© MOMENT/GETTY IMAGESIn 1993, National Institutes of Health (NIH) researchers gave Eli Lilly’s promising experimental hepatitis B drug fialuridine to 15 patients as part of a small Phase 2 trial. It did not go well. Five of the patients died; two more required emergency liver transplants.
Fialuridine had undergone toxicity testing in mice, rats, dogs, and even nonhuman primates with no apparent ill effects, so the disastrous trial outcome was shocking. Years later, research revealed that drug uptake occurred via a transporter expressed in human mitochondria but not in the mitochondria of mice. Researchers now hypothesize that fialuridine builds up to toxic levels within mitochondria of the human liver, leading to catastrophic damage.
Fialuridine’s dramatic failure is an extreme case, but it exemplifies a fundamental challenge that the pharmaceutical industry faces: model organisms are imperfect predictors of drug safety and efficacy.
One way to improve the translation of drugs from animal models to the clinic is to make lab mice more human—specifically, to genetically engineer mice to express human proteins or to implant human cells or tissues. The “humanized” mice should, in theory, better represent human physiology and thus effectively reduce the number of unexpected results that crop up in clinical trials. The mice could also allow researchers to do preclinical in vivo studies of drugs that previously had no good animal models available.
Last year, a team led by Gary Peltz, professor of anesthesiology at Stanford School of Medicine, tested fialuridine in immunodeficient mice whose liver cells had been partially destroyed and replaced by human hepatocytes, creating mice with chimeric livers. After just four days of receiving the drug at a high dose, the human cells in the mice’s livers began to fail, showing pathology reminiscent of that seen in human patients two decades earlier (PLOS Med, 11:e1001628, 2014).
“It’s essential to have a model like this as part of preclinical testing because the animal studies are clearly not sufficiently predictive,” Peltz says. He speculates that, had mice with humanized livers been available and used when fialuridine was in development, tragedy might have been averted.
Recent years have seen the development of mice with implanted human livers, tumors, pancreatic cells, diverse immune cell types, and even glial cells from the nervous system. Researchers are now using this new generation of humanized mice to study pathogenic infections, cancers, and more.
Alexander Ploss, an infectious disease researcher at Princeton University, notes that humanized mice will become all the more important in the face of dwindling support for medical research on chimpanzees. For some infectious diseases, such as hepatitis B and C, chimps are the only nonhuman animals that can be infected. But in the past few years, the NIH has declared its intention to wind down funding for research on these primates.
“There’s no question that human-liver chimeric mice and other humanized mouse systems will become very important to test curative therapies,” says Ploss.
Researchers first began to genetically engineer mice with human genes in the early 1980s. But some human systems were too dissimilar from those of mice for isolated genetic changes to make much of a difference. The first humanized mice with transplanted human cells, described in 1988, were created to host HIV infections (Science, 241:1632-39). Mike McCune, now a professor of medicine at the University of California, San Francisco (UCSF), began his medical career as a resident at the same institution six years earlier, as the AIDS epidemic was just beginning. Over the next several years, he saw staggering numbers of patients die and watched a devastated biomedical community scramble for more information about the virus. Unfortunately, chimpanzees were the only nonhuman animals that had been successfully infected with HIV. Using chimps was expensive and ethically problematic, and their HIV infections progressed to AIDS very slowly, making it difficult to use the animals to study the advanced stages of the disease. McCune knew that studying the virus in mice would be far more convenient, but HIV simply would not infect and replicate within rodent immune cells.
Fortuitously, in 1983, Melvin and Gayle Bosma, a husband and wife team working at the Fox Chase Cancer Center in Philadelphia, Pennsylvania, had developed a new mouse strain with a mutation in a gene associated with severe combined immunodeficiency (SCID) that eliminated the rodents’ mature T and B cells. As the HIV/AIDS epidemic ramped up in the mid-’80s, Leonard Shultz of the Jackson Laboratory in Bar Harbor, Maine, sent some of these immunodeficient SCID mice to McCune, who, with his Stanford postdoc advisor Irving Weissman and their colleagues, successfully injected them with human immune progenitor cells. McCune also implanted human thymus and lymph node tissue in the mice to direct the progenitor cells’ differentiation into functional human T cells and B cells. The resulting mice could harbor HIV. Around the same time, researchers from the Medical Biology Institute and Veterans Administration Medical Center in La Jolla, California, used SCID mice to create mice with humanized immune systems by injecting them with mature white blood cells (Nature, 335:256-59, 1988).
Since that time, researchers have created variations on these immune-reconstituted mice to understand the basics of HIV infection and to test HIV treatments. In 1990, the NIH awarded McCune a contract to use humanized mice to test HIV drug candidates, a project UCSF’s Cheryl Stoddart took over in 2000. “Our work basically showed more often than not that a new kind of drug just wasn’t promising,” Stoddart recalls. “I was proud of that because, from my point of view, I kept potentially dangerous or worthless drugs from going into Phase 1 clinical trials.”
