One team of scientists has coaxed mouse embryonic stem (ES) cells to grow into motor neurons—the specialized nerve cells that control the movement of muscles. When the team inserted these newly made motor neurons into a chick embryo's spinal cord, they found that the neurons grew long extensions called axons and made contact with muscle cells just as well as the embryo's own motor neurons did. This extraordinary finding by HHMI investigator Thomas M. Jessell at Columbia University's College of Physicians and Surgeons and his colleagues was given early publication online by Cell on July 17, 2002, and was published in the August 9 issue. It raised hopes that several incurable muscle diseases, including amyotrophic lateral sclerosis (ALS, also known as Lou Gehrig's disease), as well as spinal cord injuries, might eventually be treated with such cells.

Along similar lines, Ron McKay and his team at the National Institute of Neurological Disorders and Stroke in Bethesda, Maryland, have successfully directed mouse ES cells to become neurons that release dopamine, the chemical lacking in Parkinson's disease. The scientists also showed that these neurons restore some aspects of mobility in a rat model of Parkinson's, a possible harbinger of more effective therapies for this common disease.

All stem cells are immature, unspecialized cells with great potential, but some are more limited than others. ES cells—those derived from very early embryos—can become any type of cell in an organism and can renew themselves indefinitely. But adult stem cells have already started on a particular pathway of differentiation. They can still generate a variety of cell types, but the choices are restricted.

HHMI investigators Allan C. Spradling, Sean J. Morrison and Mark T. Keating (see sidebar "Will Humans Generate Replacement Parts?") and others are examining how adult stem cells might be made to treat diseases. The therapeutic potential of adult stem cells is particularly important because of current U.S. restrictions on research involving human ES cells.

THE RIGHT SIGNALS
For his systematic, decade-long study of how the nervous system functions—"specifically, how different cell types in the vertebrate nervous system actually become different"— Jessell relied initially on stem cells from embryonic neural tissue. At first, he could not even tell these neural progenitor cells apart. "At the very early stages of development, it's impossible to recognize a motor neuron or a sensory neuron by its appearance," he explains. But each class of neurons turns on a different set of genes. These genes are activated by different sets of transcription factors, proteins whose signals turn on specific genes in a cell's nucleus. Because these factors could be recognized by specific antibodies that were available to him, Jessell soon learned to identify various types of neurons at their earliest stages.

While studying the early developmental signals that normally produce motor neurons, Jessell decided to find out whether "naïve ES cells" could be converted into motor neurons by means of the same signals. With great satisfaction, he now reports that the answer is yes—at least in mice. Hynek Wichterle, a Czech postdoctoral fellow in Jessell's lab, recently proved it in an experiment that "demonstrates one can learn from embryonic development and apply it directly to ES cells. Just by exposing the mouse ES cells to developmentally relevant signals, we drove them to differentiate all the way to motor neurons."

Wichterle notes that he had "a huge chunk of luck": He managed to produce all these changes by using only two signals. One is retinoic acid, a product of vitamin A that binds to receptors in the nucleus. The other is Sonic hedgehog, the product of a gene that was first identified in developing fruit flies and that Jessell found expressed in the mouse spinal cord, where it plays a critical role in the differentiation of motor neurons.

"There are three main steps that an ES cell needs to take in order to become a motor neuron," says Jessell. "The first is the 'decision' to become a generic neural progenitor cell. The second is the decision to become a spinal cord progenitor cell. The third is the decision to become a particular progenitor cell that gives rise to a motor neuron." The retinoids "are good at converting neural progenitor cells into spinal progenitor cells, and then hedgehog is good at converting spinal progenitor cells into the kind of cells that are precursors of the motor neurons," Jessell says. "Where we got lucky is with the first step. For some reason, ES cells tend to become nerve cells almost by default, as first suggested by HHMI investigator Douglas Melton (at Harvard University)."

The third step was not so simple, however, because different concentrations, or grades, of sonic hedgehog have different effects. "Hedgehog's graded actions normally produce at least five different classes of neurons," Jessell points out. "Typically, only 25-30 percent of the cells that emerge from exposure to sonic hedgehog are motor neurons. I think it's going to be essentially impossible, in the context of neural stem cells, ever to find situations where 100 percent of the generated cells are of one particular neuronal subtype."

