Alternative Models in Biomedical Research: In Silico, In Vitro, Ex Vivo, and Nontraditional In Vivo Approaches


Introduction

In 2007, the U.S. National Research Council (NRC) formally introduced a new paradigm in safety assessment and toxicity testing on behalf of a joint consortium of U.S. federal government agencies ( ; ). This “Tox 21” initiative highlighted both a vision and a strategy for assessing the safety of underevaluated agents. For example, at present, the U.S. Environmental Protection Agency (EPA) Toxic Substances Control Act (TSCA) Inventory ( ) lists over 40,000 chemicals, most of which have little or no safety data. Risk assessment of this large and diverse universe of chemicals, and their myriad complex mixtures in the environment, is simply not feasible using in vivo animal toxicity studies, which are expensive and permit a limited throughput.

The “Tox 21” vision is to increase efficiency in safety assessment by shifting away from reliance on traditional “apical” endpoints like in vivo animal studies and toward risk predictions using data derived from alternative in silico, in vitro, and ex vivo models, pathway-based biomarkers, and “big data” analytics. The strategy was to develop and validate these innovative new models, to employ computational technologies and bioinformatics to enhance the evaluation of these models, and to translate these new types of scientific information into practical readouts that can guide risk assessment decisions even if data from traditional in vivo models are limited. In a broad sense, the “Tox 21” goal was to modernize toxicity testing and, in the process, reinvent toxicology as a more prospective and mechanism-based science.

Product discovery and development requires extensive testing to define potential toxic responses in organisms that might be exposed to a given chemical, drug, food additive, gene or cell therapy, or medical device. Such testing encompasses many different systems including computer-generated predictive algorithms (in silico) for screening and prioritization, isolated tissue elements such as cultured cells and tissue slices (in vitro), and living organisms (in vivo). In general, toxicologic pathologists spend most of their time evaluating specimens acquired from in vivo studies. However, the growing interest in reducing animal use and finding more human-like test systems is accelerating the invention and validation of alternative assay systems in toxicity assessment. Going forward, the newer in silico and in vitro model systems will have important implications for the practice of toxicologic pathology.

The goals of this chapter are to summarize current alternative models for evaluating toxicity and to provide a perspective of future directions in alternative model development of particular relevance to toxicologic pathology. Alternative platforms considered here include simple and advanced in vitro models, selected less common in vivo models, and basic types of in silico models. These models have various advantages and disadvantages, depending on the questions being asked. Models of all levels of complexity have been informative for toxicologists and toxicologic pathologists.

Nontraditional Models in Toxicity Research

Four main categories of nontraditional safety assessment have gained importance for hazard identification and characterization as well as risk assessment in recent years. These strategies are in vitro modeling, ex vivo modeling, in vivo modeling in alternative mammalian and nonmammalian species, and most recently in silico systems.

Overview of In Vitro and Ex Vivo Models

In vitro models consist of cells grown in isolation, while ex vivo models consist of structurally intact portions (usually small masses or thin slices) removed from a parent organ or tissue. In vitro and ex vivo models have been most successful at answering very specific questions. For example, are key proteins expressed? Are functional activities that are important for the question at hand maintained in the model? Does the model respond as predicted to known toxicants? By design, in vitro and ex vivo models address questions that relate to responses in a particular cell population and tissue without any attempt to consider other factors that contribute to toxicity in whole organisms ( ADME Principles in Small Molecule Drug Discovery and Development , Vol 1, Chap 3 , Biotherapeutic ADME and PK/PD Principles , Vol 1, Chap 4 ; Principles of Pharmacodynamics and Toxicodynamics , Vol 1, Chap 5 ).

Perhaps the simplest in vitro models use prokaryotic organisms. For example, bacterial models, such as Salmonella typhimurium and Escherichia coli, are staples in genetic toxicology as part of the Ames assay for mutagenesis ( ). Bacterial strains with particular deficiencies in DNA repair have helped to elucidate mechanisms of mutagenesis, and genetic modifications (introduced via plasmids) have been useful in further sensitizing some strains ( ).

