had been circulating in the circle of pathologists for a while that melanoma cells could build their own perfusion network just by themselves. Only if he could design a study in a good research laboratory, he would prove it.

Folberg’s opportunity finally arrived when Hendrix moved to the University of Iowa in 1996. Hendrix realized the potential immediately: the tools, methods, and even concepts developed in her laboratory over the past decade were perfect to answer this novel question. Folberg first showed that the channels in melanoma had different morphology from normal blood vessels, nevertheless permitted the flow of red blood cells. Interestingly, they expressed VE-cadherin (a marker of endothelial cells, the building blocks of blood vessels) together with melanoma specific markers, but lacked most of the standard markers of endothelial cells. This was good evidence, but Hendrix and Folberg wanted to test whether the channels were actively constructed by melanoma cells, or formed passively like the water scouring down the hill. She recruited Andrew J. Maniotis, who was an expert in angiogenesis, to join this project. In fact, he had done postdoctoral research in Folkman’s laboratory. Maniotis seeded aggressive and non-aggressive uveal melanoma cell lines onto Matrigel, a ECM material frequently used in 3D cell culture. Within days, aggressive melanoma cells reconstituted the same looping, interconnected architecture seen in uveal melanoma tissues from patients. These channels permit the perfusion of dye, like Folberg’s observation. In contrast, non-invasive melanoma cells from the same patients formed nothing.

The difference in the channel-forming ability between aggressive and non-aggressive melanoma cells intrigued Hendrix very much. She turned to Paul Meltzer, a renowned expert in the arising field of human genomics, for help. Meltzer applied the most advanced technology at that time- microarray- to compare poorly invasive and highly invasive melanoma cells from the same patients across five thousand genes. The aggressive cells, it seemed, had reverted toward a pluripotent, embryonic-like genotype, one echoing the way placental cytotrophoblasts adopt a vascular phenotype to build the earliest circulation of a human embryo.

Taken together, the evidence forced a reckoning the discipline had not anticipated. The patterned channels of aggressive melanoma were histologically, ultrastructurally, and immunologically distinct from angiogenic vessels; those tumor cells carried a genetic signature borrowed from embryonic vasculogenesis. Folberg, Maniotis, and Hendrix named the phenomenon vasculogenic mimicry—the tumor forming its own paravascular perfusion pathways.

Hendrix and her team submitted the manuscript of this study to Science. However, they had a hard time to convince the editor and reviewers to accept this paradigm-shifting concept, and it was eventually rejected. Subsequently, this study was published in the American Journal of Pathology. Even though, the response split the field. Mina Bissell was open to the possibility that tumor plasticity had produced an unrecognized mode of perfusion. Donald McDonald and others argued the channels could be artifacts of tissue processing or hemorrhage — blood leaking through gaps in malformed but ordinary endothelial vessels — and Robert Kerbel was associated with the same skeptical position. The objection was scientifically fair; cancer research has a long history of apparent revolutions that dissolve under sharper staining or tighter controls. Yet Isaiah Fidler, Avraham Raz, and Folkman himself remained receptive, and pathologists recalled older, unexplained "blood lakes" in tumors that had never sat comfortably within a purely endothelial account. Underneath a dispute that looked like it was about vascular anatomy sat a deeper one about cellular identity: if a melanoma cell could become vessel-like, differentiation was not the irreversible endpoint biologists had assumed.

In March 2000, this simmering disagreement finally came to a head in a Salt Lake City hotel auditorium, where hundreds of researchers gathered for a public showdown at a Keystone meeting. Hendrix faced off against Donald McDonald of UC San Francisco, who remained convinced her team had misread its own data. Hendrix later described the atmosphere as a boxing match with a championship belt on the line. The stakes were real: while angiogenesis—tumors recruiting host endothelial cells to build new vessels—was well established, Hendrix's team insisted tumor cells could also construct their own blood-delivering channels, a mechanism they had named vasculogenic mimicry. Their 1999 paper had already stirred considerable upheaval in the field. At the debate, neither side landed a knockout. Hendrix maintained the loops and networks were a genuine mini-circulatory system built by the tumor cells themselves; McDonald insisted they were merely folds of connective tissue, not true vascular channels. No consensus emerged that day. But in the years that followed, the sharpest edges of the controversy softened, as researchers gradually assembled a fuller picture of how tumors fashion their own vasculature and how this capacity shapes prognosis and treatment—even as a handful of skeptics still find the whole idea unsettling.

