Jacob Halaska, ©Index Stock Imagery

Researchers like mice. US government statistics reveal that the whiskered ones show up in 90% of all experiments. Mice come cheap, procreate often, and die fairly quickly. And although evolution separates mouse from human by an estimated 75 to 100 million years, biologically and genetically speaking, they share a lot; as much as 85% of the DNA in mice is the same in humans.

The research ground that mice have domineered for a century, however, is receding somewhat, as technologies and new discoveries make working with other organisms more efficacious. Rats replicate high blood pressure and atherosclerosis more readily than mice; in addition, their bigger brains are often easier to manipulate. Transparent zebrafish easily show circulating blood cells, making them useful for studying anemia. The physiology of rabbits more closely resembles humans than does that of mice or rats. And Caenorhabditis elegans has been manipulated to provide a simple model of Huntington disease.

"Clearly, there are not models in mice for all diseases, and sometimes when you make a knockout, you don't get a sick mouse," says pathobiologist Mark E. Haskins, School of Veterinary Medicine, University of Pennsylvania. Nephrologist Thomas Coffman, Duke University Medical Center, says it's time to find new lab denizens. "There's a general need to find better experimental models across the board."

Some are doing so. Yale University geneticist Tian Xu favors the fruitfly. About two-thirds of human cancer genes have Drosophila homologs, and a high degree of conservation is observed in neurological, cardiovascular, endocrine, and metabolic diseases, Xu says. He uses Drosophila to study tumorigenesis and neurodegeneration.

Others, like Haskins, work with larger models, which often live 15 years or more, and therefore can show consequences of treatment over time. Repeated sampling and sophisticated diagnostic work, such as magnetic resonance imaging and computed tomography scanning, are easier with larger models, he says. Moreover, model size is particularly important when testing gene therapy; the viral vector amount must be appropriate, and delivering it to a cat or dog brain presents more challenges than the tiny murine brain, adds Haskins.

But no one is abandoning the mouse, not by a long shot. Rockefeller University immunologist Jeffrey V. Ravetch says that the advantages of using mice usually outweigh the drawbacks, and better murine models continually appear. The National Institutes of Health recently funded eight animal engineering labs that together comprise the Animal Models of Diabetic Complications Consortium. The grants cover any animal, and all but one grant recipient, who prefers pigs, are using mice.

THE MURINE MODEL MATURES Ravetch's field once relied mostly on in vitro experiments, but today murine models allow in vivo study of the complex pathways involved in tolerance and immunity. Genetically engineered models have changed completely the study of human disease. The in vivo revolution began in the 1980s when scientists first devised techniques to create precise murine knockouts, such as mice with silenced or deleted genes, and transgenic mice. BioMedNet's Mouse Knockout & Mutation Database lists more than 7,500 mice, many representing human diseases, including neurological disorders, various cancers, cystic fibrosis, muscular dystrophy, and osteoporosis.

The efforts to achieve an effective Alzheimer disease model are noteworthy. Before 1995, no murine Alzheimer model existed. In that year, Elan developed a well-regarded model for amyloid deposition, says NIH neuroscientist John Hardy, but since the company was privately owned, most researchers could not use it. So, University of Minnesota researchers developed a similar mouse, colloquially called Hsiao, or Tg2576, and made it widely available. Soon after, researchers at the University of South Florida crossed this mouse with a presenilin transgenic model, creating an even better replica of amyloid deposition, Hardy says. Then, in 2000, scientists made a model for neurofibrillary tangles, the other main factor in Alzheimer pathology. In 2001, they crossed a tangles mouse with an amyloid mouse, making the most complete Alzheimer model to date.

"The Alzheimer's models are improving all the time. And just as important, they're becoming more widely available," says Hardy, previously at the Mayo Clinic. A major setback occurred, however, when Elan and its partner, Wyeth Pharmaceuticals, withdrew its Phase II vaccine trials after participants developed brain inflammation.

Tammy Irvine, Rear View Illustrations

MAKING A MODEL Simply put, creating a model is time-consuming, complicated, and often has unpredictable results, á la Elan. Many universities now have special in-house facilities where geneticists make models-to-order for staff.

