We all know people with Down Syndrome. It's one of the most common serious congenital disorders, affecting one in 700 births, and the suite of stereotypical morphological changes make it readily recognizable. It's also usually traceable to what seems to be a discrete cause: a nondisjunction in meiosis that gives individuals an extra copy of chromosome 21, so that instead of the normal two copies, Down Syndrome individuals have three. Somehow, this trisomy, which changes gene dosage, disrupts development in subtle ways and leads to a wide range of characteristic modifications.

Other chromosomes than 21 can also occur in triplicate, but the difference is that they all (except for the sex chromosomes, which are a special case) are lethal. One reason Down Syndrome may be less destructive is that human chromosome 21 is our smallest chromosome in terms of nucleotide length. The total sequence length is 46,976,537 bases, but if you discount the telomeres, centromeres, and heterochromatic regions, it's 'merely' 33,924,742 bases long, with 200-300 genes (the number is in flux; just about every paper I read reports a different value). You can browse through chromosome 21 at NCBI, if you'd like. It is a bit overwhelming and obscure; you'll find things like "keratin associated protein 19-5" and "liver phosphofructokinase" reside there. Stare all you'd like, but it's hard to get any insight into which of those genes are responsible or why having an extra copy would be deleterious.

Thirty four megabases and 225 genes is too much. We need hints. It would be useful to be able to pare that long laundry list down to a more manageable collection of items.

One way to reduce the list is to look for rare individuals who have only partial duplications of chromosome 21. This does happen; a chromosome can break, and a fragment attach itself to another, and the individual carrying that translocation can subsequently pass on both an intact chromosome 21 and the fragment to their children, and together with the intact chromosome 21 from the other parent, the children will have 3 copies of each gene contained in the fragment. This kind of analysis of 'natural experiments', in which finding genetic accidents helps us associate smaller bits of the genome with a phenotype, is a useful research strategy. One result, for instance, is that it has allowed us to identify a smaller region, about 4-5 Mb long, that seems to be important in producing at least some of the Down Syndrome effect. That region of chromosome 21 is called the Down Syndrome Critical Region, or DSCR.

In many ways, this is not an entirely satisfying sort of analysis. It depends entirely on what genetic variations fate hands us, and you certainly don't get to tweak the methods, shave off a few thousand nucleotides here, add a few thousand there, or ask someone with an interesting deletion to breed a few dozen times with this other person with a promising duplication…people object to that sort of thing. We need an animal model.

You can't just make a trisomic mouse, though. Here's the catch: recall from my discussion of synteny that blocks of genes get scrambled around to different chromosomes during the course of evolution. The genes that we have on our chromosome 21 are scattered onto mouse chromosomes 10, 16, and 17; and each of these chromosomes contain orthologs of genes found on other human chromosomes. What you have to do is splice together a kind of artificial chromosome containing the pieces you want, and introduce that into a mouse. We have such a beast: it's called the Ts65Dn mouse, and what it contains is an extra piece of mouse chromosome 16 spliced to a small piece of chromosome 17, and contains 104 of the genes we find on human chromosome 21. It's not a perfect copy of Down Syndrome, but it produces mice that live into adulthood, with characteristic skeletal changes (a shorter, broader skull and jaw, for instance) and changes in brain physiology. There is also another mouse strain with a smaller duplication, called Ts1Cje, that has extra copies of 81 of the genes found on human chromosome 21—and it exhibits a similar phenotype.

These are powerful tools! Try searching PubMed for "Ts65Dn" or Ts1Cje". There is a bloom of papers dissecting these mouse models of Down Syndrome, digging into the precise mechanics of the various deficits and working out the details of the disorder.

The diagram below illustrates what these artificial segmental trisomies contain. On the left is the human chromosome 21 for comparison, with the DSCR marked. Next to it is the Ts65Dn mouse chromosome, showing that it isn't a complete copy of Human 21, but it does encompass the DSCR. On the far right is the still smaller Ts1Cje construct, which also includes the DSCR.

In a paper by Olson et al., this reduction has been carried a step further. They have made a mouse construct that is trisomic for only the genes in the Down Syndrome Critical Region, labeled Ts1Rhr above, and they've also made one with a complementary deletion, called Ms1Rhr/Ts65Dn, which allows them the ability to very thoroughly control the dosage of those genes.

Here's a small surprise, however: the Ts1Rhr mouse, which is trisomic for the DSCR, doesn't seem to be a very good model for Down Syndrome.

