By David N. Leff

Among the numerous strains of strange laboratory mice is one variety afflicted with shivering. Its body vibrates nonstop at high frequency. This aberrant behavior derives from a neurological mutation on a gene somewhere in the animal's genome. But where?

To link that mutant gene with its neurodegenerative trait, molecular geneticist Edward Rubin employed a novel genome-searching technique.

"Mapping studies," he told BioWorld Today, "had indicated a candidate region of the genome containing the locus where the vibrator mutation was present. It had been mapped to between 500,000 and 1 million base pairs [bp]."

So Rubin and the biology group at the University of California's Berkeley National Laboratory, which Rubin leads with molecular geneticist Eric Lander, of Massachusetts Institute of Technology's Whitehead Institute, "created what we call an in vivo library of this region. What this means," Rubin explained, "is that we broke the candidate region into large DNA fragments, and propagated them in a series of normal transgenic mice. Then we bred these mice, carrying the different pieces to mice that contained the vibrator mutation."

Among their progeny were animals that carried that mutant gene, but which no longer vibrated because they had also inherited the piece of DNA containing the normal gene.

"That was fun," Rubin observed. "But the vibrator experiment also served to demonstrate our new departure in matching unknown genes of interest with the complex traits they create.

"Constructing transgenic mice with random pieces of DNA," Rubin said, "differs, I think, from how it's been done previously. Previously, we introduced defined genes into transgenic mice and then studied these genes. Here, we looked at the behavior of animals carrying large segments of undefined genes, and then looked at their behavior, to identify sequence with function."

His report of a more human-oriented gene search appears in the May issue of Nature Genetics, under the explicit title: "Functional screening of 2 mb of human chromosome 21q22.2 in transgenic mice implicates minibrain in learning defects associated with Down's syndrome."

Here, Rubin's quest was to track down the gene or genes responsible for mental retardation, the complex trait that is one of the many hallmarks in Down's syndrome (DS).

DS, which affects an estimated one million Americans, is the leading genetic cause of mental retardation in developed countries. The syndrome's genetic cause is a triplicate chromosome 21, rather than the pair bequeathed by father and mother. Rarely, only a part of 21 is present in triplicate.

The Key: A Living DNA Library

This suggested that DS may be due to a compact number of genes, and led Rubin's group to construct an in vivo library consisting of sets of transgenic mice containing different adjacent segments of human chromosome 21, microinjected into their early embryos.

They cloned these anonymous human DNA fragments in artificial yeast chromosomes (YACs). "YACs," Rubin pointed out, "can propagate up to several million bps of human DNA, so we were able to make many copies of the two-million-base-pair DS region. It allowed us to clone much larger pieces of DNA than in bacterial host cells."

One murine equivalent of human I.Q. testing measures the transgenic animals' ability to learn and remember the location of hidden platforms in an opaque swimming pool.

Two of four transgenic cohorts, carrying different fragments of DS genomic DNA, evinced poor learning and memory on the water maze, indicating that they were expressing a human gene for DS mental retardation.

An interesting event further refined this evidence.

Because YACs sometimes break apart during microinjection, a separate panel of transgenic mice turned up with only 180 kilobases of human DNA, rather than the 570 in another set of animals. Both performed badly on learning and behavior testing, a clue that the errant gene lurked in the shorter segment.

In fact, Rubin's team discovered that about 100 kilobases of that stubbier segment matched the DNA sequence of a gene in fruitflies (Drosophila melanogaster), which suffer the insect equivalent of poor learning and memory. Fruitfly geneticists have named this the minibrain gene. It has a homologue in the human genome.

"There's a long history of looking at expression of human or mouse genes in the mouse," Rubin pointed out, "and seeing that, the impact of those genes on the behavior or physiology of the mouse can almost invariably be linked to the impact of the same genes on the physiology in humans."

Finally, The Gene Defined

His DS study revealed a gene on the murine counterpart of human chromosome 21 "that is a dual-specificity protein kinase. We're not sure exactly, mechanistically," Rubin continued, "how an extra copy of it, as in the human trisomy 21, causes abnormal neuronal development in mice, or in fruitflies. But it suggests that normal levels of this gene are required for normal neuronal development."

The Berkeley researchers also compared neurons in brain slices from their transgenic mice with controls. "We did not see any histological abnormality in these animals," Rubin said. "Most disorders of mentation are not associated with gross anatomical changes."

The DS project, he observed, "was sort of a test case of our approach. It demonstrated that we were really able to identify genes that are involved in complex traits, based on function rather than prior assumptions."

He continued: "I think the DS results with regard to chromosome 21 provide information in trying to understand one of the features seen in this disorder. I think it proves this genetic approach, and that it will help us in the difficult task of linking sequences to function, which is one of the great challenges of human genetics, medical genetics and genomics."

An editorial accompanying Rubin's article in Nature Genetics suggested that his in vivo library strategy "has other important applications in mammalian genetics -- for example, in identifying oncogenes."

"Right now," Rubin concluded, "our group is interested in trying to improve the technique, so we can more efficiently look at larger regions of the genome. We're interested in applying this approach to complex human traits, such as asthma, hypertension and schizophrenia." *