Isolated eukaryotic cells in culture are another simple in vitro model for toxicity assessment. The advantages of such models include the ability to conduct experiments using cells from multiple species, including humans, at relatively low cost and with tight control of experimental conditions. Two-dimensional (2D) cell culture models allow for complete isolation of the cell type of interest. Importantly, in vitro models can be more readily standardized than in vivo studies and therefore may be implemented with greater consistency across laboratories. Furthermore, these high-throughput models permit incredible rates of product testing due to increases in the detection sensitivity for innovative modern endpoints and the expansion of traditional 96-multiwell plates to accommodate 384 or 1536 wells per plate ( ; ). These recent advances in simple in vitro eukaryotic models are ideal for screening large libraries of chemicals or drug candidates to get an early estimation regarding efficacy or possible safety liabilities ( ). One key criticism of 2D eukaryotic cell models is their inability to predict in vivo toxicity. This criticism often arises from poorly formulated questions (such as trying to use in vitro models to assess cellular toxicity in the absence of in vivo factors such as absorption, distribution, and metabolism in other organs) or from trying to ask questions that are too complex for a simple cell type–specific model to answer.

To improve physiological relevance and answer more complicated biological questions, more complex in vitro and ex vivo systems have been developed ( ; ). These systems include cell combinations (e.g., cells of interest cocultured in 2D or three dimensions [3D] with their support cells); tissue slices; miniorgans (e.g., organoids, embryoid bodies [EBs], micromasses); and microphysiological systems (MPS) (i.e., “organs on a chip”). These more complex in vitro and ex vivo eukaryotic models have the potential to address more advanced questions because of their multiple interacting cell types and their more intact cytoarchitectural structures and microphysiological processes. Multiorgan interactions can be simulated as well ( ). Such 2D and especially 3D models can be assessed readily using conventional morphological techniques available to toxicologic pathologists, including light and electron microscopy as well as numerous molecular pathology techniques ( Special Techniques in Toxicologic Pathology , Vol 1, Chap 11 ).

Overview of In Vivo Models in Alternative Mammalian and Nonmammalian Species

In vivo testing for product discovery and development traditionally relies on a limited number of mammalian species, but alternative in vivo models may offer significant advantages in certain applications. With rodents as an example, these alternatives may include less commonly used species (e.g., guinea pigs and hamsters; Animal Models in Toxicologic Research: Rodents , Vol 1, Chap 17 ) or bioengineered disease models ( Genetically Engineered Animal Models in Toxicologic Research , Vol 1, Chap 23 ). In particular, genetically modified models (rodents and more recently nonrodents) have advanced the disciplines of toxicology and toxicologic pathology significantly in the last 3 decades ( ; ). Alternative in vivo models of disease have been instrumental in understanding the genetic roots of disease development ( ; ; ). Likewise, animals with genetically altered metabolic capacity have been used to identify, confirm, and characterize mechanisms of toxicity as well as to better understand clearance and distribution of novel chemical entities ( ; ; ).

The value of nonmammalian in vivo models to many scientific disciplines, including toxicology, is undisputable ( Models of Toxicity: Non - mammalian , Vol 1, Chap 22 ). Key species used to gain deeper insights into genetic influences in biology and toxicology include such model organisms as nematodes ( Caenorhabditis elegans ), the fruit fly ( Drosophila melanogaster ), many fish species, and birds. For instance, zebrafish ( Danio rerio ) are a prominent non-mammalian in vivo model in toxicology ( ), valued since they share many conserved physiological processes with more traditional vertebrate models of toxicity. These alternative species have been shown to be excellent models for studying the link between genetics, toxicant-induced mechanisms of disease, and variations in sensitivity or resistance to toxicity due to their short life spans, small sizes, minimal husbandry costs, fecundity, and ease of cultivation. A primary advantage of such alternative in vivo models is that they may be maintained at very high stocking densities in petri dishes (for nematodes), multiwell plates (for fish larvae), or tanks (for adult fish), which permits much larger group sizes than can be managed for mammalian vertebrates.

Overview of In Silico Modeling

In silico methods are progressing to the point that commercially available computer software now provides substantial contributions to toxicity and safety assessments ( ). Quantitative structure–activity relationship (QSAR) algorithms have become a standard part of genetic toxicity screening as well as early toxicity predictions ( ). Extensive and more complex toxicity databases utilizing advanced computational approaches such as artificial intelligence (AI) and machine learning are capable of identifying patterns in large datasets to build useful predictive models ( ). Bioinformatics is a growing aid to toxicity testing, making major contributions to many “omics” technologies ( Toxicogenomics: A Practical Primer , Vol 1, Chap 15 ). In many institutions, in silico models of toxicity are superintended by scientists with computer science or information technology backgrounds. That being said, the modern practice of toxicology requires that toxicologic pathologists and toxicologists have at least a basic comprehension of in silico strategies for evaluating toxicity.

