Four Genes Are Enough – How to Make a Cell Form a Tooth

Your research group has been focusing on stem cells and the development and regeneration of tissues using the model of continuously growing mouse teeth. Can you put this current study into the context of your previous work?
We had previously managed to describe in detail how odontoblasts develop — from stem cells through several intermediate stages to the cells that actually form the tooth. At the level of gene expression, we identified which genes turn on and off at various stages of this process. In the current work, we focused on so-called regulatory or controlling genes.

Could you explain what these genes are and why you focused on them?
They are often referred to as transcription factors or regulators. These genes are responsible for controlling how other genes behave. They can influence dozens or even hundreds of other genes — hence the name “controlling genes.” We selected four of them that had never before been associated with tooth development and examined what happens when we insert and specifically activate them in pluripotent stem cells, i.e., cells capable of giving rise to any cell type or tissue in the body.

What exactly were you trying to find out, and why is it significant?
We wanted to see whether activating these genes would cause stem cells to differentiate into cells that form teeth. It turned out that they do. This study demonstrates the importance and power of fundamental research. We have shown that understanding a specific developmental process in a living organism can pave the way for targeted applications in tissue engineering or for obtaining specific cell types.

If humans have between twenty and twenty-five thousand genes, how many of these are regulatory genes? And how did you select your four, given that they hadn’t previously been linked to tooth development?
There are roughly around a thousand regulatory genes. We selected our four based on single-cell gene expression analysis. Thanks to our previous analysis of continuously growing mouse teeth, we can now determine the precise gene expression in each individual cell. We can identify which cells are at the beginning, middle, and end of the developmental process — and which genes are expressed at each stage.

On what basis do you divide this development into phases? Are there any milestones?
That’s a good question, because the process is completely fluid. However, based on processes such as differing gene expression, we can divide it into various intermediate stages. Of course, even between two such defined stages there exist other transitional states that cannot be easily classified. It also depends on how we set our parameters — and ironically, the more data we have, the more we see how transient the process actually is. If we tracked only one or two genes, it would be easier to define those stages.

So it’s like when anthropologists claim to have found a missing evolutionary link, but we can never be sure how many intermediate forms there really were…
…or we could compare it to a rainbow. At first glance, you see red and yellow, but when you look closer, you also see orange — and between yellow and orange, countless shades.

You talk about switching genes on and off — their activation and deactivation. How do you actually do that? It might be obvious to other scientists, but not to everyone…
It’s not that obvious — there are actually several approaches. We worked with lentiviruses, which belong to a group of retroviruses that can include, for example, HIV. For us, the crucial property of these viruses is that they can permanently integrate their DNA into the genome of their host. They have the capacity to carry foreign genetic information — a so-called cargo space — into which we insert the sequence of the chosen gene. We then infect stem cells with the resulting virion, and in this way, we introduce the desired genetic information into their genome. After that, we simply select the population of modified cells — for instance, using antibiotics or cell sorting — and obtain a relatively pure population carrying the genes we’re interested in.

Where do you obtain these lentiviruses?
In the past, each lab had to clone them “in-house,” which was a very time-consuming process. Fortunately, there are now companies that can produce a virus to your exact specifications. We chose the easier, commercial route. In practice, this means we design the viral sequence on a computer — and then simply wait for the shipment, which arrives as a test tube we can immediately start working with.

Since you focused on four regulatory genes, did you work with multiple variants?
Yes — the cargo space of the virus is fairly limited. We infected cells with three different virions, creating several combinations. The four genes were divided into pairs — two “early” genes and two “late” genes — and a third construct served to activate them. When we introduce our foreign DNA — these four genes — into the cells, they are initially inactive by default. Only treatment with doxycycline, an antibiotic, turns them on and allows them to function.
In the first combination, we activated the two early genes; in the second, the two late ones; and in the third, all four together. Over the course of several weeks to a few months, we analyzed how each setup affected the cells.

If we use IT terminology, you were basically writing a kind of script for the stem cells by modifying gene activity…
Exactly — though a very rough script. It might not seem that way at first glance, but these are major interventions for the cells. The roughly 20,000 genes we have — which code for all our proteins — are not actually such a large number. In addition to them, there are many non-coding genes and regulatory sequences that determine gene expression, and these sequences themselves are influenced by even higher-level regulators… In total, we’re talking about billions of variables. So we’re still a long way from being able to perform any kind of fine-tuning. (laughs)

In any case, you succeeded in getting the stem cells to start producing molecules typical of odontoblasts, and later even collagen and mineralized tissue. What is the connection between these processes?
An odontoblast is a cell that forms dentin, the main hard substance of the tooth, composed largely of collagen, which binds hydroxyapatite molecules — this mineralization process makes dentin hard. So, when a cell starts producing collagen, that’s the first sign that it may become an odontoblast. In the second phase, we found that our cells were capable of producing mineralizing deposits, accumulating calcium within them. A similar, though not identical, process occurs during the formation of other hard tissues.

