In the Spotlight
Stem Cells and Toxicity Testing, Part III: Bioengineered Human Tissue Models
Published: February 3, 2014
Bioengineered human tissue constructs have the potential to replace animals as species- and tissue-specific models for certain types of toxicity testing, but only if they provide biologically relevant data. Three-dimensional tissue models that replicate the physiology of the in vivo human tissue/organ should provide more relevant information on the toxicity of drugs and chemicals to humans than similar testing performed using monolayer cultured cells or animal models. This article will discuss some of the current and emerging concepts involved with the development of biologically relevant human cell-based models for toxicity testing.
Cultured Human Cells
Cells grown in the laboratory are typically cultured on plastic plates where they form a layer of cells (sometimes several layers depending on the cell type) that are overlaid with a liquid media containing the nutrients needed for their maintenance and growth. This technique is called monolayer cell culture, and it allows for the rapid expansion of cells so that large numbers are available for experiments. The problem with this approach is that cells grown under these conditions are in a very foreign environment and behave differently than they do in the intact tissue/organism.
Until relatively recently, the cells/cell lines used to generate human cell-based models have been either primary or immortalized cells. Primary cells are those isolated directly from a tissue and then grown in the laboratory. Immortalized cells, on the other hand, contain gene modifications that increase their ability to proliferate and survive in culture. Both of these types of cell lines have their advantages and disadvantages. Primary cells are usually better models of the in vivo state, but many cell types do not survive long in culture, and the isolated cells can have substantial variability when isolated from different donors. Immortalized cells can provide sufficient numbers to establish reproducible in vitro models, but they have a different genetic makeup than the original tissue. Both primary and immortalized cell lines can change over time in culture, so the number of times a particular cell line can be “passaged” (i.e., grown and split into new culture plates) must be defined when a cell line is characterized.
An emerging capability is to utilize a third type of human cell, the stem cell, in cell culture models. Basic information on the types of stem cells and their use in toxicity testing formed the basis of the previous Spotlight articles, Stem Cells and Toxicity Testing, Part I and Part II.
There are also a number of quality issues to address when using any cell line for research or testing, including thorough cell line characterization and identification, as well as whether to implement the use of good laboratory practice (GLP) and good cell culture practice (GCCP) guidelines.
Bioengineered Human Tissue Models for Toxicity Testing
The first step toward the development of more physiologically relevant human cell models (for many cell types) is to culture them in a three-dimensional (3D) format that more closely mimics how the cells grow in the relevant in vivo tissue (see Figure 1). Such 3D cultures are sometimes called organotypic models. Cells grown in a 3D format typically differentiate more than the same cells grown as monolayer cultures, and are therefore closer to the state of the majority of cells in the in vivo tissue. Differentiated cells in 3D cultures have been found to be more resistant to chemical injury than similar cultures of undifferentiated cells (whether monolayer or 3D cultures), thus providing a more biologically relevant test model (Ward, Walker & Dimitrijevich, 1997).
Figure 1: Human pluripotent stem cells cultured as (A) a monolayer (2D) culture on polystyrene plasticware (looking down at the top of a layer of cells), and (B) as a multilayered 3D culture on the surface of Alvetex®Strata (looking at a cross-section of cells and the culture membrane). Scale bars: (A) 25 μm; (B) 100 μm. Photos courtesy of Reinnervate.
Three-dimensional tissues can be developed using a number of physical substrates to induce cell stratification/multilayered growth, which include growing cells on permeable membranes, hydrogels, scaffolds, or in hanging drops. Manipulation of media components (such as growth factors, hormones, nutrient molecules, and ions) and other factors (such as temperature, gradients, and oxygen levels) may also be used. Many publications and reviews are available that describe various scaffold technologies that support 3D cell growth, including methods for 3D culture in multi-well high-throughput screening (HTS) assays (e.g., Maltman & Przyborski, 2010; Yoshii, et al., 2011; Tung, et al., 2011; Chang, et al., 2011; Fennema, et al., 2013; Wu, et al., 2013; Turner, et al., 2014).
Obtaining true physiological relevance, however, can be much more complicated. Cultured cells are not typically surrounded by the same extracellular molecules, substrata, and neighboring cell types as found in the in vivo tissue. Exact replication of the microenvironment has not been a realistic goal (at least not previously), but cultured cells/tissue models can be optimized to more closely mimic the in vivo cellular phenotype by certain modifications of the culture conditions and the cellular microenvironment.
Before use as a research or testing model, an in vitro cell-based model should be thoroughly characterized for how well it replicates key physiological parameters of the same tissue in vivo; especially those features involved with the endpoint being assessed. These characterization parameters will vary by the cell/tissue type and possibly by the intended use of the in vitro model. The key parameters determined to be useful/necessary in characterizing a cell-based model can then serve as the baseline for evaluating changes to the cell’s microenvironment, as culture conditions are optimized. Rigorous characterization is often shortchanged due to cost and time; however, the data derived from poorly conceived/characterized models can provide irrelevant and even misleading results.