The diagram to the right illustrates the kind of morphological analysis that was done. Mice don't look like people, obviously, and while we may be good at recognizing the overall appearance of a Down Syndrome person, it's harder to do in a mouse. What you can do is identify specific landmarks in the skeleton, measure proportions, and make a mathematical comparison. For instance, the bottom diagram is a human jaw, and the lines represent angles and distances between points on that jaw that can be measured and compared in different individuals. The purple lines indicate measurements that are significantly different between Down Syndrome individuals and euploid (or unaffected) individuals. In Down Syndrome, all of those purple lines are shorter than they should be, so that the person has smaller jaws.

In the segmentally trisomic Ts65Dn mouse, in the top diagram, similar parameters can be measured (the purple lines again), as well as others (the red lines) that all indicate a net reduction in jaw size. This is comparable to human Down Syndrome.

The middle figure, though, is the Ts1Rhr mouse, the one that is segmentally trisomic for just the DSCR. Those blue lines are parameters that are longer than they are in the wild type mouse—instead of a shorter, smaller jaw, these mice have the opposite, a longer, slenderer skull and jaw.

Scientific American has a short summary of these results, and they draw a conclusion we really ought to take for granted.

In an accompanying commentary, David L. Nelson and Richard A. Gibbs of the Baylor College of Medicine note that the findings refute the notion that possessing three copies of the DSCR is the sole cause of the cranial and facial features of Down syndrome. For their part, Reeves and his colleagues posit that genes contained in the DSCR interacting with other genes could be to blame. “The simplistic idea that just one of the hundreds of genes on chromosome 21 affect development no longer holds up,” Reeves remarks. “Now researchers can take a deep breath, accept that the syndrome is complex, and move forward.”

The key concept there is that development and morphology are consequences of interactions between genes (and with the environment, I would add). We shouldn't expect that we can reduce a phenotype to the simply additive effects of a few genes, either. The paper I use to illustrate this in my classes is by Elkins et al., in a far simpler system, Drosophila. A gene that was expressed in a suggestive pattern in the nervous system, Fasciclin I, could be knocked out, but it had no apparent effect. Another gene that was thought to be important, a signal transduction molecule called Abelson tyrosine kinase, could also be mutated, but with again no visible effect. Knock out both genes in the same animal, though, and wham, gross errors in nervous system formation occurred.

It is entirely conceivable that you could give mice extra copies of each gene found on human chromosome 21 one by one, and see no effect at all. There is instead some threshold network of genes that need to be perturbed in order to see a change.

One other comment bothered me a bit, but I know what they mean.

The notion that a few genes might be of critical importance in this syndrome is particularly attractive because such a simple model would bode well for possible therapeutic intervention.

I think we should be clear on one thing: there will never be a cure for Down Syndrome. Down Syndrome is a genetic and developmental disorder; the phenotype is the result of complex genetic and epigenetic interactions during embryonic development, not the discrete expression of a defective gene product late in development. The only possible 'cure' for the wide spectrum of problems in Down Syndrome would require breaking an individual down to a few cells and replaying all of those embryonic events over again, and that's the kind of treatment nobody wants.

However, these studies do promise to help isolate pieces of the syndrome and identify treatments that could correct symptoms: Down Syndrome individuals have an increased frequency of childhood onset leukemia, testicular cancer, and early onset Alzheimer's pathologies, for instance, and identifying the parts of the genetic puzzle that contribute to those problems could help figure out strategies for reducing them in Down Syndrome individuals. And understanding how these factors induce such problems in people trisomic for chromosome 21 might also reveal the mechanisms that cause the same problems in us euploids.

Akeson EC, Lambert JP, Narayanswami S, Gardiner K, Bechtel LJ, Davisson MT (2001) Ts65Dn—localization of the translocation breakpoint and trisomic gene content in a mouse model for Down syndrome. Cytogenet Cell Genet. 2001:270-6.

Elkins T, Zinn K, McAllister L, Hoffmann FM, Goodman CS (1990) Genetic analysis of a Drosophila neural cell adhesion molecule: interaction of fasciclin I and Abelson tyrosine kinase mutations. Cell 60(4):565-75.

Nelson DL, Gibbs RA (2004) The critical region in Trisomy 21. Science 306(5696):619-620.

Olson LE, Richtsmeier JT, Leszl J, Reeves RH (2004) Chromosome 21 critical region does not cause specific Down Syndrome phenotypes. Science 306(5696):687-690.