Tooth Development

Is it possible to build a tooth? That’s a question that many giants of 20th century dental research no doubt considered, and it’s a conceptual puzzle that continues to capture the imaginations of the nation’s oral health scientists. But there is a key difference between the musings of then and now. Today’s scientists possess for the first time the needed laboratory tools to plumb the molecular depths and developmental biology of tooth formation, and some already have begun to do so in earnest.

The research follows two broad but complementary tracks. One seeks to define in sequential detail the genetic programs underlying tooth development and, moving to the next biological level of activity, to map the protein biocircuitry in tooth-forming cells that carry out the genetic program. This research delves into the genetics of the initial tooth placode, a thickened patch of ectoderm near the fetal head that arises as migratory neural crest cells arrive early in development. It then tracks the sequential development of the tooth cap, the tooth bud, and ultimately the maturation of the individual dental tissues therein, from enamel down to the cementum of the tooth root.

The other path aims to take this fundamental information and, like a minimalist artist, deconstruct the complexity of tooth development and define its essential molecular requirements. By stripping away the redundancies and other non-essential molecular chaff of the process, scientists hope to match or possibly improve upon nature’s instructions to engineer replacement tissues for damaged teeth.

Offering their perspectives on the road ahead are two NIDCR grantees. They are: Dr. Richard Maas, a scientist at Brigham and Women’s Hospital and Harvard Medical School in Boston, and Dr. Malcolm Snead, a scientist at the University of Southern California in Los Angeles.

Richard Maas

As a part of the NIH Roadmap Initiative, you have begun a project that, in part, explores tooth development. Could you tell us about it?

Sure. The project focuses on three structures: the tooth, pancreatic islet cells, and the heart valve. The central premise here is we now know enough about organ development to use this knowledge base as a template to assemble a far more comprehensive biological picture of the process.

You want to put more meat on the bone?

Exactly. The program’s acronym is SYSCODE, which stands for Systems-Based Consortium for Organ Design and Engineering.

So, it’s a systems biology approach, or studying the cell as an integrated system of biological circuits and data processing?

Well, the systems-based aspect arises because we want to integrate the various data sets that the new generation of research tools now can generate. These include: comprehensive gene expression profiles; extensive catalogues of protein expression, chip-on-chip analyses to figure out where transcription factors bind in the genome; assembling data on common inherited genetic alterations; and logging the results of RNAi experiments to inhibit an individual gene’s expression and thus study its function and thereby dissect relevant signaling pathways within the system.

What will this data integration produce?

Our hope is a coherent molecular blueprint to build a tooth. The idea is, once the computer and genome scientists have assembled this large body of information in an intelligent format, it will be amenable for tissue engineers to use.

How would they use it?

The set of instructions would be in a user-friendly format for a scientist to say, “Okay, I need to add Factors A, B, and C in this particular sequence, at these concentrations, and in this particular combination. I need to couple them with a scaffold of extracellular matrix materials X, Y, and Z. And because this information is based on how the tooth normally forms, if we reconstruct those types of parameters as best we’re able in vitro, we should have a molecular blueprint that will yield a structure that approximates a tooth.”

How does a tissue engineer do it today?

Well, the current paradigm for tissue engineering consists of taking Factor X, adding it to some cells, and seeing what happens. I’m in no way denigrating that approach. I think it’s been incredibly successful. But it is empiric, and a molecular blueprint would be extremely helpful.

In generating the data sets, on one level, wouldn’t you want to tease out the evolutionary biology of the tooth? By analogy, a mechanic needs to know the make and model of an automobile before lifting up open the hood.

You certainly want to know how a tooth or any part of the body came to be. So, yes, I agree. And that information is being assembled. From DNA sequencing projects to improved mouse models, a major investment has been made over the last decade or so to perform comparative analyses among species and tease out these evolutionary motifs.

But, on another level, how detailed will the instruction manual need to be? In other words, do we need to recapitulate all of the moving parts and redundancies that are built into the system? Or can the process be streamlined in the laboratory and remain functional?

The answer is we probably don’t need to know all of the moving parts. There are some basic organizing principles at work in the tooth bud that can be mastered and hopefully exploited to form enamel, dentin, cementum, and the other constituent parts of the tooth. That means we don’t need to account for every last gene in the human genome to build a tooth, and that’s why this is a doable task.