Many types of human cells have been cultured as 3D tissues for use as research and/or testing models. A review by Elliott and Yuan (2011) described 3D models for “liver, breast, cardiac, muscle, bone, and corneal tissues as well as malignant tissues in solid tumors,” and explained how the microenvironment established in the 3D models made them more relevant in studying disease progression and drug responses. Gibbons et al. (2013) identified pre-clinical models for testing drugs, gene therapies, and medical devices that used bioengineered human tissues to mimic “blood vessels, skeletal muscle, bone, cartilage, skin, cardiac muscle, liver, cornea, reproductive tissues, adipose, small intestine, neural tissue, and kidney.” The models reviewed by these authors were not limited to those that incorporated stem cells. However, substantial progress has been made in the development of bioengineered human tissues that incorporate and/or are derived from stem cells, including, for example, those for bone, liver, kidney, brain, cornea, musculoskeletal, bronchial, and hypertrophic scar.
A Few of the Current Limitations and Opportunities
Even with high-fidelity 3D human cell-based models, additional issues such as metabolism and toxicokinetics often need to be assessed. The in vitro assessment of metabolism and toxicokinetics are starting to be addressed with the use of devices that include 3D and/or co-culture and/or microfluidic systems (Chang, et al., 2008; Chang, et al., 2011; Li, et al., 2012; Yum, et al., 2014). Although stem cells may find an important role in in vitro metabolism and toxicokinetic models, most are based on primary cultured cells at this time.
Blood vessels that supply nutrients and oxygen to organs are needed for the same purpose in tissue-engineered constructs. Only very small 3D tissues can be maintained in culture without some type of nutrient network. Among the many reports on developing vascularized tissues, two main approaches have emerged. One involves the addition of angiogenic growth factors to stimulate microvessel endothelial cell (EC) growth, and the other involves co-culturing the ECs with another cell type that releases factors that promote microvessel growth. The leakiness or permeability of tissue engineered capillaries/blood vessels is an important indicator used to assess their functionality. In an interesting study by Zheng, et al. (2012), 3D microvascular networks of human umbilical vein ECs were grown in collagen microfluidic scaffolds to demonstrate their potential utility in developing vascularized tissues for regenerative medicine. In another study, Grainger and Putnam (2011) developed a 3D co-culture method for growing interconnected networks of pericyte-invested capillaries. The human umbilical vein ECs were less permeable/leaky when co-cultured with bone marrow-derived mesenchymal or adipose-derived stem cells than in the typical co-culture of fibroblasts and ECs. Another study by Pati et al. (2011) suggested that a soluble factor released by mesenchymal stem cells modulates EC junction integrity by increasing levels of the cell-cell junctional protein VE-cadherin. The development of functioning vascularized in vitro tissues/organs is essential for realizing their full potential in regenerative medicine and for some alternative toxicological models.
One of the features of cell culture models that could limit their use for some types of toxicological testing is their inability to regenerate/recover from chemical-induced injury to the same extent seen in vivo, thus limiting comparison of the in vitro results to animal data. Stem cells could be the key to addressing this limitation. Many (possibly all) human tissues and organs contain regions of adult stem cells called “stem cell niches,” which are thought to be the cells that replace damaged and/or senescent cells, thus maintaining the tissue/organ. Three-D models with appropriate stem cell niches, or models fully developed from stem cells, provide the interesting possibility for developing in vitro models with greater longevity and the ability to recover from toxic injury. Reproducing the correct microenvironment may not be easily achieved, but as indicated by Kuhn and Tuan (2010), “The balance between the naïve stem state and differentiation is highly dependent on the stem cell niche…. Ultimately, the fate of stem cells is dictated by their microenvironment.”
Stem Cells in 3D Human Tissue Constructs
Stem cells technologies could be important in several different aspects/approaches to the development of functional and sustainable bioengineered human tissue models. These include:
1 – The directed differentiation of induced pluripotent stem cells (iPSCs) to create 3D in vitro models has the potential to provide the large numbers of cells and the many different cell types needed for the development of highly reproducible models for testing purposes.
2 – Recent research indicates a factor produced by stem cells is important to the integrity of the microvasculature developed in bioengineered tissues.
3 – Adult stem cell niches could be the missing ingredient needed for 3D models that can regenerate and recover to mimic the in vivo tissue/organ.
4 – Adult stem cells appear to contribute differently to the extracellular matrix/tissue architecture than differentiated stromal cells (see the example of Wu et al., 2014 described in the following section). If this is found to be a general principle, at least some types of bioengineered tissues would require appropriate stem cell niches/microenvironments to sufficiently replicate the in vivo tissue.
