Evolutionary Biology: Cichlids, Gene Networks, and Teeth

February 2009

In the journal PLOS Biology, NIDCR grantees report they have deduced a network of dental genes in fishes called cichlids that likely were involved in building the first tooth half a billion years ago.  The researchers say their finding introduces into the scientific literature a core evolutionary list of molecular pieces needed to make a tooth.  These original parts were then gradually rewired, replaced, or left in place to produce the various shapes and sizes of teeth now found in nature, from shark to mouse to monkey to human.  The Inside Scoop spoke with Todd Streelman, Ph.D., a scientist at Georgia Tech University in Atlanta and a senior author on the study, to learn more about his group’s discovery.

Cichlids belong to a large family of fish called Cichlidae.  No one knows how many species of cichlids exist in nature.  Our best estimate is somewhere between 1,300 and 3,000 species.  Cichlids can be as small as 2.5 centimeters in length and as long as nearly a meter.  Several species are actually quite well known.  Tilapia, for example, is a cichlid species that people like to eat.  And people with aquaria at home certainly are familiar with the angelfish, discus, and oscar.  They’re cichlids, too.

Well, Malawi cichlids offer an unprecedented opportunity to study the evolution of a number of traits from color patterns and brain structure to the shaping of teeth, jaws, and other craniofacial structures.  The lake is largely a closed ecosystem, and the various species have rapidly adapted to fill their ecological niches.  It’s estimated that about 1,000 different cichlid species live in Lake Malawi.  So, from a scientist’s perspective, it’s almost like being a kid in a candy shop, if you like candy.  The biological diversity is amazing.

Let me give you a little background.  The ancestors of most East African cichlids probably originated in Lake Tanganyika, the eldest of the lakes in the region’s so-called rift valley.  Lake Tanganyika is between eight and 10 million years old, and it served as an evolutionary reservoir for Lake Malawi and the other nearby lakes and rivers.  So, our best estimate is cichlids have inhabited Lake Malawi from between one to two million years.  And much of the present-day diversity is thought to have evolved in the last ten- 100 thousand years.

And they are early vertebrates?

That’s right.  You raise an important point.  A lot of really great research has been conducted in mice over the years to tease out the genetics of making a tooth.  But mice are mammals that sit higher up on the evolutionary ladder than cichlids.  If you study mammals only, you’ll miss things that happened early in the evolutionary process.  It’s like walking into a theater halfway through the movie.  Cichlids, zebrafish, and other lower vertebrates allow you to see the opening scenes. 

Last year, we described a genetic network that seems to control tooth size, number, and spacing in the oral jaws of three closely related Malawi cichlids.  The network involved 10 genes – bmp2, bmp4, eda, edar, fgf8, pax9, pitx2, runx2, shh, and wnt7b.

The genes already have been reported in the scientific literature as playing a role primarily in mammalian tooth development and in other structures, such as hairs and feathers.  So, we didn’t discover them.  I also should add that we use the term “network” to mean genes that coordinate their expression during a window of developmental time.  We don’t use the word in the systems biology sense of an interconnected network.  Clearly, there’s still a great deal of biology to work out.  What’s important about our discovery is the timing of differences in gene expression.  We show that variation in the network kicks into gear very early in the developmental process, when the initial tooth pattern is laid down.

Teeth are evolutionarily old structures, correct?

Exactly.  Teeth are ancient.  While we often think of teeth as being inextricably linked with jaws, they evolved first in the pharynx of jawless fish about half a billion years ago.  As strange as this might sound, teeth predate jaws. Like hair and feathers, it’s possible to study fish teeth as patterned, iterative structures that are constantly replaced throughout life.  That’s certainly not the case in mammals.  But it holds true in cichlids.  Some cichlids have a total of about 3,000 teeth.  Every single tooth gets replaced every 50 to 100 days.  This is accomplished via a stem cell niche associated with each functional tooth. The ability to replace teeth throughout life has been lost in mammals.

We don’t know the answer.  But you can still see the evolutionary transition in nature.  Some lower vertebrates, like the zebrafish, have teeth only in the pharynx.  Mammals, such as mouse and human, have teeth only in the oral jaw. Cichlids have teeth on both the pharyngeal and oral jaws.  This unique evolutionary feature allows us to ask a question that is the starting point for our current PLoS Biology paper.  Is tooth number regulated similarly across the pharyngeal and oral jaws?

It was biologically intriguing.  The two jaws not only are functionally distinct and evolutionarily decoupled, but the teeth on these jaws have different developmental precursors.  A tooth forms from mutual signaling interactions between a cell layer called the epithelium and one called the mesenchyme. Pharyngeal teeth likely use endoderm as their epithelial layer and oral teeth use ectoderm. If tooth number was regulated, or controlled similarly among these jaws, it might suggest that teeth are made in the same way, regardless of how and where they develop.

