HHMI's Thomas M. Jessell:
High Concentration on Cell Differentiation

   One of the salient lessons to come out of developmental biology over the last two decades is that the mechanisms that control invertebrate development are conserved to an extraordinary extent, from fruit flies to humans. This lesson was put to the test in 1992 when researchers from Harvard, Columbia, Johns Hopkins, the University of California at San Francisco, and Oxford set out to find the mammalian equivalents of the hedgehog gene known to play an important role in pattern formation in the developing fruit fly. The result, within the course of a year, was the discovery of three new mammalian genes—known as sonic, indian, and desert hedgehog—and the realization that the proteins they coded for could account for a significant fraction of all the developmental interactions known to occur in the vertebrate embryo.

"The hedgehog protein provides a striking case in which a complex group of neurons that eventually function together are all induced by the same signaling molecule, just by its acting at different concentrations," says Thomas M. Jessell of the Howard Hughes Medical Institute. "It’s a very economical way of generating cell diversity."

   Among the major players in this developmental watershed was Thomas M. Jessell, a Howard Hughes Medical Institute investigator at Columbia University College of Physicians and Surgeons, New York, whose work helped elucidate the role sonic hedgehog plays in the development and differentiation of the spinal cord and nervous system. Jessell's remarkable influence in developmental biology is evidenced by a steady and methodical production of high-impact papers, including two dozen in the past decade with over 100 citations each, and six that have each been cited over 300 times. Along with Andrew MacMahon, Douglas A. Melton, and Cliff Tabin of Harvard, Jessell is also one of the founders of Ontogeny, Inc., a high-profile Cambridge, Massachusetts, biotech company that aims at using the approaches of developmental biology to come up with new treatments for diseases ranging from diabetes to Parkinson's.    Jessell, 47, received his undergraduate education at the University of London, and went on to get a master's at London Hospital. In 1977, he earned his Ph.D. in neuropharmacology from Cambridge University and in 1981, after three years of fellowships, became an assistant professor of neurobiology at Harvard Medical School. In 1985 Jessell moved to Columbia University College of Physicians and Surgeons, where he also became an Investigator at the Howard Hughes Medical Institute. Since 1989, Jessell has been professor in the Department of Biochemistry and Molecular Biophysics, at the Center for Neurobiology and Behavior at Columbia. From his office at Columbia, Jessell spoke to Science Watch correspondent Gary Taubes.

: How would you describe the overall aim of your laboratory?

Jessell: Our interest is in understanding how the nervous system functions—specifically, how different cell types in the nervous system actually become different. One basic problem is simply identifying these different cell types. At the very early stages of development, it’s impossible to recognize a motor neuron or a sensory neuron by their characteristic adult appearance. So we need to have an independent method of distinguishing different classes of cells. This is where transcription factor biology becomes very important, since the major molecular distinctions that exist between different classes of neurons can be recognized by their different expression of specific sets of transcription factors. So we can generate, for example, antibodies against these proteins. Then, at very early stages of development—just as the earliest precursor of the nervous system, the neural plate, is beginning to form the neural tube and the spinal cord—we can start to recognize molecular differences between different cell types. That is a key step in even defining an assay to identify the factors that control these distinctions in cell identity.

: Can you explain what you mean by transcription factor biology?

Jessell: The whole problem of developmental biology is how to turn specific sets of genes on in specific cell types. The way to activate gene expression is through transcription factors—the proteins that bind DNA and then lead to activation of target genes. Over the last decade, from thousands of different labs, there has been an explosion of information on different classes of transcription factors.    A breakthrough from our point of view was the observation than when one cell expresses a particular protein and its neighbor doesn't, it gives you a molecular handle to ask questions about how a given pattern of cell types is generated. We’ve spent a lot of time mapping the early expression of different classes of transcription factor. In particular, we’ve been focusing on the spinal cord, and we can now do a rather good job of accounting for each of the developmental stages, proceeding from a uniform group of precursor cells in the neural plate, where all the cells are the same, to the point halfway through nervous-system development where you have 20 different cell types.

: How do you find out when those transcription factors are turned on and exactly what they do?

Jessell: One type of experiment, for instance, is to take a piece of neural tissue at a time before it's been exposed to any environmental factors. You then examine the fate of those cells under that default program, so to speak. The next step is to add a potential source of an inductive signal and show that the program is changed—the cells become something else. That extrinsic signal might turn out to be, for example, hedgehog. An extrinsic signal is a protein that is excreted by one cell and which influences or changes the fate of a neighboring cell. That's a process of cell-cell communication. It changes the cell’s fate by turning one transcription factor on and turning another one off. This experimental method uses transcription factors as sort of molecular beacons to assay the existence of a diffusible or a cell-cell signal.

: The big breakthrough in your field was the discovery of the hedgehog genes in mammals. How has that work been developed?