Realizing the potential of immunodeficient rodents for developing models with more humanlike immune systems, researchers began to cross SCID mice with nonobese diabetic (NOD) mice—inbred mice that develop autoimmune diabetes due to impaired immune systems—to create models lacking natural killer cells as well as T cells and B cells. Then, beginning in the early 2000s, researchers at Japan’s Central Institute for Experimental Animals (CIEA) and the Jackson Laboratory’s Shultz independently developed NOD-SCID mouse strains that also had nonfunctional interleukin-2 receptor gamma chains, causing the dysfunction of a wide variety of cytokines and making it much more efficient to implant human immune cells in the animals. The Jackson Lab’s mice, called NSG mice, are sold through their website, while the CIEA version are called NOG mice and are sold by Taconic Biosciences. These NSG and NOG strains have served as receptive backgrounds for the infusion of both adult and immature human immune cells, as well as the implantation of human bone marrow, liver, and thymus tissues, creating what are known as BLT mice. These implanted tissues can help train infused human immune precursor cells to differentiate into mature human immune cells.
If the humanized mouse answers the question and helps in the evolution of a therapeutic or in the understanding of the disease, we’ve won.—Leonard Shultz,
The new immune-humanized mice have sparked a growing interest in the drug-development community, says Stoddart, who estimates that in recent years there has been a marked increase in the number of groups using these animals to study diverse infectious agents, including Ebola, dengue, and herpesviruses. “There was a huge revolution, or renaissance, because of newer kinds of immunodeficient mice that support much greater immune reconstitution,” she says. Researchers are also using the mice to study transplantation and the immunology of tumors, and to create models of autoimmunity, which may be useful for studying diseases such as lupus, rheumatoid arthritis, and multiple sclerosis.
In addition to the implanted human bone marrow, liver, and thymus, BLT mice develop human immune cells in the gastrointestinal tract, as well as in the reproductive tract, says University of North Carolina at Chapel Hill’s J. Victor Garcia, who in the mid-2000s helped develop BLT mice and was the first researcher to employ them for the study of HIV. “It opens the door for experiments that were literally impossible to do before.” Using these mice, Garcia and his colleagues have studied oral, rectal, and vaginal HIV transmission, as well as treatments to prevent transmission. For instance, Garcia was able to use BLT mice to show that the HIV drug Truvada, which was in clinical trials for prophylactic use at the time, prevents HIV infection via the vagina (PLOS Med, 5:e16, 2008). He later showed that it also prevents infection following rectal and intravenous exposure (PLOS ONE, 5:e8829, 2010).
Garcia suspects that in the coming years, there will be many more examples of BLT mice guiding the development of drugs preclinically. “If you think about it, humanized mice for HIV research like this are not even ten years old,” he says. “It’s going to take a little while for you to feel the impact that these new technologies have on drug development.”
PLOS MEDICINE, 4: e1001628, 2014While immunodeficient mice were initially furnished with humanlike immune systems for HIV research, researchers quickly realized the value of such immunodeficient mouse models for studying all manner of diseases. Cancer researchers, for example, can implant human tumors or cancer cells into the mice to create patient-derived xenograft (PDX) mice. Because PDX mice lack immune systems, they do not reject human tissues. Researchers can use these mice to observe the responses of individual patients’ cancers to therapies. (See “My Mighty Mouse,” The Scientist, April 2015.) “We know pretty well from years of research that testing [drugs] in the dish is not really predictive,” says cancer biologist James Mulloy of Cincinnati Children’s Hospital Medical Center in Ohio.
Newer humanized mice on the scene carry transplanted human hepatocytes, or liver cells. Hepatocytes are notoriously tricky to grow outside the body, and they generally do not divide in culture. And some human liver-infecting viruses such as hepatitis C won’t infect rodents. Researchers studying these viruses thus had to instead rely on finicky human cell cultures or face the ethical challenges of using chimpanzees.
In 2001, a team of researchers led by the University of Alberta’s Norman Kneteman developed a mouse model with a humanized liver and became the first group to infect liver-humanized mice with hepatitis C. The researchers started with immunodeficient mice engineered to express a transgene that kills liver cells. They then surgically implanted human hepatocytes into the dying mouse liver. Because of the liver’s natural regenerative capacity, the cells thrived and multiplied, leaving the mice with functional new livers that were partly human (see photograph)—and readily infected with hepatitis C (Nat Med, 7:927-33). This liver-chimeric mouse is now sold by KMT Hepatech, an Alberta, Canada–based company cofounded by Kneteman. Since then, other researchers have used similar strategies to develop different mouse strains with humanized livers. Oregon-based Yecuris and Japan-based PhoenixBio also sell mice with humanized livers, and Taconic will soon release a new liver-humanized mouse model.
