Organoids and Mini-Organs

By Jamie A. Davies

Organs and Mini-Organs combines contributions from leading practitioners who work under the editorial control of an acclaimed researcher who also served for eight years as Editor-in-Chief of the journal Organogenesis, the first journal on this topic. The book begins with an introduction, but then delves into chapters that present advice on how to make organoids for many systems. In addition, case studies that illustrate the uses of organioids are presented, along with discussions on future directions and specific problems that need to be solved.

• Collects the best protocols of organoid cultures from diverse tissues
• Covers a wide range of organs
• Includes troubleshooting cases for common, but specific problems for each culture conditions
• Provides an entire section on the application of organoids

• Language: English
• Publisher: Academic Press
• Release date: Mar 9, 2018
• ISBN: 9780128126370

Book preview

animals.

Part I

Introduction

Outline

Chapter 1 Organoids and mini-organs

Chapter 1

Organoids and mini-organs

Introduction, history, and potential

Jamie A. Davies,    University of Edinburgh, Edinburgh, United Kingdom

Abstract

Organoids are three-dimensional assemblies that contain multiple cell types, arranged similarly to the cells in a specific tissue, at least at the micro-scale; mini-organs add to this micro-realism a realistic macro-scale anatomy as well. Researchers have been producing organoids for at least 60 years, initially to explore basic mechanisms of development, but more recently as tools for medical research. Organoids made from human cells are particularly valuable for preclinical studies, because they avoid the need to extrapolate results from one species to another. This chapter outlines the history of organoid research, describes briefly how typical organoids are produced, and goes on to describe their use in pharmacology, toxicology, oncology, and microbiology. It includes an indication of how organoids can be turned into mini-organs, and ends with some cautionary notes on the limitations of current techniques.

Keywords

Organoids; mini-organs; research

Chapter Outline

Introduction 3

Origins of Organoids 5

Mechanisms of Organoid Formation 7

The Path to Human Organoids 9

The Construction of Organoids 11

Organoids as Tools for Clinical Research 12

From Organoids to Mini-Organs 16

Limitations of Organoids 17

Final Remarks 18

Acknowledgments 18

References 18

Further Reading 23

Introduction

It is important to begin this book with a definition of terms, because the word organoid has been coined several times in the history of biomedicine, and has been used to convey at least three distinct meanings. In the 20th century, organoid was sometimes used as a synonym for organelle (e.g., Duryee and Doherty, 1954), a sub-cellular structure: this use is now obsolete. In oncology, organoid is sometimes used as an adjective to imply a tumor with a complex, tissue-like structure, for example a gland-like carcinoma or a teratoma: this use remains current (e.g., Nesland et al., 1985; Heller et al., 1991), if somewhat obscure. Neither of these meanings is relevant to this book. Here, organoid is being used in its most common modern sense, to refer to a three-dimensional assembly that contains cells of more than one type, arranged with realistic histology, at least at the micro-scale. An organoid might be made from cells of humans or other animals, and these might be differentiated cells, stem cells, or a mixture of the two.

Interest in organoids has increased significantly in the 21st century (Fig. 1.1), fueled on the one hand by rapid developments in stem cell derivation to provide human progenitor cells, and on the other by a strengthening desire to refine, reduce or replace the use of animals in research (reviewed by Davies, 2012). Organoids are being produced for basic research into development and neoplasia, for industrial and medical applications such as toxicology, and—ultimately—for transplantation. At the end of 2013, The Scientist named organoids as one of its advances of the year (Grens, 2013), not because the technology was new, but because it had suddenly become much more pervasive and visible thanks to some high-profile research papers. About a year and a half later, and for much the same reason, Nature published an overview of the potential of the technology (Willyard, 2015). Organoids have now been developed to represent many different parts of the body, for many different reasons, with applications expected to grow still further over the next few years.

Figure 1.1 The rising use of organoids in biological research. The graph shows the number of papers returned from a Pubmed search for Organoid, and screened manually for organoid being used in the sense in which it is used in this book.

Being published in a period of such intense research, this book has two purposes. The first is to help newcomers to the field to set up working organoid systems, whether by using the exact techniques of other researchers, or by using these techniques to inspire the creation of novel organoid systems. The second purpose is to help researchers with working organoid systems find inspiration and advice to help them use their organoids as tools to solve a range of biological and medical problems.

Origins of Organoids

Organoids have been an important area of research for much longer than most recent reviews imply, particularly the annoyingly large number of reviews that appear to have been written for the sole purpose of portraying their authors as founders of a field. The first mammalian organoids were in fact produced over 60 years ago, and organoids have contributed steadily and significantly to developmental cell biology ever since.

Techniques for constructing organoids have their origins in basic research about the nature of biological organization. A good starting point for this brief history would be the work of H.V. Wilson (1910), who showed that, if a sponge is dissociated into its constituent cells, and these cells are reaggregated randomly, they reorganize to make a realistic and viable new sponge (Wilson, 1910). Though not aimed at making organoids in the modern sense, this experiment was an important demonstration that cells of an adult organism can contain sufficient information to specify a multicellular structure, without any need for outside instructions, and without the need for cells to start from some specific anatomical arrangement contingent on their embryological history. This point is critical to organoid production. In the 1950s, several laboratories began to use the same basic method—disaggregation followed by reaggregation—to determine whether the cells of more complex animals such as vertebrates also had the ability to self-organize (this term was already in use) or whether, for these complex animals, the spatial relationships contingent on past developmental history were critical. Early examples of the disaggregation−reaggregation method of organoid construction being applied to higher vertebrates were provided by Moscona (1952) and by Moscona and Moscona (1952), who made suspensions of chick mesonephric kidney cells. The source tissue included both tubular epithelial cells and mesenchymal cells. On reaggregation and incubation, epithelial cells made small clusters that went on to make tubules surrounded by mesenchyme-derived stroma. This arrangement was reminiscent of the small-scale anatomy of normal mesonephroids, although the overall gross-scale organization of the organ was absent: the structure met the modern definition of an organoid. These experiments demonstrated that at least some of the cells of embryonic chicks, like the cells of sponges, carried sufficient information to organize themselves realistically, even when their original spatial relationships had been erased.