To get around this problem, he explains, "we used a genetic trick. Wichterle, together with postdoctoral fellow Ivo Lieberam, marked the motor neurons with green fluorescent protein. Then by putting all the cells through a fluorescence-activated cell sorter, we could separate the fluorescent ones from the others and get essentially 100 percent purity."

RATIONAL DESIGN
Ultimately, Jessell is seeking precision, control. He points to several clinical trials in which people with Parkinson's disease were treated with human cells that produce dopamine, "not of ES cell origin but of fetal brain origin." A few patients benefited from these attempts, but some actually got worse. "That may be because the cells introduced into those patients were a heterogeneous mixture," Jessell says. "It's not surprising that the clinical outcome is variable if you cannot control the precise number or proportion of dopamine-producing cells that you are putting in."

Such control would be even more important in dealing with other diseases, he believes. "In Parkinson's, at least you know that you really need dopamine-producing neurons," he says. "But in other diseases, it is not clear which cell type is needed to reconstruct a circuit or prevent neural degeneration. In the context of ALS, for instance, do you want to put in motor neurons or would you be better off with spinal interneurons or glial cells, which normally surround the motor neurons and may actually support the survival of the remaining ones?" Or, for that matter, should all classes of spinal cord cells be used?

Having a way to purify the appropriate cells "puts you in a position now where you can ask such questions," Jessell says. He is also pleased that "we have manipulated mouse ES cells simply by exposing them to different environmental signals—we have not changed the genetic makeup of the ES cell itself," he says. While some other scientists introduce genes directly into the ES cell, "we have just added factors that we know are involved in the normal developmental process and let the ES cell do the rest." As a result, Jessell has gained a tool that might eventually be used with human ES cells as well.

"If we wanted to apply this strategy now to a human ES cell, we could simply take the same factors," Jessell says. "It turns out that the two factors we use—Sonic hedgehog and retinoic acid—cross species barriers. Furthermore, one of our collaborators, Jeff Porter at Curis, in Cambridge, Massachusetts, has identified a small molecule that activates the hedgehog pathway, so we can now direct ES cells to become motor neurons simply with small, synthetic chemicals. You don't even need mystery factors anymore."

Eager to test whether the neurons they grew from mouse ES cells were functional motor neurons, Wichterle decided to put them into live chick embryos. Why chicks? "Because their embryos grow in eggs rather than wombs," Jessell explains. This makes it much easier to gain access to the neural tube and "introduce the ES cell-derived motor neurons into the spinal cord at exactly the same time that the host motor neurons are being generated." In a sense, Jessell says, "we are giving the ES cell-derived motor neuron an even chance."

The scientists found that "even in this completely different host species, the ES cell-derived motor neurons settle in the right place in the spinal cord, extend axons out and form differentiated synapses with target muscle" over the same time course as the chick's own motor neurons. In fact, Jessell says, "it was a dead heat." The fluorescent green markers continued to function, so he could follow the path of the ES cell-derived motor neurons.

His team has already started to explore whether human ES cells will behave in the same way. "We are collaborating with a group of scientists in Israel [led by Nissim Benvenisty of the Hebrew University in Jerusalem] who have experience in turning human ES cells into neurons," Jessell says. The goal, he says, is "to find out whether you can efficiently convert human ES cells into human motor neurons."

If so, "one could really start to evaluate whether this would be a sensible way of approaching ALS, spinal muscular atrophy or spinal cord injuries," Jessell says. But he warns that there are big hurdles ahead. "The motor system relies heavily on the precision of connections and circuits. Simply having generated a motor neuron is not sufficient. I think the challenge will be to reconstruct appropriate circuits," he declares. Partly for this reason, Jessell is now studying the next steps in the differentiation of motor neurons.

To help develop new therapies, Jessell has also started to collaborate with neurologists who are doing research on ALS as well as other spinal cord disorders and injuries. "There are many groups interested in cell-based treatments of diseases like ALS," he says, "but one of the difficulties in comparing results is that most people have been using different cells in different systems. If we can generate motor neurons under fairly standardized conditions, we can provide them to anyone who is interested, and it will be easier to compare their results."

Jessell is working particularly closely with Robert Brown, director of the Neuromuscular Disorders Unit at Massachusetts General Hospital. "Bob Brown is a world expert on ALS," Jessell says, "and he has mouse models of ALS. It will be interesting to test whether introducing motor neurons derived from mouse ES cells in these models actually does any good."