Taken together, these trends affirm the growing importance of alternative models of toxicity as aids in hazard identification and characterization as well as safety assessment. Challenges in creating, and effectively using, more common alternative in vitro and in vivo models need to be considered in their application. Nonetheless, many standard toxicologic pathology techniques may be effectively used in their evaluation.

In Vitro and Ex Vivo Models

In vitro and ex vivo models need to be characterized and validated carefully to confirm their appropriateness for the question being asked ( ; ). By intent, in vitro and ex vivo models focus on evaluating toxic responses in cells and tissues after they have been isolated from systemic factors (e.g., absorption, circulation, and excretion). Recent innovations have restored some of the absent systemic parameters to in vitro systems, thereby allowing them to better approximate the conditions found in cells and tissues maintained in their natural in vivo setting. The incomplete restoration of normal structural (for in vitro) and physiological (for in vitro and ex vivo) attributes means that even the best in vitro and ex vivo models are not fully equivalent to tissues in vivo. In particular, the absence of normal tissue organization may prohibit expansion of some cell populations in vitro, and may preclude analysis of processes in vitro that require a particular tissue architecture to permit effective cell interaction and function.

Cell Cultures

For decades, in vitro 2D cell cultures have been an integral part of basic biological research for the study of cellular function and metabolism, mechanisms of disease initiation and progression, and investigations of efficacy and toxicity for environmental chemicals and potential pharmaceutical candidates ( ; ; ). Categorically, cell cultures may be characterized as continuous cell lines or primary cell cultures. Under the right conditions, such cultures may be maintained for several weeks or a few months. Both categories are widely utilized in academic, government, and industrial research laboratories, and data from both categories have proven valuable in various regulatory settings. Each category has its advantages and shortcomings.

Continuous (immortalized) cell lines have the ability to proliferate indefinitely due to either a spontaneous and random gene mutation or by deliberate genetic alteration. Continuous cell lines have several advantages when cultured such as their indefinite proliferation, low propagation and maintenance costs, purity of cell population (vital for experimental reproducibility), and well-established culture conditions and protocols ( ; ). Therefore, continuous cell lines are still widely used in screening assays in drug discovery. However, such cell lines are not without problems. The inherent defects leading to immortalization mean that they do not typically resemble the native functions and metabolic pathways of the tissue from which they are derived nor are their responses to external modulations typical of their original cells. In addition, passages of continuous cell lines result, over time, in genetic drift from a growing stable of additional gene mutation that further alters their phenotype in culture. Perhaps the most problematic attribute is the wide spread contamination of continuous cell lines by other cell types (e.g., HeLa [cervical carcinoma] cells), which confounds interpretation of results from both current studies and prior scientific literature ( ; ; ; ; ).

In contrast, primary cell cultures are derived directly from freshly excised tissue or biopsy specimens harvested from humans and animals. Primary cells have a wild-type genotype (i.e., without major genetic mutations) and phenotype, so they recapitulate the attributes of their cell of origin. According to Good Cell Culture Practice, primary cell cultures are defined as “ the initial in vitro culture of harvested cells and tissue taken directly from animals and humans ” ( ). In reality, since appropriately cultured primary cells retain their genetic integrity, morphology, and native cellular functions even after limited subculture passages, certain commercially available “primary” cultures are in fact low-passage number subcultures of the initial tissue isolation. The high structural and functional fidelity of primary cell cultures enables the study of normal physiological and biochemical processes in essentially native cells as well as the metabolic fate, chemical interaction potential, and effects (efficacy and toxicity) of environmental chemicals and drug candidates on such processes ( ; ). In addition, the ability to culture primary cells from equivalent organs of animals and humans allows for direct comparison of chemical potency and toxicity across species, which aids substantially in translation of animal-derived data to guide human benefit/risk assessment. Primary cell cultures are not without their own problems. They typically have limited life spans in culture, may be relatively hard to establish and maintain in culture, and generally are more expensive than immortalized cell lines. Because of the limited yield and finite life span of freshly derived cells, primary cultures typically are obtained from multiple donors, so they have inherently divergent—and not necessarily modest—differences in function and genetic makeup based on interindividual variations among donors.