In addition to aiding in hepatitis C research, the mice may also serve as models for malaria or for hepatitis B, a disease that drugs currently can control but not fully eradicate from the liver. German researchers used Kneteman’s humanized-liver mice to evaluate the lipopeptide drug Myrcludex-B for fighting hepatitis B, and found that it inhibited the virus from infecting and spreading among the animals’ implanted human liver cells (Nat Biotechnol, 26:335-41, 2008; J Hepatol, 58:861-67, 2013). Humanized mouse studies have also provided evidence of efficacy that was submitted as support for the drug’s clinical development and helped researchers determine proper doses, according to Stephan Urban, the hepatitis researcher at University Hospital Heidelberg in Germany who first identified Myrcludex-B’s potential. The drug is now in Phase 2 clinical trials for treatment of hepatitis B and for treatment of hepatitis B–hepatitis D coinfections.
For an even more humanlike model, Ploss and others are working to create mice with both humanized livers and humanized immune systems. These animals will provide a better model of hepatitis pathogenesis than immunodeficient mice and make it possible to study vaccines or immunotherapies for any liver-dwelling pathogen. “Many times . . . the rodent parasites mislead us,” says the University of Washington’s Stefan Kappe, a malaria researcher at the Center for Infectious Disease Research in Seattle. “Now we can directly work with human parasites in humanized mice, which I think brings us closer to making predictions for clinical studies.”
Drugs and disease
This June, researchers created mice with humanized livers that modeled a human metabolic disease (Nat Commun, 6:7339, 2015). Patients with a genetic disorder called familial hypercholesterolemia cannot properly metabolize low-density lipoprotein (LDL) cholesterol because they lack functional copies of the gene for the LDL receptor (LDLR). Researchers have previously engineered mice lacking the LDLR gene, but they do not fully recapitulate the disease because mice metabolize cholesterol using different pathways than humans do. A team led by Karl-Dimiter Bissig of Baylor College of Medicine in Houston, Texas, transplanted liver cells from a 7-year-old girl with the disease into mice, which then suffered from a reduced ability to metabolize cholesterol. When the researchers delivered functional copies of LDLR to the mice’s humanized livers via virus-mediated gene therapy, the animals’ cholesterol problems disappeared.
In addition to serving as models of liver-based infections and diseases, liver-humanized mice are also useful for testing drug metabolism and toxicity. Human and mouse livers have different enzymes for digesting drugs, and drug metabolites vary between the organisms. This can cause trouble if a uniquely human metabolite, or one that is produced at higher levels in humans, is toxic.
“For pharma, the disaster is that you find out late in Phase 3 . . . that you have a novel human metabolite that is not exposed to any extent in [preclinical studies],” says Ian Wilson, formerly at the U.K.-based AstraZeneca and now a professor at Imperial College London. “Then you are going to have a very difficult conversation with the regulators.”
Testing drugs in humanized mice may prevent this sort of unwanted, late-stage discovery, says Wilson, who was involved in early tests of the mice at AstraZeneca. Among other experiments, Wilson and his colleagues tested troglitazone, a formerly FDA-approved type 2 diabetes therapy that was withdrawn from the market in 2000 due to acute liver toxicity in a small number of users. “What we found was these [humanized] mice did produce a metabolic profile that was more similar to humans than to [wild-type] mice,” Wilson said (Xenobiotica, 42:503-17, 2012).
Making a purchase
Liver-humanized, or liver-chimeric, mice purchased commercially can be quite expensive, Wilson notes, limiting their general use as models for toxicity testing. But if prices fall, these animals could become a routine part of drug development, he adds. Indeed, most companies are already dipping their toes in, adds John Bial, president and chief executive officer at Yecuris. “Every single major pharma group has a program [in] chimeric mice to some level or another. It’s really gained a lot of steam in the last three to five years.”
And researchers continue to experiment with new types of humanized mice. The Jackson Lab’s Shultz, for instance, is working with virologist Jennifer Wang at the University of Massachusetts Medical School in Worcester to create a new humanized mouse model of diabetes by ablating beta cells in immunodeficient NSG mice and then implanting the animals with human islets. Islet-humanized mice infected with coxsackievirus B, a suspected trigger for type 1 diabetes, developed hyperglycemia (Diabetes, 64:1358-69, 2015). And a group based at the University of Texas Medical Branch last year engrafted human fetal lung tissues under the skin of NSG mice to study the emerging Nipah virus (PLOS Pathog, 10:e1004063, 2014). The new model allows researchers to study more specifically how the virus infects humans and interacts with the human immune system.
Researchers are even trying to humanize the brain. Investigators at the University of Rochester Medical Center in New York injected the progenitors of human glial cells into the brains of newborn mice, where the human glia matured and expanded throughout the mouse forebrain (J Neurosci, 34:16153-61, 2014). (See “When Does a Smart Mouse Become Human?” The Scientist, July 2015.)
But Shultz says that no matter how many human cells and tissues are incorporated into these models, they will never themselves be truly human. “They’re tools,” he says. “If the humanized mouse answers the question and helps in the evolution of a therapeutic or in the understanding of the disease, we’ve won.”