(5)

By the late 1980s, something was bothering Herlyn. The cell lines were invaluable — he did not doubt that for a moment. But there was a limitation built into the model that he had been aware of for some time and could no longer set aside.

Melanoma cells in a tissue culture flask grow as a flat monolayer on plastic. They are bathed in a chemically defined medium that bears no resemblance to the biochemical environment of the skin. They have no neighbors of other cell types. They do not experience the mechanical cues, the oxygen gradients, the paracrine signaling, or the spatial organization of an actual tumor growing in actual tissue. In a two-dimensional monolayer, the movement of melanoma cells of radial growth phase and vertical growth phase could not be distinguished, because those behaviors are inherently three-dimensional. RGP and VGP are spatial description. You cannot study invasion in a system with no depth.

What Herlyn needed was a way to restore enough of the skin architecture that the cells could behave as they did in the body, at least to a biologically meaningful degree. The technology he reached for had not been invented for melanoma. Reconstructed human skin had been developed in the context of wound healing, as a potential graft material for burn patients. It was a medical engineering achievement, and by the late 1980s several groups had produced reliable protocols for creating it. The resulting structures were not perfect replicas of living skin, but they were impressively faithful: a dermal layer of fibroblast-contracted collagen, an epidermal layer of differentiated keratinocytes sitting above it, a developing basement membrane at the interface between them, and an air-exposed surface that induced the keratinocytes to stratify and differentiate as they would in the body.

Herlyn's team was the first to ask what would happen if you added melanocytes — or melanoma cells — to this system. When melanoma cells were introduced into this system, they behaved in ways that corresponded precisely to what Clark had described from the microscope decades earlier. Normal melanocytes and RGP melanoma cells integrated into the epidermis and stayed there, dispersing among the keratinocytes in the basal layer, as normal melanocytes do in living skin. They kept a distance from basement membrane, the boundary between epidermis and dermis that normal cells respected absolutely. VGP melanoma cells began to push downward. They disrupted the basement membrane, protruding into the collagen matrix below, moving with a tentative but unmistakable directionality. Metastatic melanoma cells invaded with efficiency. They crossed the basement membrane, proliferated in the collagen matrix of the artificial dermis, organized into nests and cords, patterns recognizable to any pathologist who had looked at invaded tissue. They behaved, in a dish, the way they behaved in a patient.

The skin reconstruct confirmed something important: the biological differences between tumor stages that Herlyn had identified in flat culture, and that Clark had identified in sectioned tissue, were not artifacts of either system. They were real properties of the cells, expressed consistently across contexts. The cells carried their stage identity with them, and that identity governed how they behaved when they were given back the spatial freedom to act on it.

More than confirmation, though, the skin reconstruct was an engine of discovery. It allowed questions about invasion — arguably the most therapeutically important property of a melanoma cell — to be asked meaningfully. Using the system, Herlyn's team introduced bFGF into RGP cells by gene transfer, those cells, which had previously sat quietly in the epidermis of the reconstruct, began to invade. The biology of the flat flasks and the biology of the three-dimensional tissue told the same story, but the three-dimensional tissue made the causal logic legible.

(6)

Through all of this — the cell lines, the skin reconstructs, the growth factor biology, the dormancy hypothesis, the therapeutic antibodies — Herlyn had been doing something that was easy to overlook from the outside: he had been building an institution as much as a collection of scientific results.

For more than twenty years after arriving at Wistar in 1976, Herlyn had never directly encountered a melanoma patient. His work had been conducted entirely at the level of cells and tissues and animal models. This is was the way that basic researchers believed,  that the deepest understanding of a disease came from the most fundamental questions, from knowing the cell, its biology, its vulnerabilities behind the noisy complexity of a patient's symptoms or a clinical trial.

But in 2003, something shifted. The Foundation for Melanoma Research approached him about organizing a scientific congress — the first meeting ever devoted specifically to bringing together melanoma researchers from across disciplines: basic scientists, clinical oncologists, pathologists, immunologists, epidemiologists, and patients. It was in preparing for this meeting that Herlyn, for the first time, met people who were living with the disease he had studied for a quarter of a century. He sat with patients and their families and heard accounts of diagnosis, of treatment, of the peculiar hope and exhaustion that accompany a melanoma prognosis. He listened to what it felt like to be told that the drug that had seemed to be working had stopped working — that the slow-growing, drug-resistant cells he had identified in his laboratory cultures were real in a way he had understood intellectually but not experientially. The encounter changed something in how he talked about his work, if not in the work itself.

The first International Melanoma Research Congress was held in Philadelphia in 2003. The following year, Herlyn co-founded the Society for Melanoma Research and served as its first president. It is the first organization dedicated exclusively to uniting the global melanoma research community across the divide between laboratory and clinic. It was, in its way, a continuation of the principle that had organized the WM cell line collection and that had sustained the triangular relationship with Halaban and Bennett for two decades: that resources held only by one person were resources wasted, and that science advanced by sharing, not hoarding.

By the end of the twentieth century and into the first decade of the twenty-first, the framework that Herlyn, Halaban, and Bennett had spent decades constructing  was bearing fruit in ways that would have seemed extravagant to anyone in 1976. It is worth pausing to consider what kind of scientist Herlyn, Halaban, and Bennet turned out to be, because the answer is not entirely obvious from a list of their brilliant publications. They were builders. They built the foundations  that other people's research were based on — cell line collections, skin reconstruct protocols, a research congress, a scientific society. 

"We need models," Herlyn said simply, "because we cannot do everything in patients."

That sentence carried everything: the recognition of disease as a human reality, the acknowledgment of human limits, the practical insistence on building systems that could substitute for what could not ethically or logistically be done directly. It was the sentence of a veterinarian turned microbiologist turned cancer researcher — a scientist who had spent his career translating the visible into the workable, the clinical into the biological, the map into the territory.