References

• Hendrix, M. J. C., Seftor, E. A., Seftor, R. E. B., & Fidler, I. J. (1987). A simple quantitative assay for studying the invasive potential of high and low human metastatic variants. Cancer Letters, 38(1–2), 137–147.
• Hendrix, M. J. C., Seftor, E. A., Seftor, R. E. B., Fidler, I. J., & Meyskens, F. L. (1990). Retinoic acid-induced modulation of human melanoma cell invasion. Journal of the National Cancer Institute, 82(20), 1656–1657.
• Welch, D. R., Bisi, J. E., Miller, B. E., Conaway, D., Seftor, E. A., Yohem, K. H., Gilmore, L. B., Seftor, R. E. B., Nakajima, M., & Hendrix, M. J. C. (1991). Characterization of a highly invasive and spontaneously metastatic human malignant melanoma cell line. International Journal of Cancer, 47(2), 227–237.
• Hendrix, M. J. C., Seftor, E. A., Chu, Y. W., Seftor, R. E. B., Nagle, R. B., McDaniel, K. M., Leong, S. P., Yohem, K. H., Leibovitz, A. M., & Meyskens, F. L. (1992). Coexpression of vimentin and keratins by human melanoma tumor cells: correlation with invasive and metastatic potential. Journal of the National Cancer Institute, 84(3), 165–174.
• Seftor, R. E. B., Seftor, E. A., Gehlsen, K. R., Stetler-Stevenson, W. G., Brown, P. D., Ruoslahti, E., & Hendrix, M. J. C. (1992). Role of the αvβ3 integrin in human melanoma cell invasion. Proceedings of the National Academy of Sciences, 89(5), 1557–1561.
• Hendrix, M. J. C., Seftor, E. A., Chu, Y. W., Trevor, K. T., & Seftor, R. E. B. (1996). Role of intermediate filaments in migration, invasion, and metastasis. Cancer and Metastasis Reviews, 15(4), 507–525.
• Hendrix, M. J. C., Seftor, R. E. B., Seftor, E. A., Gruman, L. M., Lee, L. M. L., Nickoloff, B. J., Miele, L., Sheriff, D. D., & Schatteman, G. C. (2002). Transendothelial function of human metastatic melanoma cells: Role of the microenvironment in cell-fate determination. Cancer Research, 62(3), 665–668.

The verdict delivered

As the skepticism demanded more evidence from different perspectives, Hendrix was trying to identify the environmental cue to induce vascular mimicry of melanoma in vivo. She reasoned that this function should be needed when melanoma cells grow in an oxygen deprived environment, so she chose an old, stressful model: the ischemic limb by severing a mouse's femoral artery and tying off its branches. Hendrix's team injected two sister melanoma cell lines, one aggressive and the other non-aggressive, into the ischemic muscle. Five days after the injury, the aggressive melanoma cells occupied the vessel's inner, blood-facing surface, exactly like part of the blood vessels. The nonaggressive cells, in contrast, vanished within five days. Even more interesting, aggressive melanoma cells expressed genes required for formation of embryonic vasculature.  Twenty days after the injury, once the vasculature had stabilized, the melanoma cells had largely vacated the vessel walls; ordinary melanoma tumors began forming elsewhere in the muscle instead. The embryonic-like profile from Meltzer's microarray proved to be an operational toolkit of melanoma cells. They opened and used it when needed and closed it when getting back to regular environment.

In the following years, Hendrix’s team collaborated with Meltzer to compare gene expression profiles of more non-aggressive and aggressive melanoma. They found that, while the former were recognizably melanocytic, the latter cells were in an uncommitted or developmentally regressed state, expressing melanocytic, endothelial, epithelial, and neural progenitor genes. Furthermore, she and collaborators showed how melanoma cells adapt to the environment by switching their identity. In one study, human metastatic melanoma cells were transplanted into the neural tube of chick embryos. They did not form tumors at all. Instead, they dispersed along the host's own migratory pathways, arriving unbidden at the branchial arches, the dorsal root ganglia, the sympathetic ganglia - exactly where neural crest cells travel in normal development (Kulesa et al., 2006). In another model, an aggressive human melanoma line was introduced into zebrafish embryos at the blastula stage. The embryos responded by growing duplicate body axes, as if the tumor had seized the master switch of development itself (Topczewska et al., 2006). The agent responsible for this was Nodal, an ancient morphogen that embryos use to distinguish head from tail; silencing it stripped the melanoma of its borrowed power and coaxed the cells back toward pigment production, back toward the melanocyte they had once been.