5 – Various approaches of combining appropriate primary cells and/or differentiated iPSCs with stem cell niches, and relevant topographic cues, matrix, and growth factors need to be explored to optimize the development of biologically relevant engineered tissues. Biomarkers, ultrastructure, and physiological features will need to be identified for the characterization of each tissue type, so that success can be recognized.
Case Study: Human Cornea
Human cornea is one tissue where an understanding of stem cell niches and their microenvironments has led to progress in the development of bioengineered human tissues.
Bioengineered human corneal models have a long history of development for applications in ocular toxicity testing, ophthalmic research, ocular product research and testing, and human transplantation. Unlike most tissues, the cornea is avascular, making it a potentially easier tissue to replicate with in vitro models. However, even with this “relatively simple” tissue, efforts to develop in vitro models of the human cornea for ocular toxicity testing have been underway for over 20 years (AltTox, 2007).
The corneal epithelial stem cell niche was one of the earlier stem cell niches to be identified (Schermer, Galvin & Sun, 1986). The corneal epithelium contains a stem cell niche in the basal cells of the limbal region (Stepp & Zieske, 2005) (see Figure 2). In addition to cell division to repopulate the cells of the corneal epithelium, the corneal limbal stem cell population has been proposed as a “barrier” that prevents the conjunctival epithelial cells from migrating into the cornea (Dua & Azuara-Blanco, 2000). The limbal stem cells are the ones predominately surviving in corneal epithelial cell cultures; however, they lose their ability to proliferate in culture after several passages unless the cells are expanded on feeder cell layers or genetically modified to have an extended life span.
Figure 2. Cross-sectional diagram of the human limbus. (Reprinted from StemBook, “Limbal epithelial stem cells of the cornea” by Genevieve A. Secker and Julie T. Daniels, licensed under a Creative Commons Attribution 3.0 Unported License.)
For over a decade, a variety of ocular cell and tissue transplantation techniques for treating ocular surface diseases and injuries have been developed based on the regenerative properties of the limbal epithelial stem cell population in donor corneas. On the other hand, cell culture models derived from isolated corneal limbal cells have been observed to be slower or incapable of repairing superficial damage than what would take place in the in vivo cornea (personal observation, author). Research into various ways to prolong the stem cell-like properties of the limbal stem cells, as they were expanded in culture for use in human transplantation, have met with limited success (for example, Barbaro et al., 2009). Menzel-Severing et al. (2013) concluded “it is unlikely that [cultured limbal epithelial transplantation] will yield fully satisfactory clinical results until the factors that govern limbal stem cell maintenance and differentiation are identified.”
Stem cells also exist in the corneal stroma (Du et al., 2005), the transparent connective tissue layer beneath the corneal epithelium that comprises around 90% of the total corneal volume and is sparsely populated with differentiated, fibroblast-like cells called keratocytes. Pinnamaneni and Funderburgh (2012) reported that “a small population of cells in the mammalian stroma displays properties of mesenchymal stem cells… [and they] undergo extensive expansion in vitro without loss of the ability to adopt a keratocyte phenotype.” These stromal stem cells are located near the limbal stem cells, and it was proposed that they “may function to support potency of the epithelial stem cells.” [For stem cells, potency has been defined as the range of options available to a cell.] Wu et al. (2013) engineered corneal stroma on a specialized matrix to achieve a strong and transparent corneal tissue where the stromal stem cells differentiated into cells that expressed high levels of keratocyte gene products. The authors concluded that their findings demonstrate “the importance of topographic cues in instructing organization of the transparent connective tissue of the corneal stroma by differentiated keratocytes.” Additional studies have shown the human stromal stem cells generate a more structurally correct stromal extracellular architecture than the differentiated keratocytes when they are cultured in the presence of appropriate substrate and growth factors (Wu et al., 2014).
The many attempts at reconstructing the human cornea and corneal epithelium for toxicity testing purposes have been somewhat successful, but current models have been rejected by others based on their inability to replicate all of the features evaluated by the in vivo Draize rabbit eye test. This may not be a reasonable or necessary standard, but science is expected soon to prevail in the development of even better in vitro ocular models. The emerging capability to incorporate corneal limbal and stromal stem cell niches into a 3D tissue that incorporates appropriate microenvironments, ocular cells, and extracellular matrix could lead to in vitro models that can recover from topical chemical injury and that will be even more predictive of human ocular toxicity than the existing in vitro and animal models.
In vitro models derived from human stem cells hold great potential for the development of more biologically relevant models to evaluate the toxicity of substances to humans, compared to earlier generation efforts. Due to the concurrent and more lucrative demand for stem cell technologies in the fields of regenerative medicine and drug development, researchers are rapidly surmounting the technological and scientific obstacles.
Current developments in bioengineered corneas are beginning to take into consideration stem cells niches, microenvironments, and cell matrix and cell-cell signaling within the 3D tissues, and these efforts are just the beginning of discovery into the important roles that stem cells will play in a systems approach to toxicity testing.