How many genes would make the cut?

Well, although there are roughly 23,500 genes in the human genome, probably only one tenth of those are expressed during tooth development. That is to say, they satisfy the condition of being necessary or sufficient.

And of those genes, some likely will be more critical than others?

That’s right. Not all genes and proteins that plug into a developmental pathway are of equal importance. This relates to the structure of networks.

How so?

There are two general types of biological networks hardwired into our cells. One is called a universal random network, where every component, or node, is equal in importance and weight. The second and more prevalent form is called a scale-free network. In scale-free networks, not all nodes are created the same. There are very important centralized nodes, or hubs, that act as convergence points and processing centers for incoming biological information. Think of the spoke-and-hub system in aviation. If you identify the hubs – the Chicago O’Hare’s -- it’s possible to predict the behavior of the system to a large extent and without needing to piece together every single element of the network.

What might these predictions reveal?

Let me give you an example. We study a gene that, if inactivated, results in the formation of supernumary, or extra, teeth in the mouse. That suggests that this gene and its protein product are very high up in the regulatory cascade that controls tooth development. By manipulating that protein, you wouldn’t necessarily have to control all of the others that are activated subsequently, or downstream, of it. Because that master gene would take care of them for you. You see? So, this shows the great simplification that is possible as some genes turn on entire programs of downstream events.

Are there other basic organizing principles?

There’s a corollary principle. I would call it the principle of autonomy. By that I mean, if an early tooth germ reaches a certain developmental stage, it will continue to develop all the way to the latter stages of mineralization. There’s actually a precedent for that. Dr. Paul Sharp and colleagues at Guy’s Hospital in the UK showed some time ago that if they took what’s called a cap-stage tooth germ and grafted it into the jaw of an adult mouse, it will in fact develop much, much further.

That would mean we don’t necessarily have to worry about mastering all stages of tooth development. Once a developmental tipping point is reached, biology could take care of the rest?

That’s right.

What impact might this and related work have on practitioners in the coming years?

Well, let me just say that current prosthetics work relatively well. But millions of Americans still lose a significant number of teeth during their lives, and dental disease remains a significant health problem. So, if we could generate a biomimetic substitute, it would be welcome. To do that at a reasonable cost and with good efficiency, I think you’re looking at least a decade into the future. On the other hand, if you could generate enamel matrix in a test tube from cells that you’ve programmed, that would be very exciting. How that would figure into clinical practice, dentists no doubt would decide. But, clearly it would be a wonderful natural product.

Malcolm Snead

The term “building a tooth” suggests creating a bicuspid or incisor from scratch. But that’s not the focus in your laboratory?

Most of my interest and expertise developmentally lie downstream of those initiating events. In other words, I’m not interested in day six or seven of gestation, although I think early development is very interesting. My research focus is on the problems of tissue specification during late gestation and early postnatal development.

How do these problems flow into building a tooth?

In our case, the focus is on learning to engineer new tissue to replace damaged or diseased tooth structures.

Why engineer?

Let me back up a bit. Biology is now in a golden age of discovery. We can knock out a specific gene, modify another gene, and ask a variety of profound questions about the circuitry of the cell that just weren’t on the table a decade or so ago.

For example?

Well, you could ask what happens systemically if you remove 100 percent of transforming growth factor X? Does the cell – the biological system - have the ability to compensate for the loss via a redundant signal? If there’s no compensation and thus the effect is uniform within the system, what then happens downstream when the circuit is shut down? It’s kind of like a caveman holding a pocket watch. You smash the pocket watch and say, “Great, look at all of these parts in it.” But can you reassemble the watch? The next step is to go back and say, “Okay, I know that I need this piece, but how far can I turn it down,” so that it functions at 10 percent of its normal level and still get an outcome?

Addition by subtraction?

Right. If you’ve identified all of the pieces, can you also define the ones that you don’t need? It’s a matter of relevance, and that is now framed within the context of our expanded scale of discovery. What I mean is we used to try to understand how the cell worked at some level when we could perform a Western blot assay and detect a protein. We advanced to a Northern blot assay to process RNA and that gave us greater sensitivity to quantify gene expression. In the 1990s, we entered a PCR state of affairs that enabled us to look at five or six molecules in a cell. Now you must start pulling out the noise, the chaff in the system. What is relevant? And what is spurious noise? It’s oftentimes a matter of understanding what you know relative to how you think you need to know it.

