Preclinical models of cancer need to raise up their game. They already perform at a high level, increasing the efficiency of drug development, individualizing patient treatment, and uncovering anticancer immune mechanisms. Going forward, however, preclinical models will also have to contend with the challenge posed by cancer immunotherapy. Essentially, cancer immunotherapy calls on mouse makers, both commercial and academic, to capture the complexities of the human immune system.
Preclinical model development has progressed from mice that lack an immune system and support the growth of implanted tumors from human cancers—that is, patient-derived xenograft (PDX) mice—to mice that incorporate humanized immune systems.
PDX models that simulate human tumor biology, allowing natural cancer progression, represent a powerful translational research tool for evaluating drug or treatment efficacy. In these models, tumor fragments from patients are directly implanted into immunodeficient mice and then passed in vivo directly from mouse to mouse, maintaining the cellular and histological structure of the original tumor.
Scientists continue to develop more sophisticated animal models, offering researchers a range and breadth of products that capture interpatient heterogeneity. Such products can be predictive of tumor drug responses and antitumor immune mechanisms.
Available mouse models differ in their ability to support the engraftment of functional human immune systems. Furthermore, the nomenclature surrounding the various strains and models is complex, as investigators are the first to acknowledge.
A real game changer in the use of mouse models, says Michael Seiler, PhD, vice president, commercial products, Taconic Bioscience, is the ability to combine key elements of a human immune system with PDX models to enable the assessment of novel immunomodulatory and other agents that affect tumor response in the context of a functioning immune system.
“It has been a profound change,” Seiler emphasizes. “With small-molecule drug testing, the standards have been common inbred mouse strains, or genetically engineered mice with gain-of-function (that is, transgenic expression) or loss-of-function (that is, knockout) mutations that expand the researcher’s toolkit.”
But immuno-oncology therapies have moved into several new classes of treatment, including monoclonal antibodies (mAbs), tumor vaccines, bispecific antibodies (bsAbs), and chimeric antigen receptor (CAR) T cells. “If we use drugs that target proteins on human immune cells, it creates an exquisite challenge,” Seiler points out. “Keytruda® (pembrolizumab) recognizes only the PD-1 receptor on human lymphocytes. It does not recognize the mouse PD-1 receptor, which limits the utility of a traditional mouse at a crucial stage in drug development.”
“This necessitates a different experimental model, one that supports human immune cell functions in a living system,” Seiler explains. “We have learned from studying stem-cell humanization in the CIEA NOG® mouse that endogenous mouse cytokines can’t drive human hematopoietic stem-cell differentiation.”
According to Seiler, Taconic’s huNOG-EXL combines the background of the CIEA NOG mouse with transgenic low-level expression of two human cytokines, GM-CSF and IL-3, both of which are known to limit myeloid lineage commitment due to cross-species limitations. “Upon engraftment with human hematopoietic stem cells, the hGM-CSF/hIL-3 transgenic-NOG host results in a human-like immune system that includes mature granulocytes, monocytes, macrophages, B cells, and T cells, extending the limits of existing engrafted human immune system models,” he elaborates.
Regarding the increasing complexity of mice required for testing immuno-oncology agents, Edgar Wood, PhD, senior research director, oncology, Charles River Discovery Services, explains that for testing standard cancer drugs, like cytotoxic or targeted therapies, the basic platform has been human tumor material grown as a xenograft in immunodeficient mice.
He cautions, however, that research into human-specific immunomodulators requires two xenografts in mice—the tumor material plus the human immune cells. “The nature of the immunodeficient mouse is also different,” says Wood. “Most tumor material will grow in simpler forms of immunodeficient mice, like T-cell-deficient nude mice or T- and B-cell-deficient SCID mice.
“Engraftment of immune cells typically requires an increased level of immunodeficiency that results from additional mutations including IL-2 receptor gamma,” he continues. “The simplest model involves engraftment of peripheral blood mononuclear cells (PBMCs) from adult donors, but T cells typically dominate the engraftment due to their capacity for proliferation.”
According to Wood, if you are hoping to modulate other cell types, such as myeloid cells or natural killer (NK) cells, PBMC engraftment may not represent the desired biology. “More advanced models involve engraftment of CD34+ human stem cells that can differentiate into key cellular players including natural killer and myeloid cells,” he points out. “The level of myeloid cells can be enhanced by using mouse strains that express human cytokines including IL-3 and GM-CSF.
“These complex humanized models have disadvantages as well, including the cost and availability of the animals, the duration of the engraftment, the onset of graft-versus-host disease, and donor-to-donor variability.”
Wood adds that for therapies that recognize the mouse counterpart of a target, traditional mouse syngeneic tumor models offer a fully intact mouse immune system and a variety of tumor types and responsiveness to immunomodulators.