What did you find?  

To our surprise, we found that tooth number was regulated similarly in the two jaws.  Oral and pharyngeal jaws shared the same constraints on tooth number. 

What we found is a set of common genes that we argue form a dental gene network.  This network is common to most dentitions.  Included in the network are the genes that we described in our previous paper.  That also included eda and edar, which was a surprise.  These two genes are thought to be involved exclusively in making ectodermal tissues.  But we found these genes expressed in the pharyngeal dentitions, which we think are derived from endoderm.  So, that opens up the role of eda and edar in tissues derived from the endoderm.  It also points to the idea that before jaws, hairs, scales, feathers, and other ectodermal tissues ever arose, these genes were acting in a dental network deep in the pharynx. 

Richard Feynman, the famous physicist, once said, “I can’t understand it unless I can make it.”  I’d flip that around to answer your question.  You can’t make something unless you fully understand it.  For instance, we described two things in this paper.  One, an ancient gene network that our data indicate is active in these oldest populations of teeth.  Two, and perhaps more importantly, we describe what we call the core dental network – the set of genes conserved in all teeth that we currently know about, from fish to mouse to human.  So, what’s potentially interesting there not only are the things that fall into the network (surprises like eda and edar), but the things that fall out.  For instance, take the genes pax9 and fgf8, which are necessary ingredients for mammalian dentition.  These genes are either not expressed at all or only expressed in oral teeth, not pharyngeal teeth.  That suggests they are not evolutionarily necessary to make a tooth. 

That’s right, and when they became important in dental evolution.  If you wanted to make teeth from culture or in a test tube, you could ask which types of molecules would be necessary.  Even though some of these genes appear to be genetically necessary for mammalian teeth, there may be other ways written into evolutionary biology to make teeth. 

Right, let’s go back to fgf8.  It’s expressed in and necessary for mammalian molars.  But it’s not expressed in any fish tooth that anyone has ever examined, although there are other FGF family members, and perhaps other molecules that could serve the role of an FGF if one were trying to make a tooth.  Our analysis can start to point to some of those things.

I think this is one of the open questions for the next decade of research.  What are these models really for?  What I see happening – for teeth and really any human organ or disease – is the genomic resources now are so good for humans, we might not need model organisms for particular diseases in the future. It’s actually just as easy to go to human populations and directly ask which genes are responsible for this and that trait.  But if you follow how genomic information is now being used to find human disease genes, we still need models that allow you to assemble the map between genotype and phenotype.  Many of our current models, including mouse, zebrafish, and fruit fly, represent homogenous, inbred lines.  In other words, they have been bred this way to make the genetics easier.  Humans have heterogeneous genomes, and that’s why identifying a specific genetic cause of a disease has turned out to be difficult.  We see our cichlid fish and some other emerging evolutionary models as being appropriate to cobble together a better picture of genotype and phenotype.  These models exhibit heterogeneous genomes like humans and the genotype-phenotype map is likely to be more complex.

The second thing, as we mentioned a moment ago, people would like to biologically engineer teeth and make today’s ceramic restoratives obsolete one day.  To facilitate this, we want to understand the natural regenerative capabilities of dentitions.  This then becomes very interesting.  The primary model used to study the human dentition is the mouse, and the mouse does not replace its dentition at all. 

True.  Mice have a labial stem cell niche associated with their incisors.  But their incisors are not replaced (except in a few genetic mutants).  They are renewed via continuous growth.  Mouse incisors also tend not to take on complex shapes.  A gap in space and a gap in development exist in the mouse between the incisors and the molars.  Molars take on complex shape, but are not renewed and are not replaced.  In the fish that we study, we see all of those things actually occurring in the same teeth.  So, what you see are teeth in any single position in the mouth that are replaced, renewed, and can take on complex three-dimensional shapes during development. 

The biology is all there in the same tooth position.  We think this is really the ancestral situation.  Repair, replacement, and shape are coupled developmentally and genetically in an organism like our cichlids.  But then throughout the evolution of vertebrates, these processes have become decoupled in time and space.  What you see now in a mouse model, for instance, is that the molars are shaped but not repaired.  The incisors are repaired but not shaped. 

I think so.  It takes a meeting of the scientific disciplines.  Gareth Fraser, the lead author on the PLoS Biology paper, has an earlier degree in paleontology.  He’s an evolutionary developmental biologist, but his background is in paleo.  Darrin Holsey, an author on the paper and formerly a post doc in my lab, is a comparative evolutionary morphologist.  So it’s been great having these two guys interacting in the lab on this question.  That sort of integration takes input from lots of diversely trained biologists.  When you put them over the same microscope, it gets pretty interesting.