Mighty Mice

Scientists are still improving the humanized mouse model but are optimistic about its future role in evaluating AIDS vaccine candidates

By Andreas von Bubnoff

The flying cartoon rodent Mighty Mouse would be proud of his brethren. New mice, which are being developed by researchers, may not don capes or fight villains but they do possess other super powers, brought on by the fact that they have human immune systems. These so-called “humanized” mice represent a new frontier in the preclinical testing of experimental drugs, and possibly even vaccine candidates.

After about two decades of experimentation in transplanting human tissues into mice, the latest round of rodents can successfully harbor human immune cells and can be infected with human viruses that mice are usually not susceptible to. This is particularly significant for HIV. Although the mouse model is one of the most fundamental in all of biomedical research, its use as a model system for HIV has been severely hampered by the fact that the virus is a uniquely human pathogen. Studies in nonhuman primates are conducted with simian immunodeficiency virus, which is the closest approximation of HIV infection researchers suspect, but is still a different virus.

Humanized mice are already being used preclinically to evaluate new antiretrovirals, but most researchers say these mighty mice will require some further optimization before they can be used as a preclinical screen for AIDS vaccine candidates. Still, many are optimistic that the humanized mouse model will someday soon provide valuable preclinical information on human immune responses to candidate vaccines.

Early findings

The first humanized mice were generated around 20 years ago by transplanting human tissue into strain CB17 severe combined immunodeficiency (SCID) mice. These SCID mice have a mutation in a DNA repair enzyme that leads to impairment in the genomic DNA rearrangements, which are responsible for creating the cornucopia of different B and T cells. As a result, SCID mice can’t make B or T cells, and therefore don’t reject grafts of human tissue.

At that time, several groups transplanted human tissues into SCID mice. In 1988, a group led by Joseph McCune, then at Stanford University School of Medicine and now a professor of medicine at the University of California in San Francisco (UCSF), transplanted human fetal liver and thymus tissue under the kidney capsule, a well vascularized organ where transplants survive especially well (Science 241, 1632, 1988). The resulting mice, known as SCID-hu mice, developed a thymus-like structure containing human T cells that could be infected with HIV (Science 242, 1684, 1988). Another group led by Donald Mosier, then at the Medical Biology Institute and now a professor at the Scripps Research Institute in California, injected human peripheral blood leukocytes (PBL) into the peritoneal cavity of SCID mice, generating Hu-PBL-SCID mice (Nature 335, 256, 1988).

But these early SCID models had their drawbacks, says Markus Manz, who studies humanized mice as a group leader and vice director at the Institute for Research in Biomedicine in Switzerland. Manz says the recipient mice only produced human immune cells for a limited time, maybe for a few weeks or months. “All these [models] did not really sustain the human hematopoiesis,” Manz says. This is partly because CB17-SCID mice still maintain some innate immune responses such as natural killer (NK) cells and therefore eventually develop some mouse T and B cells, which causes them to reject some of the human cells. “The CB17-SCID mouse has quite active NK cells, and when the mice get older, T and B lymphocytes develop,” says Leonard Shultz, who studies humanized mice as a professor at the Jackson Laboratory, a leading genetics research lab in Maine.

Another drawback was that the HIV particles in infected SCID-hu mice remained mostly confined to the transplanted thymus tissue where the infection occurred, and as a result, biopsies were required to analyze the infected cells.

The next advance came in 1995 when a team led by Shultz crossed non-obese diabetic mice (NOD) with SCID mice (J. Infect. Dis. 172, 974, 1995). NOD mice develop autoimmune diabetes, but NOD-SCID mice don’t because they lack the ability to generate immune responses and, as it turned out, they were also much more accommodating for transplanted human tissue. These NOD-SCID mice supported better engraftment with human CD34+ hematopoietic stem cells (HSCs), the cells from which most other immune cells develop, for up to six months following transplantation, which is much longer than HSCs last in vitro, Manz says.