The two research teams are also investigating the basic cause of ALS. A gene whose mutation is known to result in ALS in humans is the gene for an enzyme called superoxide dismutase 1, or SOD1. Brown uses a mouse model of ALS engineered with the same mutation, with resultant motor neuron degeneration. Yet "nobody knows why motor neurons are selectively vulnerable to death in ALS," Jessell points out. He hopes to discover the answer by comparing motor neurons that are derived from normal ES cells to motor neurons derived from ES cells bearing the SOD1 mutation. "Then we can ask what has changed, biochemically, that might be a predictor of the later degeneration of the motor neurons," he says.

Scientists who work with stem cells are finding it difficult to choose among the many different research paths now opening up—and finding it particularly challenging to work with human ES cells. Although all mouse ES cell lines "behave in a rather uniform way, that may not be true in human ES cells," Jessell says. Yet researchers may have a hard time finding out. Most of them rely on federal funds, which can be used to study only a limited number of human ES cell lines. And as Jessell points out, "human ES cell lines are so poorly characterized, compared to many of their mouse counterparts, that nobody actually knows how much they vary and whether the full range of developmental potentials is going to be offered in those lines that now have federal approval."

WHAT ABOUT ADULT STEM CELLS?
The researchers who have chosen to work with adult stem cells face other problems. In many tissues, adult stem cells are relatively unexplored and still full of surprises. They become most active when tissues wear out or are damaged, yet they are often hard to find. Besides the nervous system, researchers are looking at blood, skin, nails, hair, saliva and sperm cells—where the need for replacement is particularly obvious.

Allan C. Spradling, an HHMI investigator at the Carnegie Institution of Washington in Baltimore, Maryland, has focused on what he calls niches, special microenvironments in various organs such as the skin, gut and gonads. Each niche houses one or more adult stem cells and regulates their activity. His team has identified three types of regulatory cells that form such niches and keep the stem cells from differentiating prematurely.

The strongest evidence for such regulation in mammalian tissue probably comes from studies of spermatogenesis, Spradling says. Thousands of stem cell niches have been found lining the walls of the seminiferous tubules in which sperm develop. Spradling has focused on the niches that surround germline stem cells in fruit flies and on the signals they send to the stem cells. Such signals appear to have a powerful influence on the behavior of adjacent stem cells, he says. By comparison, the stem cells "may themselves be relatively unspecialized." Therefore, the environment of stem cells will have to be controlled with care if one hopes to use them in some form of therapy.

More data on the complexity of using adult stem cells comes from Sean J. Morrison, an HHMI investigator at the University of Michigan, who found that the properties of some adult stem cells were quite different from those of embryonic or fetal stem cells. Morrison managed to isolate "neural crest" stem cells from the gut tissue of adult rats, even though such stem cells (which give rise to various tissues, including the peripheral nervous system) were supposed to exist only in the embryo and fetus. Then he either cultured them or transplanted them into chick embryos to study their activity. It turned out that these adult stem cells could do some of the same things as embryonic and fetal stem cells but not others; for example, they could not differentiate into cells that make two important neurotransmitters, serotonin and noradrenaline.

Morrison also found that neural crest stem cells he had isolated from the gut of rat fetuses differed from those that came from the sciatic nerve. After transplanting cells from both sources into the developing nerves of chick embryos, Morrison discovered that stem cells from the gut produced mainly neurons, whereas those from the sciatic nerve made only glial cells, a type of supporting cell. This finding, he says, "suggests that it's really important to match the origin of the stem cell to the therapeutic job that you're trying to do."

"There is a great debate in the field of regenerative medicine as to whether one should start with ES cells or with adult stem cells," Jessell says. "Many groups are trying to get adult neural progenitors to differentiate into particular cell types. Our study shows very clearly that a mouse ES cell can become a motor neuron in a very predictable way, but so far, no one has shown that an adult neural progenitor cell can become a motor neuron."

"I think the potential of adult progenitor cells is exciting," Jessell adds, "but in my view, the evidence that they perform as well as their embryonic counterparts is not strong at the moment."

Download this story in Acrobat PDF format.
(requires Acrobat Reader)

Photo: Marc Bryan-Brown

Reprinted from the HHMI Bulletin,
December 2002, pages 22-27.
©2002 Howard Hughes Medical Institute