At present, primary cells used in toxicological applications are isolated from most major organs. Common sites include skin, intestine, liver, lung, kidney, eye, and brain ( ; ; ; ; ; ; ; ; ). Tissue samples for primary cell culture may be procured from different sources, including post mortem donors, rejected transplants, surgical biopsies, and other voluntary donations such as blood, bone marrow, and skin biopsies. Mechanical and/or enzymatic processes may be necessary to dissociate the cells of interest from connective tissues. The resulting cell suspensions are typically a heterogeneous mixture of several cell types which may need to be further purified before the cell culture can be established. This additional purification can be accomplished by physical (microdissection, gradient centrifugation); immunological (antibody-specific capturing); and/or biochemical (specific media conditions to promote cell type–specific growth and proliferation) methods ( ). Purified cells can be used as a suspension (for anchorage-independent cells such as lymphocytes) or as an adherent culture (for anchorage-dependent parenchymal cells from solid organs). The composition of the culture medium is cell type-dependent and usually contains variable concentrations of many nutrients including glucose, essential amino acids, one or more growth factors, and sometimes antibiotics; the medium composition is optimized for proper growth and differentiation of the particular cell type. In some cases, extracellular matrix (ECM) protein–coated culture plates also are necessary to promote adhesion of anchorage-dependent primary cells. For example, sandwich coating of matrices is used to promote bile canaliculi formation when liver cells are being cultured ( ). Such coatings may be used to impart a more 3D, quasi-organized structure to the culture that provides an environment that more closely resembles the milieu the native cells encounter within intact tissue.

Because freshly isolated cells have a finite lifespan in culture, consistency in primary cell cultures requires banking of primary cells for archiving, transportation, and future usage ( ; ). Cryopreservation is a process to preserve living cells using ultralow temperatures. Cells are suspended in solutions containing dimethyl sulfoxide or glycerol and fetal bovine serum as cryoprotectants and then slowly frozen in a programmable controlled-rate freezer to minimize the formation of ice crystals (which can disrupt cell and organelle membranes) within the cells. The frozen cells can be stored in liquid nitrogen (<−150°C) for long-term preservation and transport. Cryopreserved cells can be recovered by thawing in a 37°C water bath for a few minutes.

Before their use for toxicological applications, primary cell cultures need to be characterized and qualified. Characterization usually includes the assessment of cell morphology and polarity, expression of relevant marker proteins, and demonstration of relevant functional capabilities. After satisfactory characterization, the cultures are then qualified using known toxicants (positive controls) and nontoxicants (negative controls) to ensure the recapitulation of specific cellular toxicity responses to known sets of test agents. One underappreciated aspect of qualification is to ensure that any effects seen in the toxicant-treated cells result from exposure to the test article and not to poor medium quality (due to inappropriate formulation and/or bacterial/fungal contamination). These steps are necessary to provide context to the toxicity data obtained from these cultures so that appropriate translation can be made to in vivo outcomes in the same species (e.g., in vitro rat hepatocyte cultures with in vivo liver toxicity in rats) or to similar cell populations across species (e.g., in vitro rat hepatocyte cultures to in vitro human hepatocyte cultures). An additional manipulation to assist animal-to-human translation when screening molecules for toxicity in vitro is to add the S9 (microsomal) fraction from homogenized human liver to cell cultures to add human xenobiotic-metabolizing activity (harbored predominantly in microsomes) to the system. This adaptation has been used successfully in Tox21 toxicity testing to show that metabolic activation of genotoxic agents increases their capacity to induce mutations in the DT40 (chicken cell) DNA damage response assay ( ).