It was known that lower NK cell activity contributed to the increased ability of NOD-SCID mice to support human HSC engraftment, according to Shultz. But “the reason why this worked so much better was never clear,” says Manz. Then, a study published last year showed that the transplanted human cells engraft better in these mice than in SCID mice because of an additional impairment in their innate immune system. Macrophages in NOD-SCID mice have a version of a receptor that can recognize the human cell surface protein CD47 in the transplanted human cells and as a result, the NOD-SCID mouse macrophages don’t reject the transplanted human cells (Nat. Immunol. 8, 1313, 2007). “Now it’s become clear [that not only] is it the adaptive immune system that is impaired, but also the innate immune system [that] needs to be impaired to have the graft surviving,” says Manz (Nat. Immunol. 8, 1287, 2007). Still, even in NOD-SCID mice, the transplanted human HSCs did not develop into a functional immune system, Manz says, precluding any meaningful human-like immune responses. But a few years ago, several groups developed strains of mice with further impairments in their innate immune responses that allowed these mice, for the first time, to develop functional human B and T cells from the transplanted human HSCs. Some groups achieved this by crossing mice lacking the common gamma chain cytokine receptor—a common element of several different cytokine receptors—with NOD-SCID mice.

Manz transplanted human tissues into mice that lacked part of the recombinase activating gene (RAG2) and the common gamma chain cytokine receptor and as a result, these mice lacked B and T cells, as well as NK cells. His group also injected human CD34+ HSCs from cord blood into the liver of newborn mice instead of adult mice, which Manz says allowed the transplanted cells to be in a better growth environment (Science 304, 104, 2004). This is similar to the natural situation in humans where blood cells are formed in the liver during intrauterine development and immediately following birth. In the recipient mice, just like in humans, the transplanted human CD34+ cells then migrate from the liver into the bone marrow.

Still, even the new RAG2/common gamma chain mice can’t maintain transplanted human HSCs forever. Six months after transplanting 100,000 human HSCs in a mouse, most of them are gone, Manz says. By comparison, when the transplanted HSCs come from another mouse, they survive much better and expand 10- to 100-fold in just a few months. However, Ramesh Akkina, a professor of microbiology, immunology, and pathology at Colorado State University, says he has been able to maintain human cell engraftment in these mice for up to a year and a half by using CD34+ cells derived from fetal human liver (see Figure 1).

Figure 1. One approach for generating humanized mice. Injection of human CD34+ hematopoietic stem cells derived from human fetal liver into the liver of a newborn RAG2/common gamma chain mouse. Ramesh Akkina, a professor of microbiology, immunology, and pathology at Colorado State University, says human cell engraftment in humanized mice generated using this method can be maintained for up to a year and a half, generally the life of the engrafted mouse. Image provided by Akkina.  

The new RAG2/common gamma chain humanized mouse model allows researchers to maintain HIV infection in the mice longer and with higher viral titers than before (Proc. Natl. Acad. Sci. 103, 15951, 2006; Retrovirology 3, 76, 2006). In addition, HIV can also be found in the blood of these mice, with intraperitoneal HIV injection resulting in an HIV titer of about 10,000-100,000 copies/ml of blood. The infected mice also show chronic viremia, because the transplanted human HSCs keep making T cells that serve as HIV targets, as well as CD4+ T helper cell depletion.

“For the first time we seemed to have a real model of the adaptive immune system,” Manz says. “There was disappointment but now with the new models there is great hype.”

BLT: More than a sandwich

Meanwhile, J. Victor Garcia-Martinez, a professor of internal medicine at the University of Texas Southwestern Medical Center, developed another humanized mouse model called BLT (bone marrow-liver-thymus). In an extension of the 1988 McCune SCID-hu model, Garcia added intravenously injected fetal liver CD34+ cells that repopulate the bone marrow, which is where the reconstituted human immune cells come from, he says, adding that the bone marrow is a niche for the stem cells to grow and proliferate. In addition, he uses adult NOD-SCID mice as the recipients of the human cells.

Like RAG2/common gamma chain mice, the BLT mice also developed a human adaptive immune system, including human B and T cells, dendritic cells, and macrophages in blood, as well as in the spleen, lung, liver, and even the gut (Nat. Med. 12, 1316, 2006). “The entire gastrointestinal tract was reconstituted with human lymphoid cells, and that had never been seen before,” Garcia says. His group then showed that BLT mice could be infected with HIV by both rectal and vaginal transmission, since these sites were reconstituted with human cells (J. Exp. Med. 204, 705, 2007; PLoS Med. 5, e16, 2008). “Everything that we know is important for [HIV] transmission, pretty much was there,” says Garcia. Akkina’s group also showed successful mucosal transmission of HIV via vaginal and rectal routes in RAG2/common gamma chain mice (Virology 373, 342, 2008).