How crucial is it that you were able to observe this process also in a living mouse?
At first, our goal was to create an organoid (or spheroid) — sometimes called a mini-organ. Although these systems are widely used today, they have one major drawback: they lack vascularization, innervation, and immune-system interactions, as well as the physical influences of the surrounding environment. Therefore, they do not fully replicate what happens in a living body. To better simulate natural conditions, we implanted our organoids under the kidney capsule of immunodeficient mice — mice that can accept foreign tissue as their own. The mouse vascularized the implant, allowing it to develop further. In this way, we came one step closer to what happens inside a living organism.

Were you surprised that the processes initiated in vitro continued in the living mouse?
It was a discovery worth celebrating! (smiles) This was a very high-risk project — in the “high-risk / high-gain” category. We hoped our hypothesis would hold true, but there were thousands of things that could have gone wrong. The fact that it worked wasn’t exactly unexpected, but it was certainly a very pleasant surprise.

How disappointing is it when things don’t work out that way?
Disappointments are inevitable — they happen at many stages of any project. The key, in my opinion, is to be prepared, experimentally robust, and not get desperate when an experiment fails. When it comes to high-risk / high-gain projects, from my experience perhaps eight out of ten fail or lead nowhere. But the remaining two reveal something new, something fascinating — something that perhaps no one has seen before — and that’s worth pursuing further. Doing experiments when you don’t know the outcome in advance is what makes science exciting.

Was there anything else that surprised you during this project?
Definitely. Some of the newly formed cells exhibited features of other cell types. In addition to expressing genes typical of odontoblasts and producing collagen, they also expressed genes characteristic of epithelial cells, such as those forming the skin.

What does that mean?
It’s interesting for several reasons. Cells expressing such a combination of genes don’t exist naturally in our bodies. We essentially created a kind of cellular hybrid. In mice, for example, if there’s even a small shift in gene expression in enamel-forming cells — something that can happen even in otherwise healthy mice — the animal can grow hair instead of teeth. This shows that even a minor change in gene expression can lead to a major change in the resulting phenotype — tooth versus hair. Isn’t that fascinating?

Two years ago, we discussed your BEE-ST method, which enables you to observe the rate of hard-tissue growth in real time. You mentioned collaboration with other institutions and applying the method to other model organisms, not just mice. Does this new project have a similar broader impact?
Yes, definitely. We demonstrated, using tooth development as a model, a new way to derive terminally differentiated, or fully functional, cells from stem cells. In our case, these are odontoblasts. But I believe that with the right combination of transcription factors, it might be possible to obtain other cell types in a similar way — for instance, neurons or pancreatic beta cells.

Scientists have reportedly already managed to modify pancreatic beta cells to produce insulin. What makes your approach fundamentally different?
That’s true. But in those cases, gene expression is typically influenced externally — through the action of a protein or a small molecule. Our approach is different in that we went directly inside the genome. We intervened in pluripotent stem cells, which have the potential to develop into any cell type, by introducing selected transcription factors — the regulatory genes. We demonstrated that by choosing the right transcription factors, we can directly control differentiation, i.e., the transformation of these cells into a desired type.

When working on such projects, how do you deal with the risk that the cells you manipulate might initially become odontoblasts but later prove unstable?
That certainly happens. It’s important to emphasize that our goal wasn’t to invent a new therapeutic procedure, but to understand how things work and to demonstrate a potential direction for future research. Not all cells modified by our method became odontoblasts — we always ended up with a mixed population. It wasn’t as if every single cell transformed at once. The key outcome of our work was showing that the genes Sall1, Etv4, Nupr1, and Gsc are essential for odontoblast differentiation.

Do you have an explanation for why some of the cells didn’t differentiate?
Biology, unfortunately, isn’t like mathematics, where every step is precisely determined — at least according to my modest knowledge of math! (laughs) In cell and molecular biology, natural mechanisms like gene silencing occur. This process can cause some genes — including those we’ve inserted — to become silenced by the cell’s own internal mechanisms. It’s part of the organism’s internal control system that ensures it functions properly. It also depends on where in the genome the inserted sequence lands, which can influence other factors as well. There are many variables, and the outcome can vary widely.

You mentioned this project two years ago, but at the time you couldn’t say much more. Can you hint at what we might be discussing in another two years?
(smiles) This project was, so to speak, one-off. We don’t expect a direct follow-up in our own lab. However, we’ve discussed our results with, for example, a colleague from Canada who’s interested in how our four selected genes influence gene expression through other regulatory sequences — and he’d like to collaborate with us on that topic. Altogether, this project took about six years, during which we launched dozens of other projects, so I hope we’ll meet again in the future — hopefully over something just as interesting.

---

More articles

All articles