In general, characterization and qualification consists of biochemical and functional (i.e., quantitative) endpoints. The parameters typically are preferred since they can be measured using automated, high-throughput instrumentation. Morphology of cultured cells following toxicant exposure may be undertaken by toxicologic pathologists using conventional histopathologic techniques as an additional means for characterization and qualification ( Figure 24.1 ). Endpoints of interest subject to microscopic evaluation include culture “anatomy”—cells undergo a reproducible structural evolution as they develop—and biomarker expression ( ). These features may be assessed in live cultures throughout their life span using fluorescent or phase-contrast microscopy. In the authors' experience, toxicologic pathologists participate in evaluation of cell cultures only on an occasional basis, and chiefly during basic discovery and investigational toxicology studies.

Figure 24.1, Neurotoxicity and its prevention in primary cerebrocortical cell cultures of near-term embryonic mice, as evaluated by fluorescent microscopy. The neurons were treated with pentylenetetrazol (PTZ, in mM), a stimulant that can induce seizures at high doses, or PTZ plus otophylloside N (OtoN, an herbal extract used to treat epilepsy, in μg/mL) for 24 h after 7 days in culture. Cells were immunostained with neuronal antiβ-tubulin marker (green) and nuclear DAPI (blue).

Tissue Slices

Tissue slices have considerable attraction as ex vivo models of toxicity. Ex vivo slices consist of thin tissue slabs that have been removed from an organism for evaluation in an external environment, thereby retaining near-natural tissue architecture for cells within the piece. The key advantage of this ex vivo assay is that 3D tissue architecture is well maintained for most cells in the specimen except for those nearest the cut margins. Thus, appropriate cell–cell and cell–matrix interactions are retained, which permit such cultures to function in a more physiologically relevant setting relative to 2D monolayers or suspensions of isolated primary cells obtained from the same location. The thinness of tissue slices permits nutrients, oxygen, signaling molecules (e.g., cytokines and growth factors), and waste product levels to approximate those found in the intact organ, though over time the quantities of these constituents diverges from the norm in the deepest parts of the slices.

Tissue slices can be produced for many organs utilizing well-controlled cutting to generate thin tissue slices of reproducible thickness. The use of tissue slices (of inconsistent thickness) was first described by Otto Warburg in 1923 to measure cell metabolism in tumor tissues, and later found success in the study of amino acid metabolism by Hans Krebs in 1933 ( ). Widespread use of tissue slices was enabled in the 1980s with the development of instrumentation to allow more precise cutting of thin slices in parallel to dynamic culture methodologies to prolong the ex vivo viability of tissue slices. The use of precision-cut tissue slices (PCTS), which yields more consistent specimen characteristics, has been demonstrated in various organ types including liver, kidney, intestine, lung, heart, and brain ( ; ; ; ; ). This method is widely applied to study disease modeling, drug metabolism, pharmacologic efficacy, and toxicity ( ; ; ; ).

In order to consistently obtain high-quality PCTS, tissue cores are prepared by using a sharpened metal cylinder to remove samples from isolated tissues/organs ( Figure 24.2 ). The tissue cores are then mounted in a cylindrical tissue holder which is submerged in cold, oxygenated slicing buffer. Slices are produced by moving the tissue cylinder across a blade at the bottom of the cylinder, while a stream of cold buffer helps sweep away the slices for collection. For tissues that are difficult to cut after isolation (such as intestine and lung), prefilling the hollow spaces within the tissue core with agarose is necessary before the tissue can be cut to prevent distortion of the tissue architecture and damage to the cells that line the spaces. Freshly prepared slices are then either placed onto a titanium grid inside 25 mL glass scintillation vials containing culture media or directly introduced into tissue culture dishes; the choice of which approach to use depends on the applications. The typical thickness of slices varies with the tissue type, but normally it is from 100 (liver) to 500 (lung) μm. The viability of PCTS cultures is typically assessed by such methodologies as measuring the rate of protein synthesis, leakage of intracellular enzymes, and histopathological evaluations.

Figure 24.2, Representative diagram for producing precision-cut tissue slices with the Krumdieck/Alabama Research and Development Tissue Slicer.