Putting them to use

Since these current humanized mice can be infected with HIV, they can already be used to test the efficacy of antiretrovirals (ARVs), according to Harris Goldstein, a professor of pediatrics, microbiology, and immunology at Albert Einstein College of Medicine. “We have tested ARVs that have been shown to work in humans and they all work in the mouse model,” she says. “I would suspect that that indicates that it’s predictive [of the human situation]. HIV is HIV, and it’s replicating in a human cell, so [if a] drug blocks replication it’s going to block it in the model.”

But researchers have observed that different doses of ARVs are required for humanized mice because they have a larger renal filtration surface area in proportion to their weight and therefore excrete drugs faster than humans. A study last year with the four licensed classes of ARVs showed that SCID-hu mice need to be dosed with about 12 times more of a given drug per kg body weight per day than humans to reduce viral load in the mice to a similar degree (PLoS One 2, e655, 2007).

That study’s lead author Cheryl Stoddart, an assistant professor in the division of experimental medicine at UCSF, has also been using humanized mice to pre-clinically test drug candidates as part of a program funded by the Division of AIDS at the National Institute of Allergy and Infectious Diseases, which enables small companies to have drug candidates screened free of charge. As part of that program, Stoddart has tested dozens of potential drugs, including the candidate drug Bevirimat, an HIV maturation inhibitor developed by Panacos Pharmaceuticals. This drug was able to reduce viral load in humanized mice and protect them from T cell depletion at plasma concentrations that are also achievable in humans after oral dosing (PLoS One 2, e1251, 2007). This drug candidate is now in Phase IIb clinical trials, according to a company spokesperson.

Fatah Kashanchi, a professor of microbiology and tropical medicine at the George Washington University Medical Center, is using RAG2/common gamma chain and NOD-SCID mice to test drug candidates such as an inhibitor of the kinase CDK9, a host cell factor that is needed for HIV transcription. He says using humanized mice for such experiments is far better than using nonhuman primates. “[Using nonhuman primates] would have taken me probably twice the amount of time [and] at least 100 times more money,” Kashanchi says. “These animals are almost like test tubes with four legs.”

The fact that both the BLT and the RAG2/common gamma chain deficient humanized mice can be infected rectally and vaginally with HIV may also enable researchers to test preventive approaches such as pre-exposure prophylaxis (PrEP) and microbicides. For example, Garcia recently showed that BLT mice are protected from vaginal HIV challenge after PrEP with truvada, a combination of the antiretrovirals emtricitabine and tenofovir (PLoS Med. 5, e16, 2008). “If you give the humanized mice that drug on a daily basis for seven days and in the middle of that expose them to HIV intravaginally,” Garcia says, “they are completely protected.” Garcia and Akkina have both started to use the humanized mice to test microbicides. “It offers additional relevant information regarding their potential to actually work when they are eventually tested in humans,” says Garcia, referring to previous microbicide candidates that have failed to protect against HIV infection in clinical trials. “There was very little, if any, evidence that these products would actually work.”

Others are using humanized mice to test novel approaches to treating HIV infection. For example, Pin Wang, an assistant professor of chemical engineering at the University of Southern California, in collaboration with David Baltimore, a professor of biology at the California Institute of Technology, is using the RAG2/common gamma chain deficient humanized mouse model to test if a lentiviral gene therapy vector is able to bind human target cells and introduce genetic material into them (see Engineering immunity, IAVI Report, July-Aug. 2008). And recently Shultz was involved in a study that used NOD-SCID/common gamma chain deficient humanized mice to show that anti-HIV small interfering RNA (siRNA) delivery to human T cells could suppress viral load in these humanized mice (Cell 134, 577, 2008). “Even though these humanized mice might need some further development [to test] primary immune function from HIV vaccines,” Shultz says, “I think they are ready right now to test other approaches such as siRNA.”

Limited response

Before these humanized mouse models will be viable for screening AIDS vaccine candidates, further optimization is required since the immune responses produced in response to infection are somewhat limited. For example, when Manz and his colleagues infected the humanized RAG2/common gamma chain mice with Epstein Barr Virus (EBV), which only infects human B cells, some of the humanized mice eventually developed B cell lymphomas (Science 304, 104, 2004). In EBV-infected humans, this only occurs in immune-suppressed individuals, suggesting the human immune cells in the mice could not control EBV infection in the same way as humans with a fully functional immune system. “This is proof that this [human immune system in the mice] is not the real thing yet,” Manz says.

