A Model Mouse

When it comes to biomedical research, it’s a zoo out there. To understand and develop treatments for human diseases, scientists are increasingly turning to monkeys, dogs, rodents, and other animals with seemingly related conditions or symptoms. In part because of biologists’ growing capability to genetically engineer mice, the number of these animal models has exploded over the past decade. Jackson Laboratory in Bar Harbor, Maine, which sells mice for research, lists more than 2,500 rodent strains in its catalog and adds about 100 annually. Scientists have used these animals to study such diverse conditions as diabetes, osteoporosis, obesity, cancer, Alzheimer’s disease, and infertility, and the work has led to new treatments.

Yet how completely does an abnormality in a particular mouse or other animal actually emulate a human disease? That’s a question that biomedical investigators regularly confront. Sometimes it’s difficult to determine whether the animal model and human disease share an underlying mechanism. Consider the controversial mouse strain that develops a condition closely resembling rheumatoid arthritis, an inflammatory disease that degrades the joints of more than 2 million people in the United States.

Seven years ago, immunologists Christophe Benoist and Diane Mathis of the Joslin Diabetes Center in Boston created the mice for their research on the immune system.

The researchers were surprised when the genetically engineered rodents spontaneously developed inflamed joints. As they investigated the mice further, Benoist and Mathis slowly began to figure out why these rodents’ immune systems triggered arthritis.

For a time, buoyed by a surprising discovery, they were even hopeful that they could use the mice to resolve what goes on in human rheumatoid arthritis.

Other researchers have been dubious. “As far as I’m concerned, the evidence for the mechanism of rheumatoid arthritis has to come from rheumatoid [patients],” says Jonathan Edwards of University College in London. “You can make a mouse have arthritis 100 ways, but it doesn’t tell us anything, I’m afraid.”

Indeed, recent experiments suggest that the Benoist-Mathis mice won’t provide direct insight into rheumatoid arthritis. That doesn’t mean the mice are a bust, however.

Far from it. Even if the animals don’t reflect what occurs in human rheumatoid arthritis, many scientists, including Edwards, are optimistic that the mice will greatly aid studies of the disease and other inflammatory conditions.

Joint attack

Scientists are eager for a good animal model of rheumatoid arthritis because the disorder continues to baffle them. It strikes the synovium, the thin layer of tissue lining the area of a joint where two bones meet. This tissue secretes a lubricating fluid that aids joint movement. In people with rheumatoid arthritis, inflammation within the synovium causes pain and limits the flexibility of a joint. As the disease progresses, it can debilitate a person by slowly eating away the joint’s cartilage and bone. This condition differs from the more common osteoarthritis, which results from wear and tear on the cartilage within aging joints.

Benoist and Mathis became intrigued when their genetically engineered mice, within several weeks of birth, spontaneously developed joint inflammation with many features of rheumatoid arthritis. The inflammation focused on the synovium. Moreover, as in rheumatoid arthritis, certain immune sentinels called T cells invaded the inflamed rodent joints. Bone and cartilage began to erode, and the condition incapacitated the mice by the time they were a few months old.

At first glance, the mice studied by Benoist, Mathis, and their colleagues seemed to support earlier research suggesting a central role for T cells in rheumatoid arthritis. The arthritic rodents had been created by breeding a strain of diabetes-prone mice with a strain genetically engineered to have altered T cells.

By the mid-1990s, however, many scientists had become skeptical that T cells are key to rheumatoid arthritis. Indeed, in later studies with the mice, Benoist and Mathis found that T cells weren’t central to the mouse condition. In early 1999, they reported that another class of immune cells, B cells, and the antibodies they produce are at the heart of the problem. For example, the researchers can trigger arthritis in healthy, typical mice by simply injecting them with antibodies from the engineered mice.

That caught the attention of other arthritis researchers, and later that year, the Boston scientists reported something even more startling. They determined that the antibodies that triggered arthritis target an enzyme called glucose-6-phosphate isomerase, or GPI. This enzyme helps metabolize sugar into energy. What made the discovery so puzzling is that GPI is produced within virtually every cell of an animal. How could something so widespread lead to a disease focused on the joints?

Benoist recalls that he and Mathis initially had trouble believing that GPI was the target of the arthritis-generating antibodies. He told his coworker who had made the initial lab finding, “No, that can’t be. It must be [from] contamination.”

Unwelcome complement

Antibodies had appeared in explanations of rheumatoid arthritis that were popular in the 1960s and 1970s. That’s in large part because one of the most common features of the disease is the presence of an antibody called rheumatoid factor in a person’s blood. More than 70 percent of people with the disease produce this factor, which is an antibody to other antibodies, but people with inflammatory diseases other than rheumatoid arthritis also frequently have rheumatoid factor in their blood. Moreover, other types of antibodies have been implicated in rheumatoid arthritis. Most recently, scientists have focused on antibodies to joint-specific molecules, such as the collagen.