The most common method for making transgenic mice involves microinjecting DNA into the pronuclei of one-celled embryos, says geneticist Roberta Franks, who directs the University of Illinois' Chicago Research Resources Center Transgenic Production Service. First, scientists construct a transgene with conventional recombinant DNA. Tissue-specific transgene expression requires a well-characterized promoter specific for the tissue of choice, or if widespread gene expression is desired, a ubiquitous promoter may be used, Franks says. To precisely control temporal expression patterns, she explains, coding elements for conditional, or inducible, transgene activation may be included on the transgene construct. Then, using an ultrathin pipette, a technician injects the purified transgene fragment into the male zygote pronucleus, which is larger than the female's. Surviving embryos are reimplanted into the oviducts of surrogate mothers; efficiency ranges between 5% and 30%, Franks says. The successfully created pups are identified using PCR amplification or Southern hybridization of tissue samples.

Making knockout mice involves targeted DNA insertion, typically using a vector in which a marker gene is "flanked by large stretches of cloned genomic DNA homologous to the endogenous target gene," Franks says. The vector reaches the embryonic stem cells (ESCs) by electroporation, which delivers strong, brief pulses of electric current that punch holes in cell membranes. These pluripotent cells, which derive from the murine embryonic blastocyst, can be grown in culture while still retaining their potential to repopulate the entire embryo.

Following electroporation, the ESCs are grown in a solution that selects for those cells that have incorporated the vector, and the surviving cells are collected and expanded. The cells that incorporated the altered genetic material into the homologous target gene (usually an essential exon), effectively replacing the wild-type allele, are then microinjected into blastocyst-stage murine embryos, which are reimplanted into the surrogate mothers' uteri. If the ESCs contribute to the resulting animals' germ lines, the genetically altered, targeted alleles can be transmitted to offspring.

Mice engineered with a deleted gene in all their cells often die early, Franks says, indicating that the protein encoded by the targeted gene may be crucial for normal embryonic development. Other methods now exist, however, to delete genes in specific tissues, at specific times, in the life cycle.

All in all, creating a relevant model is usually possible, says neurogeneticist Huda Zoghbi, Baylor College of Medicine. In her first murine model of Rett syndrome, she deleted the gene MECP2 responsible for the syndrome; the animals developed a severe form of the disease and died very young. For her research she needed a more typical model of Rett syndrome, which is the leading cause of mental retardation in girls.

After reexamining the disease's genetics, Zoghbi realized that if she developed a male model, it would lack the complication of the first model's X chromosome activation. So rather than totaling deleting MECP2, she designed a mutant mouse that produced a partially functional MECP2 protein. "Lo and behold, we got all of the features we were looking for," she says.

FROM BUZZ TO BOWWOW In his research, Xu induced tumors in flies by genetic screens in mosaic animals (a technique not possible in mice) to remove a tumor-suppressing gene from a small number of the insects' somatic cells. As in patients with cancer, the flies carried the mutation in some cells, and tumors consequently developed, Xu says. He then saved the flies from tumorigenesis by introducing a human LATS1 (large tumor suppressor) gene into a model lacking the fly homolog. His lab also generated mice deficient for LATS1; tumors developed in these mice as well.

The work showed that an invertebrate tumor suppressor is also a tumor suppressor in mammals, Xu says. It also uncovered evidence that LATS1 is a negative regulator for enzymes that drive the cell cycle, he adds. Once initial studies are finished on mice or other small organisms, researchers sometimes turn to larger models, such as cats or dogs. Besides the advantages of size and longevity, diseases that affect humans will sometimes spontaneously develop. Haskins works specifically with these kinds of cat and dog models, running an NIH-funded resource that makes the models available for research.

The animals "come to us because they have been sick, so they're relevant models," he says. He oversees colonies of cats and dogs with 25 different genetic diseases, including lysosomal storage diseases, congenital heart disease, skin disease, erythrocyte enzyme disorders, congenital deafness, and X-linked severe combined immunodeficiency. His research focuses on applications of gene therapy for lysosomal storage diseases, which are caused by inherited genetic defects that result in an enzyme deficiency. Haskins has successfully prevented clinical manifestations of one such disease, mucopolysaccharidosis VII, a rare and inherited degenerative disorder that is generally fatal in both dogs and humans.

Most researchers do not work with larger models, despite the benefits. "It's not easy to take an animal like a dog into the lab. It's expensive, and dogs and cats are not easy to manage," Haskins says. Despite the difficulty of developing models that perfectly mirror human disorders, they can provide endless clues to human and animal disease. Says Haskins: "Animal models can be sentinels."

Jennifer Fisher Wilson (jfwilson@snip.net) is a freelance writer in Haddonfield, NJ.