The utility of tissue slices in toxicological research has been recently reviewed in detail. Representative reviews include liver, intestine, lung, heart, brain, and kidney ( ; ; ; ). Recent trends include the incorporation of PCTS into dynamic flow devices to prolong the culture life time ( ), the study of immunotoxicity in lymphoid organ PCTS ( ), the addition of physiological electromechanical stimulation for long-term functional preservation of myocardium slices ( ), and virus-assisted gene transfer studies ( ). The preservation of all cell types in their native positions and with their usual interactions makes PCTS a great model for investigating toxicities that may involve multiple cell types. Importantly, the architecture in PCTS resembles that of the intact tissue in vivo, which makes this model well suited for histopathologic evaluation; indeed, the desire to perform a detailed microscopic assessment of a specimen that retains normal cell and tissue organization is a feature that often drives the choice to employ tissue slices rather than cell cultures as the investigational tool ( Figure 24.3 ). The sharp margins of PCTS preserve the function of most metabolizing enzymes and transporters, which sometimes is critical in the manifestation of toxicant-induced effects.

Figure 24.3, Cell heterogeneity in precision-cut human liver slices as assessed by transmission electron microscopy is preserved to a degree that matches that which occurs in native liver tissue.

However, the use of PCTS does have some disadvantages. First of all, the slices are typically short-lived, with viability in culture ranging from a few hours to a few days depending on the tissue type and medium. Second, because this model relies on fresh tissue isolation, both the throughput and reproducibility are limited by the availability and quality of fresh tissues—especially for human specimens—which typically come from surgical procedures. Finally, although the slices retain the native architecture of the original tissue, the ex vivo culture conditions are not representative of the tissue environment and regular molecular turnover (e.g., replenishment of blood-borne nutrients and oxygen, circulation of signaling hormones and cytokines, and removal of metabolic by-products) that occurs in vivo. By combining PCTS with microfluidic techniques (see Section 3.5 ), it is reasonable to anticipate that microphysiological systems (“organs on a chip”) might be generated that allow for longer-term cultures under more physiological culture conditions.

Bioprinted Microtissues

Bioprinting is a technology that is driving significant innovation in several areas including tissue engineering, regenerative medicine, tissue transplantation, and the development of complex in vitro models ( ; ; ; ). Bioprinting tissues involves significant complexities including the selection of cell types, growth media (with differentiation factors), minimizing damage to printed cells, and the composition of extracellular scaffolds that support a defined (2D or 3D) architecture ( ). The proper selection and control of these factors are key to successful tissue/model construction. Bioprinting has been used to produce multiple complex in vitro models from patterned 2D-printed cultures to organoids to microphysiological systems with fluidics ( ; ; ; ); all these have all been used for assessing toxicity. The versatility of bioprinting is further demonstrated by the wide array of microtissues that have been generated including cardiac and skeletal muscle, lung (alveolar–capillary interface), gastrointestinal villi, liver, and kidney (proximal tubules) ( ; ; ; ; ; ; ).

Three main strategies—biomimicry, self-assembly, and microtissue approaches—have been used for creating bioprinted 3D in vitro models. Each process involves preprocessing, processing, and postprocessing stages ( ). Preprocessing encompasses the proper design and selection of appropriate materials and reagents. Processing represents the printing and assembly of microtissue models. Postprocessing generally incorporates an incubation period that allows cell attachment, proliferation, interaction, and functioning prior to the use of bioprinted tissue in testing ( ). In addition to proper printing and assembly, other factors such as signaling molecules, structural elements, and appropriate environmental factors (e.g., temperature, pressure, sheer, and electrical forces) may need to be optimized during the processing and postprocessing stages ( ). All these factors may be addressed in the choice of bioink components (where a bioink is a complex brew of cells and/or organic materials that provide a suitable ECM environment to support the adhesion, proliferation, and differentiation of living cells).

Biomimicry

Biomimicry represents perhaps the most straightforward approach to bioprinting. This strategy assumes that proper form results in appropriate function ( ). Essentially, cells and extracellular components to support them are printed in a 3D pattern that mimics the microanatomy of a tissue or organ. However, accurate recapitulation of biologic tissue is extremely difficult. Several challenges have been noted with the biomimicry approach including incomplete and/or variable attachment, proliferation, migration, and maintenance of proper cell types and proportions ( ). The selection of scaffolding material is crucial in this process in order to best mimic the native structural and mechanical requirements for the target tissue being represented. The scaffold choice also significantly influences signaling through cellular interactions with the ECM component ( ). Common options for scaffolding materials include polymers of natural materials such as decellularized ECM, peptides, and DNA strands; synthetic materials such as polyethylene glycol and polyurethane; or combinations of natural and synthetic materials ( ; ).

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here