Jessell: It turns out that there are three hedgehog genes in higher vertebrates, in mammals. Different groups cloned different hedgehogs early on. They're all interesting, but the one that is the most interesting to us is a gene that Cliff Tabin at Harvard eventually named sonic hedgehog. It turns out to be expressed in precisely those cells in the limb bud that have the limb-patterning activity. And it turns out to be expressed in just those cells under the spinal cord, in the notochord and later in the floorplate, that have the spinal-cord-patterning activity, which is what we were looking for.    By now, many papers have established the function and the necessity of hedgehog in these signaling processes. In the context of the spinal cord, one of the surprising things to emerge is that, at least in the ventral half of the spinal cord, which generates many different cell types, hedgehog is responsible for the generation of each of those cell types. It does this by acting at different concentration thresholds to establish different cell fates.

: Different concentration thresholds?

Jessell: There is now evidence that two-fold differences in the concentration of sonic hedgehog proteins to which a cell is exposed produce very dramatic changes in the eventual fate of that cell. So we can do in vitro assays where we can essentially dial in the concentration of hedgehog and generate a predictable cell type at each concentration. There are many precedents for this, both in Drosophila and in vertebrate development. The hedgehog example, however, provides a particularly striking case, where a complex group of neurons that eventually function together—motor neurons and different types of interneurons—are all induced by the same signaling molecule, just by its acting at different concentrations. It’s a very economical way of generating cell diversity from one signaling molecule.

: How do the other two hedgehog genes fit into the developmental picture?

Jessell: Desert and indian? They're expressed in different places in the body, but to a large extent it appears that each of the proteins can substitute for one another. Basically the general property of all three proteins is the ability to be provided by one cell type and influence or change the fate of a surrounding cell type. And you use these proteins repeatedly during development to control the development of many, many different tissues. Clearly, hedgehogs are not the only proteins. And there are many other classes of inducing molecules. Hedgehogs turn out to be one of the major families, and they have unmistakably important roles in many different systems.

: The company that you helped found, Ontogeny, was started shortly after the discovery of the hedgehog genes. What was the basis for its founding?

Jessell: The true impetus was primarily Doug Melton and Andy MacMahon at Harvard. Cliff Tabin came in slightly later. We were all working on different developmental systems. Doug was working on very early stages of mesoderm formation using the frog as a system. Andy was working on the kidney and nervous system, while I was working on the nervous system. But because of all the molecular genetics that had become available, it became clear that one could think about developmental biology as a rather unified field, and that the central role of inductive factors was inescapable. And, given that there was a relatively restricted set of induction factors, it seemed that there must be some application of those factors other than just basic scientific interest.

: How does this differ from what most biotech and pharmaceutical companies are doing?

Jessell: At that time most biotechnology research had really used molecular cloning as a major impetus to the generation of biopharmaceuticals. Genentech with t-PA, and Amgen with erythropoietin (EPO), are good examples. Those are not so dissimilar. t-PA is tissue plasminogen activator. It is a very successful Genentech product for immediate treatment of heart attacks. EPO is used in a variety of different conditions. Those companies essentially took a factor that was suspected of being involved in a disease, and cloned it. I think what we wanted to do is stay more closely wedded to the developmental process and understand an entire program of differentiation, rather than supplying one factor that ameliorates a particular condition.

: Can you give us an example of how understanding the developmental process might lead to a therapy?

Jessell: Probably the best example, and one that has a high probability of coming to practical fruition in the context of Ontogeny, is in the case of pancreatic differentiation and its applicability to diabetes. Diabetes clearly depends on the production of a single protein insulin from pancreatic BETA cells. Despite the availability of insulin by injection, it turns out that the long-term prognosis for diabetics is not so great because it is very difficult to stabilize the concentration of insulin through injection. If one could understand developmentally the mechanisms that control the generation of BETA cells, in the same way that one now understands, for example, the mechanisms that control motor neuron differentiation, then one might in principle be able to take a progenitor cell which exists even in the adult animal, expose it to the right combination of factors in the right regimen, and get that cell to turn into a BETA cell.    And so one can then think in a not-too-distant future about taking a BETA-cell progenitor cell, amplifying that cell so that you get a large stock of progenitor cells, and then exposing them to an appropriate differentiation signal that will turn them into BETA cells. The next step would be to use cell implantation methods to provide those cells as an endogenous source of insulin and thereby achieve a more stable glucose-insulin relationship.

: Doesn't this require that you have the necessary progenitor cells as well as the inducing molecules to differentiate them?

Jessell: There are two problems here, and this is an important distinction. There is no point in being able to take three adult progenitor cells and add a factor to get them to turn into BETA cells if the end result will be something like three BETA cells—such a small quantity is not going to be enough to do anything for anybody. So there are two steps to the process: one is to find a cell that has proliferative division capacity and then expand that cell, using the one cell to generate a hundred million cells, each of which retains the ability to respond to the inducing signal to turn into BETA cells. The second step is to make those cells do what you want them to do.    And so the stem-cell companies are now focusing on how to make a lot of these cells in the hope that once they’ve made them, something is going to come along and get the cells to differentiate. Whereas at Ontogeny, at least initially, we’re working to understand the process of getting a stem cell to do what we want.