With so many model systems to choose from, deciding which will work best for your research can be daunting. No one model system is perfect, and no matter which you choose, there will be tradeoffs. To help make the selection process easier for our readers, Biocompare interviewed three experts to learn more about the model systems that are available and the circumstances under which they’re most appropriate. The panel included Jarkko Huuskonen, Ph.D., Director of R&D at ATCC; Steve Festin, Ph.D., Director of Scientific and Commercial Development at Charles River; and James Hickman, Ph.D., Chief Scientist at Hesperos.
What factors should a researcher consider when selecting a model system?
Huuskonen: The key to choosing a model system is to consider the purpose of your experiment and the limitations of the different model systems. As an example, if the goal is to have a preliminary look at the toxicity of compounds, a high-throughput 2D screening using a transformed cell line might be adequate. Down the line, however, more complex models such as 3D spheroid or organoid models will provide even more details and physiological relevance. However, more complex model systems are typically more labor intensive, lower throughput, and higher cost when compared to simpler 2D culture models.
If you are working with cells, another thing to consider is the tissue origin of the cell model being used. The use of a tissue-specific model may be a better choice than a non-representative continuous cell line.
Festin: Model systems can range from biochemical assays to cellular-based systems to in vivo “live” animal models. Ideally, the best system will most closely reflect the biological system being studied. First, biochemical and cellular systems can be used to validate assumptions of mechanism and function. Once validated, in vivo research models provide a complex living system, encompassing the process, pathway, or function under study while being part of a larger “living test tube.” Further, lab animal research models may be modified using genetic technologies that enable testing of specific pathways.
Hickman: The major factor a researcher needs to take into consideration is whether the model system reflects the human response. And, when possible, it’s better for the system to have been validated with human clinical data. For an in vitro system, an interconnected multi-organ system will enable interactions between organs that are normally seen in the body. And if it is metabolically competent, you’ll be able to see the effect of not only the parent drug but its metabolites as well. Additionally, if it has the ability to create a PKPD model of the system, it’ll help determine how the results could project to human clinical questions.
What can researchers do to ensure translatability of the system?
Huuskonen: One tactic to evaluate the reliability of your model system is to test a panel of compounds with a known effect in humans on your model system. This typically gives you at least an idea whether or not your model system appears to be reflective of in vivo behavior. Another approach is to look at the biology beyond the cell model, including shear stress, nutrient circulation, and interplay between cells or compartments. These are often recapitulated in more complex microphysiological systems (MPS) that are, in principle, closer to the in vivo situation. However, it is good to recognize that the presence of these factors is not a guarantee of performance and that extensive characterization and qualification of the model system is still required.
Festin: Simply, researchers need to work with a model that most closely resembles human systems. For example, today, immunodeficient rodent laboratory animal models with reconstituted human immune systems are available that enable the study of human immune processes. These chimeric models enable researchers to study human systems, drug interactions, and cell (tumor) growth in a manageable research model system. Ideally, the platform a researcher uses needs to be as close to the human system as it can be and anticipate the impact of complex physiological systems on study targets.
Hickman: As indicated above, comparison to clinical results or clinical observations is the gold standard for evaluating model systems. A secondary method of evaluation is to compare the systems to previous literature—although this should be done with care, as not all studies are fully trustworthy.
What are some of the ways that model systems are made into disease models?
Huuskonen: There are multiple ways to generate disease models, ranging from simple to extremely complex. The following are a few examples:
(A) You can isolate primary cells from healthy and diseased donors. For example, ATCC offers human primary airway cells derived from both normal tissues and donors diagnosed with asthma, fibrosis, cystic fibrosis, or COPD.
(B) You can use gene-editing technologies such as CRISPR/Cas9. By using CRISPR/Cas9 technology to make precise changes to the genome of a target cell, ATCC is able to develop a series of isogenic cell models harboring mutants of key oncogenes, such as a lung cancer cell line with an EML4/ALK fusion gene product and melanoma cell lines with KRAS, NRAS, or MEK mutations.
(C) You can access cancer models derived from patient samples as well as clinical data and full sequence information of the particular model through the Human Cancer Model Initiative (HCMI), a joint effort among multiple institutes and labs worldwide.
Festin: Traditionally, disease models were identified through observation and subsequent breeding to isolate or enhance disease traits. More recently, modern genome editing and analysis technologies, like CRISPR, enable construction of highly complex models in both in vivo and in vitro models. Gene-modifying technologies can be used on any genomic living cell or organism across species.
Genetic modification of a target within an organism may result in unexpected effects. For example, when a gene that is required for development is knocked out, the mutant organism may die as an embryo—resulting in a model unable to reproduce. It is very important that a genome-edited model undergo significant validation of genetics, health, and phenotype before being considered a model of biological function or disease.
Hickman: The best way to utilize organ-on-a-chip and body-on-a-chip systems for disease research is to use them as phenotypic models in which deficits are due to the disease or condition, so that treating it with a potential drug will bring it closer to healthy function. This method allows researchers to establish drug efficacy without requiring that the exact target be known. These systems can be created using induced pluripotent stem cells from patients that are differentiated into the appropriate organ mimics. This is especially useful when researching rare diseases for which animal models do not exist or are not possible.
What are some alternatives to animal models?
Huuskonen: There are multiple ways to construct cell models that can be used as alternatives to animal models. One consideration is the cell type used and another is the complexity of the model.
Some examples of cell types that can be used include transformed cell lines, iPSC-derived cells, primary cells, immortalized primary cells (such as using hTERT technology), and gene-edited cells. In a general sense, transformed cell lines are the easiest and lowest cost to use, but they also have the lowest resemblance to physiology. Primary cells provide the closest resemblance to physiology but are more costly, are limited in number and life span, and require special media and more technical expertise than continuous cell lines. A model that bridges the gap between continuous cell lines and primary cells is hTERT-immortalized primary cells—cells that offer the unlimited supply and life span of continuous cell lines while their genetic makeup and performance is equivalent to the primary cells.
The complexity of the model ranges from simple 2D cultures to complex co-cultures, spheroids, organoids, and organ-on-chip and body-on-chip models. In a general sense, when the complexity of the model increases, so does the physiological resemblance to the in vivo situation. On the other hand, the more complex systems are typically lower throughput and more costly. It is a balancing act between the speed, cost, and purpose of the model that dictate its selection and use.
Festin: In all cases, the “best” model should be selected. In some cases, advanced 3D culture, organ-on-a-chip, and in silico technologies are the best models. In other cases, the best choice is a well-established, well-validated traditional biochemical or animal model. There is no single model, in vivo, in vitro, or in silico, that answers every question. It comes down to a thorough analysis of the purpose of the study and the questions being asked. Using Occam’s razor as a guideline, the system should only be as complex or contain as many entities as necessary to answer the question. When applying this in biological systems, the simple assumption should be verified in a minimum two representative systems to ensure the validity of the model and the meaning of the resulting outcomes.
Hickman: Cell-based models can be good alternatives to animal models, but they come with their own set of pros and cons. For example, with organoids, anatomical differences can be observed, especially in toxicity or disease modeling of embryonic or childhood development. But the difficulty with organoid-based systems is that it is difficult to assay readouts of the systems through any means besides biomarkers or image analysis. With body-on-a-chip systems, you can introduce function—such as muscle contraction, electrical activity, and membrane resistance—but anatomy is generally not reproduced in these systems.