Sheng Ding, Scripps Research Institute

Embryonic stem cells are hard to grow, study and use. In a review1 in Nature, Sheng Ding of the Scripps Research Institute, in La Jolla, California, describes identifying small molecules that not only help improve techniques used to manipulate stem cells in culture but also answer questions about the cells' basic function.

(Note: This is part of a series of interviews conducted to accompany Nature Insight Regenerative Medicine .)

What can small molecules teach you about stem cells?

For a lot of stem cell phenotypes we don't understand their mechanism, so we don't really know how to generate a phenotype or how to control that process. The small molecule has to interact with a partner to have a biological effect. So we look for the protein the molecule binds with. In that regard, there's a lot of discovery after you've identified the molecules.

Pluripotin is sort of a magical molecule with two functional targets [RasGTPase activating protein (RasGAP)and extracellular signal–related kinase 1 (ERK1)], and only by regulating those two targets will that generate such a specific phenotype [the ability to grow well in culture without serum or feeder cells]2. That's a fundamental understanding in embryonic stem cell self-renewal.

Does pluripotin make mouse embryonic stem cell culture easier too?

Under conventional feeder-cell conditions, some of the cells differentiate, so it's not a homogenous population of cells. With that small molecule your culture is more homogenous; the cells grow better in terms of morphology, markers, function. Sheng Ding

Yes, absolutely. It also makes the culture more robust. Under conventional feeder-cell conditions, some of the cells differentiate, so it's not a homogenous population of cells. With that small molecule your culture is more homogenous; the cells grow better in terms of morphology, markers, function.

What's the point of replacing biological components with small molecules?

Using chemically defined conditions is really about removing the variables. The conventional way of culturing embryonic stem cells or other types of pluripotent cells is basically growing those cells on feeder cells and in the presence of serum. The problem is that those conventional conditions can be very variable. Feeder cells can secrete different factors at different concentrations depending on their quality. Also the serum from batch to batch is very different; it depends on how people process those serum products. So people find inconsistent results from one experiment to another.

The advantage of using chemically defined conditions is you know every single component in your culture. When you observe a phenotype, you know the exact condition that controls the environment. With serum or feeder cells, it could be an unknown factor.

Small molecules that can actually help stem cell culture are not hard to make. In the near future, they will become standard media components.

What else can small molecules do?

ROCK [Rho-associated kinase]inhibitor is good for human single-cell survival. The MEK [mitogen-activated protein kinase] inhibitor will be very useful. GSK [glycogen synthase kinase] inhibitor would also be useful for self-renewal or differentiation.

Reversin is very interesting conceptually. These days we all talk about induced pluripotent cells. Reversin also induces reprogramming, not to the pluripotent stage but to a multipotent state. That's still a new frontier in reprogramming.

It seems like differentiation would be more difficult than promoting self-renewal because there has to be a sequence of steps.

You just need to do screening in a stepwise way. Certainly you use those small molecules for a specific time window.

We and others have used small molecules for differentiation. In some cases it's trial and error, in other cases we have substantial knowledge from embryology and can use that to try to recapitulate those steps: we model in vitro differentiation on embryonic development. That has happened for the neural, cardiac and pancreatic cell types.

How does this work apply to the clinic?

Right now, most people are talking about stem cells as regenerative medicine; many people are talking about cell-replacement therapies. But that's just one option. In our opinion, it won't ultimately be mainstream medicine. More conveniently, we can develop small-molecule therapeutics that people can take in conventional ways.

Neuropathiazol is a molecule we identified that works on adult neural progenitor cells; it can redirect cells to become neurons. Typically, adult neural progenitor cells are glial restricted. In vitro those cells can become neurons, but the dominant type is astroglial. You can imagine a disease where you want your endogenous neural cells to become neurons, [and] so that's the motivation to identify small molecules that can make the neural progenitor cells become neurons more specifically. It's sort of a proof of principle that you can find molecules ex vivo to control the stem cell fate.

What's the range of chemical classes that still needs to be probed?

What is really enough? That is a very hard question. Even in the pharmaceutical industry, people are still looking for new ways [to synthesize molecules]. Chemical diversity is restricted by synthetic accessibility, so if you can't make the molecule, you won't be able to access chemical space; so people need new synthetic methods to generate new chemical compounds, but no one really has a chemical collection that will saturate every possibility.

On the other hand, in practice, you do not want to have your chemical collection just based on diversity. There's no reason to generate every chemical compound in every corner of chemical space. Chemical space is very huge, and generally chemical compounds do not exist for biological interaction.

So for what's being screened now for small molecules, is that enough?

It's a good starting point.

What about screening technologies?

The screening technology has been lensing pretty fast. Now a lot of people are talking about high-content screening, looking at several parameters of a live cell; it's more functional than using artificial constructs like reporter genes. With automated or imaging technology, you'll be able to get a lot of information up front and also cheaper.

Can small molecules do anything that adding genes or proteins can do?

I would think it can but probably in a different way. I wouldn't say that one small-molecule would function identically like a protein but small molecules can work in a different way to replace or compensate that specific protein function.

What accomplishments and challenges do you expect to be describing in ten or fifteen years?

Fifteen years from now we should have examples of small molecules or drugs that have been developed for in vivo cell regeneration.

Challenges would be hard to predict; it's easier to predict what can be done. Ten or fifteen years down the road, we may still have to deal with those molecules' specificity and how to precisely deliver the molecules only to the right cells. The challenge will always be how to control the molecules' action on the specific cell types.

Sheng Ding is a founding scientist for the start-up company Fate Therapeutics.