The NSG mouse
James G. Keck, PhD, senior director, innovation and product development, JAX® Mice, Clinical and Research Services (JMCRS), The Jackson Laboratory, says that the lab’s “platform mouse” is the immunodeficient NSG mouse. It is used to engraft human umbilical cord stem cells for the development of a partial human immune system in the mouse.
“The stem cells grow and differentiate into human immune cells, mostly functional CD4/CD8 cells,” he explains. “We have about a 95 percent success rate in humanizing the NSG mouse. The mice are healthy and robust, and live for about a year.
“The introduction of patient-derived tumor gives you a mouse with a human immune system and a human tumor, and a model in which we can evaluate PD-1 inhibitors, CTLA-4 inhibitors, IDO1 inhibitors, and other drugs in involved in T-cell recognition and response to the tumor.”
Regarding the predictive reliability of such models, Keck asserts that “many pharma and biotechnology companies are accepting the value of the NSG platform to better understand how their drug will behave in the clinic.” He also declares that the platform allows investigators to “look at the effects of immune-cell infiltration into the tumor to determine whether a drug affects the immune response to the tumor.”
The Jackson Laboratory plans to make additional mouse models that will, upon engraftment with umbilical cord stem cells, lead to additional human immune cell populations and allow researchers to ask more specific questions about the immune response to tumors.
“We don’t know how many different human immune cells we can develop in a mouse and still make it a good model,” continues Keck. “We, for example, are introducing a new model in January, the NSG-IL15 mouse. It will allow the development of NK cells along with CD4 and CD8 T cells. This is important because of the development of more sophisticated drugs and drug combinations that target these cell populations.”
Future of immunotherapy
Federica Parisi, PhD, manager of scientific product marketing, Crown Bioscience, says her company believes that the future of immunotherapy will likely be in combination reagents. Crown produces preclinical models in which pairs or combinations of human-specific immuno-oncology agents can be tested.
The company’s HuGEMM platform provides “double knock-in” humanized drug target models (such as PD-1 inhibitor and CTLA-4 inhibitor models). They feature two humanized immune checkpoint inhibitors within an immunocompetent mouse for combination immunotherapy development.
Parisi notes that like all of Crown’s models, HuGEMM mice undergo strict validation before being launched for client studies. “This includes,” she elaborates, “validation of the human protein expression and ex vivo binding assays to verify that the chimeric human/mouse protein expressed by the models is effectively recognized by the human antibody or by human recombinant proteins.
“In some instances, depending on the model type, we also validate the activation of the relevant immune cell population upon agent binding. In all instances, we test the model response in vivo to the relevant human-specific antibody currently in the clinic.”
Making a Humanized Mouse
According to Lenny Shultz, PhD, professor, The Jackson Laboratory, there are many ways to make a humanized mouse. “The work that started with CB17-SCID mice has advanced to genomic editing of multiple humanized mouse models,” he says. “We now have models that can support engraftment of a human immune system.”
To illustrate how far these models have gone toward achieving that goal, he cites work in which he and colleagues transplanted non-obese diabetic SCID-interleukin-2 receptor gamma null (NSG) mice with human (hCD34+) hematopoietic progenitor and stem cells (HPSCs) leading to the development of human immune system–humanized (HuNSG) mice.
In a paper that was printed last March in The FASEB Journal, Shultz and colleagues described a humanized mouse model bearing human cancer cell line–derived xenograft (CDX) or patient-derived xenograft (PDX) tumors. The model uses allogeneic but human leukocyte antigen partially matched CD34+ HPSC donors and tumors.
“Tumor growth curves were similar in HuNSG compared with nonhuman immune-engrafted NSG mice,” the paper indicated. Treatment with pembrolizumab, which blocks PD-1, significantly inhibited growth in both CDX and PDX tumors in HuNSG but not in NSG mice. Inhibition of tumor growth depended on huCD8+ T cells, as demonstrated by antibody-mediated depletion. Thus, tumor-bearing HuNSG may provide an important new model for preclinical immunotherapy research, the paper noted.
Shultz adds, however, that while it is relatively easy to genomically edit a mouse, the optimal model is unclear: “You can’t put every human immune gene into a mouse or eliminate every mouse immune gene. We’ve knocked out toll-like receptor gene pathways and seen that the mice can die of their own microbial flora.”
The lab has made 40 or 50 genetic alterations in the NSG mouse model to support, for example, the development of human mast cells and natural killer cells.
“We develop our models to answer specific questions,” Shultz declares. “Right now, we are trying to [find out] why checkpoint-blockade works for some human cancers and not for others.”