In a later study, only one of 25 HIV-infected humanized mice showed HIV-specific antibodies and none showed obvious T-cell responses to the virus (Proc. Natl. Acad. Sci. 103, 15951, 2006). None of the other studies with the same mouse model showed strongly convincing HIV-specific immune responses, Manz adds.

In response to Dengue virus, however, Akkina’s lab has observed neutralizing antibody responses in RAG2/common gamma chain mice (Virology 369, 143, 2007). Still, in general the immune responses are sometimes limited, in part because the human T cells mature in the thymus of the mouse and are therefore “educated” on a mouse major histocompatibility complex (MHC) instead of a human MHC. This should be less of an issue in BLT mice because they contain human thymic tissue, and in fact Garcia has already observed immune responses to EBV infection. “They mount a classical MHC class restricted human T-cell response,” he says, adding that these were the first HLA-restricted cellular immune responses seen in a humanized mouse model (Nat. Med. 12, 1316, 2006). “[These] are probably the most comprehensive and most clear-cut human immune responses to a specific pathogen in a way that you measure it for humans,” Garcia says. He also demonstrated the production of human antibodies to HIV proteins in BLT mice (J. Exp. Med. 204, 705, 2007). Still, he points out that measuring immune responses in humanized mice will require developing new, more sensitive assays. “You need to miniaturize everything because the volumes and the number of cells are small,” says Garcia.

A mightier mouse?

To ready the humanized mouse model for evaluation of preclinical AIDS vaccine candidates, researchers are now trying to improve upon the immune responses seen in the current models. Manz is collaborating with Richard Flavell, a professor at Yale University School of Medicine, and the Tarrytown, New York-based company Regeneron Pharmaceuticals, on such a project with support from a grant through the Grand Challenges in Global Health Initiative. To further improve the immune responses, researchers generate RAG2/common gamma chain immunodeficient mice that express certain human cytokines important for maintaining human HSCs and supporting myelopoiesis. These mice will also express human MHC class I and II so that the T cells that develop from the transplanted human HSC are “educated” properly. The strategy is to replace the respective mouse genes with their human counterparts in mouse embryonic stem (ES) cells and then make transgenic mice from these ES cells.

The project is now in its third year and Manz says his group and his colleagues at Yale are now testing the first mice. So far Manz says he has some “promising” data. “We think it’s going in the right direction,” he adds. Eventually Manz says this model will be made available to researchers who want to test candidate vaccines.

Another project—also funded by a grant through the Grand Challenges in Global Health Initiative—led by Rudi Balling, scientific director of the Helmholtz Centre for infection research in Braunschweig, Germany, is trying to achieve similar goals by crossing existing mouse strains, according to Manz.

Still, even without the full and appropriate immune responses, there is a way the current humanized mouse models could already be used for HIV vaccine research, says Shultz. “You could take PBL from people who got experimental HIV vaccines and put the PBL in the mice and then challenge with HIV,” he says. “People should be doing that.”

Got stem cells?

Even if researchers succeeded in developing a humanized mouse model with a perfect immune response, researchers will still face another challenge—a limited supply of human CD34+ HSCs. “From one cord blood [sample] you can maybe transplant four to eight mice,” Manz says. As a result, experiments with humanized mice are hard to compare with each other because of the low numbers of animals and/or genetic variation between different human HSC sources. The situation is better with SCID-hu mice, in which thymus and liver tissue from one fetal donor is sufficient for 50-60 mice, according to Stoddart. “It remains to be seen yet whether those kind of numbers can be made of any of these other models,” she adds. Still, fetal tissue that can serve as a source for CD34+ HSCs is hard to come by—it requires fetal tissue from pregnancies aborted after three months.

To find a source that’s independent from donated tissues, researchers are trying to generate an unlimited number of CD34+ HSCs in vitro. Hongkui Deng, a professor of cell biology at the University of Peking, is working on another project funded by a grant through the Grand Challenges in Global Health Initiative in which he treats human embryonic stem cells with small molecules that induce differentiation (induction factors) to make them develop into CD34+ HSCs. One problem, he says, is that there is no functional HSC marker yet; CD34+ expression alone is not enough to be sure a cell is an HSC. Still, he is able to get CD34+ cells that, after transplantation into immunodeficient mice, lead to up to 6% engraftment. This means that up to 6% of the blood cells in the recipient mice are human. That’s much less than the up to 80% engraftment one gets after transplanting human cord blood cells, he says, but “that’s already very promising.”