Finding that antibodies to a widespread protein such as GPI generate joint inflammation added a new wrinkle to the antibody-based theories of rheumatoid arthritis. “This is very intriguing because it tells you that something ubiquitous can lead to an organ or tissue-specific problem,” says immunologist David Pisetsky of Duke University in Durham, N.C.

But is it relevant to arthritis in people? Researchers studying rheumatoid arthritis are particularly wary of findings about antibodies directed at the body’s own molecules, which scientists call autoantibodies.

“Over the years, many antibodies have been identified as potentially the key factor in rheumatoid arthritis. In each case, subsequent research has indicated otherwise,” notes Gary Firestein of University of California, San Diego.

Yet the GPI story gained momentum. Last year, a research team headed by Henrik Ditzel of the Scripps Research Institute in La Jolla, Calif., reported that in one study, 44 of 69 people with rheumatoid arthritis had antibodies to GPI circulating in their blood and in the fluid bathing their joints. In contrast, only 3 of 107 healthy people tested had such antibodies.

Three recent reports fleshed out the story further. In the February Immunity, the Benoist-Mathis team revealed several immune molecules that have to be present in the mouse for the GPI-binding antibodies to cause arthritis. For example, there are about 30 proteins known collectively as complement that the immune system uses to tag and destroy microbes. The researchers found that mice lacking certain of these proteins, such as the ones called C5 and C3, don’t develop arthritis when injected with the GPI antibodies. That indicates that antibodies produce joint inflammation by accidentally triggering the complement system.

Other researchers have considered inhibiting complement proteins as a treatment for rheumatoid arthritis. In fact, as least one biotech firm is testing such a strategy with a drug against C5. On the basis of his mouse studies, Benoist speculates that inhibiting C3 might block arthritis even more safely and effectively.

In the April Nature Immunology, two other reports provide some insight into how an antibody response to GPI can afflict joints but not other tissues containing that protein. In one study, Paul Allen of Washington University in St. Louis injected normal lab mice with radioactively labeled antibodies to GPI and used positron emission tomography to follow them through the body. Within minutes, the antibodies accumulated in the joints of the animals’ limbs. Because antibodies can’t bind to proteins inside cells, Allen and his colleagues conclude that some GPI must exist outside cells in normal mouse joints and that the antibodies latch on as they circulate through the tissue.

In the second report, Benoist, Mathis, and their colleagues confirm that there is exposed GPI in normal mouse joints. It coats cells and the cartilage there. Because cartilage is a noncellular material, it lacks the proteins that cells typically use to protect themselves from the complement system. The researchers propose that when antibodies bind to exposed GPI within joints, complement proteins attach to the antibodies and set off an unchecked molecular cascade that leads to inflammation.

No easy answers

“It’s very elegant science,” Firestein says of Benoist and Mathis’ scenario for joint inflammation in their mice. Nevertheless, Firestein and other scientists continue to challenge the idea that the rodents represent a model of rheumatoid arthritis in people. Indeed, the role of GPI in the human disease has recently become less clear than it seemed to be in 2000.

Working with Hani El-Gabalawy of the University of Manitoba in Winnipeg, Benoist and Mathis have recently looked for GPI antibodies in the blood of several hundred people who are healthy or have one of several joint diseases called arthropathies. The researchers haven’t yet published the results, but they presented a preliminary analysis in October at an American College of Rheumatology meeting in San Francisco.

Ditzel and his colleagues are also now finding GPI antibodies in people with non-rheumatoid forms of arthritis.

“The bottom line is that these antibodies have no specificity for rheumatoid arthritis. They are detectable in a wide spectrum of arthropathies, and indeed in a number of healthy [people],” El-Gabalawy tells Science News.

Even Benoist has concerns as to whether GPI will prove to be central to human rheumatoid arthritis. “I would not necessarily want to stick my neck out too far,” he says.

But if the animal model doesn’t replicate the human disease, researchers may still find it useful. El-Gabalawy notes that injecting GPI antibodies into mice allows investigators to quickly and consistently generate arthritis. He points to a recent study in which another research group used this strategy to confirm that the gradual bone destruction in arthritic joints stems from the activity of cells called osteoclasts.

Edwards sees the opportunity to use the mice to investigate aspects of his own antibody-based theory of rheumatoid arthritis, in which rheumatoid factor is the central culprit. He has drawn attention to this idea by pursuing a controversial treatment in which he temporarily eliminates some antibody-producing immune cells in patients (SN: 11/4/00, p. 294: Killing immune cells thwarts arthritis). A small trial with 20 patients brought promising results, and the outcome of a larger trial is due later this year.

Although Edwards doesn’t think GPI antibodies are relevant to rheumatoid arthritis in people, he notes that the antibody-induced joint inflammation in the Benoit-Mathis mice resembles the human disease process closely enough that the rodents deserve further scrutiny.

“What’s exciting about this mouse is that we may finally find out how [autoantibodies] cause inflammation,” he contends.