Over time, more and more subsequent research across the field confirmed that the plasticity in melanoma was indeed a central engine of the disease's lethality, with concrete studies now linking metastasis, therapy resistance, and recurrence to the reactivation of different neural crest-derived programs. Researchers found that melanoma cells could reversibly toggle between a differentiated, pigmented, proliferative state and an undifferentiated, invasive one, a phenomenon formalized as "phenotype switching" and governed largely by the transcription factor MITF (Hoek et al., 2008). Studies have shown that melanoma cells underwent such switching in different environments during metastasis (Karras et al., 2022). Conversion to slow-cycling, stem-like state helps melanoma cells to survive chemotherapy, target therapy, and other drug treatment (Rambow et al., 2018; Roesch et al., 2013). Radiotherapy resistance of has also been linked to stem-like subpopulations that resist DNA-damage-induced apoptosis and retain heightened repair capacity (Marzagalli et al., 2019; Confronting Melanoma Radioresistance, 2025). Parallel work extended the pattern to immunotherapy, showing that dedifferentiated, neural crest stem cell-like states correlate with poor checkpoint-inhibitor response by reshaping the tumor microenvironment (Falletta et al., 2017; Karras et al., 2022).

Taken together, this body of work closed the circle Hendrix's research had opened decades earlier: the embryonic toolkit that let melanoma cells build their own vasculature was the same one that let them invade, resist every major class of treatment, and lie dormant before returning. Her insistence that melanoma cells were not broken melanocytes but developmentally regressed, environmentally responsive entities, was not a peripheral observation but the conceptual seed of an entire research program—one that now shapes how oncologists think about targeting the most adaptable, and most dangerous, cancer cells. In 2016, seventeen years after the manuscript of melanoma vascular mimicry was rejected by Science, the journal published a commentary to recognize the importance of this function in melanoma biology, as well as to highlight a first-in-class drug, Foslinanib (CVM-1118) that specifically targets vasculogenic mimicry.  

References

• Hoek, K.S., et al. (2008). Pigment Cell & Melanoma Research, 21(6), 665–676.
• Kaufman, C.K., et al. (2016). A zebrafish melanoma model reveals emergence of neural crest identity during melanoma initiation. Science, 351(6272), aad2197.
• Kulesa, P.M., et al. (2006). Reprogramming metastatic melanoma cells to assume a neural crest cell-like phenotype in a chick embryo microenvironment. PNAS, 103(10), 3752–3757.
• Topczewska, J.M., et al. (2006). Embryonic and tumorigenic pathways converge via Nodal signaling: role in melanoma aggressiveness. Nature Medicine, 12(8), 925–932.
• Karras, P., et al. (2022). A cellular hierarchy in melanoma uncouples growth and metastasis. Nature, 610, 190–198.
• Rambow, F., Rogiers, A., Marin-Bejar, O., et al. (2018). Toward minimal residual disease-directed therapy in melanoma. Cell, 174(4), 843–855.
• Marin-Bejar, O., et al. (2021). Evolutionary predictability of genetic versus nongenetic resistance to anticancer drugs in melanoma. Cancer Cell, 39(8), 1135–1149.
• Roesch, A., et al. (2013). Overcoming intrinsic multidrug resistance in melanoma by blocking the mitochondrial respiratory chain of slow-cycling JARID1B(high) cells. Cancer Cell, 23(6), 811–825.
• Marzagalli, M., et al. (2019). Cellular and molecular biology of cancer stem cells in melanoma: possible therapeutic implications. Seminars in Cancer Biology, 59, 221–235.
• Confronting Melanoma Radioresistance: Mechanisms and Therapeutic Strategies. (2025). Cancers, 17(16), 2648.
• Falletta, P., et al. (2017). Translation reprogramming is an evolutionarily conserved driver of phenotypic plasticity and therapeutic resistance in melanoma. Genes & Development, 31(1), 18–33.
• Ossowski, L., & Aguirre-Ghiso, J.A. (2010). Dormancy of metastatic melanoma. Pigment Cell & Melanoma Research, 23(1), 41–56.