That brings us back to engineering a tooth. What about something as complex as a tooth root? It’s a four-in-one proposition with dentin, pulp, cementum, and the periodontal ligament.

It represents a challenge, but certainly a worthy one. If you can understand root formation, you have a much better handle on regenerating a major cause of tooth loss in adults, which is the loss of supporting bone and ligament. But what’s important here is this is a challenge that we now can productively wrap our minds around.

What about regenerating enamel, one of your major research interests? Enamel is a tissue in which the whole is greater structurally than its individual parts, in this case, elongated hydroxyapatite crystals.

Enamel is a fascinating tissue. I’m actually sitting here holding a chunk of hydroxyapatite in my hand. If you made a tooth out of what’s in my hand, it would fracture and fall a part in a matter of days. Its toughness, hardness, and elasticity really are quite different than the hydroxyapatite in the enamel of my teeth. Some of that has to do with the nanoscale that nature works to weave hydroxyapatite crystals into the patterned structure that we know as tooth enamel. Another small part of that is some residual amount of protein. It’s maybe 5/10 of a percent of protein dry weight. It’s likely retained for a very specific function.

And it’s retained as a remnant of the original protein matrix?

Right. Even though the tooth erupts into the oral cavity as a white, mineralized fossil, it was not a fossil during its formation. Before the hydroxyapatite crystals elongated, properly oriented themselves, and formed mature enamel, they were seeded in an extensive protein matrix that served as a developmental lattice. All of the rules that apply to changes in gene expression, control of protein expression, response of different signaling molecules through membrane-mediated receptor events and secondary signals. All of those are happening. Enamel is very, very much alive as it’s being made.

But can you go all the way back and track the assembly of, say, the amelogenins in forming the protein matrix?

Absolutely. In fact, there are 16 different isoforms, or types, of amelogenin. We work on them in in vitro analyses, and that tells you certain things. But then you must go back to the organism and say, “In the context of the organism, does it work this way?” That’s where the approach to genetic engineering - the simplicity of a biological outcome as your measure - is very appealing. If you can achieve the same outcome with a lot less moving parts, it suggests that you understand how the system works and know its critical parts.

How’s the engineering going?

Quite well. Right now, I’m working on a manuscript that produces almost a two order reduction in simplicity in proteins of the amelogenin class. We used a genetic knock-in strategy that ends up producing an enamel that has essentially characteristics that are within 20 percent of the natural enamel. So I’ve made this enamel with 16 times less alternative proteins participating. And yet, the system seems to work adequately. Or, within 95 percent of the expected values. So, this question of how far can we go is one that I share with the rest of the team here at USC and numerous labs around the world.

And across scientific disciplines.

Yeah, I think a lot of people in materials sciences, engineering, and nanotechnology get quite excited about it. Making a mineral that behaves in the way that enamel does, or the way that the enamel actually bonds to the underlying dentin and dentinoenamel junction, or DEJ, is a phenomenal piece of engineering. When I speak to my engineering colleagues, they would like to know about the DEJ. How does the enamel stay on the dentin?

That’s of real significance to dentists.

Right, if you could replicate a DEJ , it would be a first step toward ensuring that fillings lasted longer. Restorations usually fail at the interface. It’s not the filling itself. It’s the bond of the materials. If you could make the interface more contiguous, so that the junction between the underlying dentinal surface and the filling material. If a glass-like ceramic material is more compatible with the dentin, you can increase the longevity of the restoration.

And the data are pouring in.

Yes, that’s right. It’s pouring in all areas of biology. I was at a nanotechnology meeting recently where people were talking about monitoring 40 channels of data coming out of the cell simultaneously, from their cell receptors to their oxidative state. You could see lots of different parameters of cell biology being reported in a single cell, instead of looking at a thousand cells and averaging their behavior for one parameter. You can just imagine what that means. Suddenly you must ask, “Have we been measuring the tops of mountains? There’s a much wider range to how cells respond. We have to see a 20 fold change in its activity before we even start to say we can measure and work with it. When, in fact, the system is much more sensitive, it may be changes of 20 percent that make differences. But we ignore them because we don’t see them.

Next: Periodontal Disease: Engineering the Future of Care