The observation that different types of cells in the suspension (e.g., epithelial, mesenchymal) would separate in the reaggregate raised questions about how cells choose their neighbors. It was already known from culture experiments that similar cells tend to unite even if they come from tissues of different species. In the 1940s, Harris (1943), Medawar (1948), and Grobstein and Younger (1949) all tried coculturing fragments of tissues from different species to explore the mechanisms of transplant rejection (this was before the role of the immune system in the process was well understood). They each showed that fragments of the same tissue from different species could join and behave as one: indeed, in the case of heart muscle, heart-beat synchronized across the interspecies boundary. Moscona (1956) built on these studies of interspecific combinations to ask whether the association of cells was controlled more by their being of the same histological type, or by their coming from the animal. He mixed suspended mouse and chick embryonic liver cells, and showed that they cooperated in making an organoid, epithelia associating with epithelia, and stroma with stroma, regardless of the species of origin. He also mixed different tissues, for example chick kidney and mouse cartilage, and observed that the cells sorted out reasonably well, and each resulting organoid was formed only of cells exhibiting nuclear markers of its own species. The species-specific nuclear morphologies ruled out any possible mechanism of cells trans-differentiating according to their surroundings, and supported instead the idea of cells choosing their neighbors. The author concluded that type specificity was stronger than species specificity in arranging cells, and that cells were already determined to make a particular tissue. The paper also showed an early application of organoids to problems of pathology, when its author mixed mouse melanoma cells with normal chick cells, and saw some evidence of invasive growth (infiltration in his words) by the sarcoma cells.

Part of the drive behind these early organoid experiments, acknowledged and discussed by Weiss and Taylor (1960), was to provide a counterbalance to the prevailing view that most embryogenesis was driven by inductive signaling, a mechanism that had obsessed many embryologists since its discovery in the 1920s (Spemann and Mangold, 1924). The ability of mixtures of cells, isolated from their normal anatomical relationships with the rest of the embryo, to self-organize was taken by these authors as an indication that much epigenetic information was held by the cells themselves, and did not rely on inductive instructions from elsewhere. This did not, though, rule out signaling taking place between different cell types within the self-organized aggregate and, in the 1950s, Grobstein’s laboratory used the dissociation−reaggregation technique to explore these signals. It was already known, from the experiments of Gruenwald (1937, 1942), that kidney development relies on inductive signaling between the ureteric bud (the progenitor of the urine collecting duct system) and the surrounding metanephrogenic mesenchyme (Metanephrogenic mesenchyme is Grobstein’s original term that captures the developmental potential of the tissue.). The mesenchyme induces the ureteric bud to grow and branch, while the bud induces the mesenchyme to make nephrons and stroma. Auerbach and Grobstein (1958) made cell suspensions of metanephrogenic mesenchyme, reaggregated them, and cultured them. On its own this reaggregate did nothing but, when placed in contact with inducing tissue, it made tubules. This showed that the disaggregation and reaggregation process did not itself substitute for induction: the rules of development in reaggregates apparently remained the same as in the embryo. The researchers also made suspensions of inducing tissue, mixed them with suspensions of metanephrogenic mesenchyme, and cultured the reaggregate. The cell types segregated and inductive signals passed to the mesenchymal cells, inducing them to make nephrons, and indicating that inductive activity is robust to disaggregation and reaggregation. The inducing tissue used for these experiments was spinal cord, because ureteric buds die. Only decades later, when the Auerbach and Grobstein work was revisited, was a method developed to prevent anoikis in the ureteric bud cells of a disaggregate−reaggregate in culture: in the presence of an inhibitor of Rho-activated kinase (ROCK), the ureteric bud cells survive, segregate from the mesenchyme, and make small collecting duct cysts and treelets that induce the metanephrogenic cells to make nephrons, the whole being a renal organoid (Unbekandt and Davies, 2010).

Mechanisms of Organoid Formation

An important question hanging over the work on organoids in the 1940s and 1950s was the mechanism by which cells of a mixed reaggregate sort out into distinct populations. The most determined and effective approach to this was probably that of Malcolm Steinberg, who proposed that the sorting of cells was due to differential adhesion (Steinberg, 1963). His argument rested on essentially thermodynamic grounds: he proposed that cells express adhesion systems on their surfaces, that different adhesion systems adhere with different strengths (i.e., that binding reduces the free energy in the system by different amounts), and that homotypic adhesive systems will adhere more strongly (lower free energy more) than heterotypic ones. Under these assumptions, and allowing cells to move, the lowest energy state would be one in which cells bind to their own kind in preference to remaining mixed: cells mixtures would therefore separate into homogenous populations (Fig. 1.2). What is more, cells of the more adhesive type will be in the core of a reaggregate, and the less adhesive cells will surround them. In this view, the information content of the cells, in terms of their propensity for self-organization, therefore reduces simply to a quantitative measure of their stickiness.

Figure 1.2 Computer simulation of adhesion-mediated phase separation. In each case, homotypic adhesion between green cells and homotypic adhesion between red cells is greater than heterotypic adhesion between red and green cells, and cells are allowed to change positions if this will reduce the free energy of their binding. The result is that randomly-arranged starting populations undergo partial phase separation, but separation cannot complete because islands of one color become separated by the other, and cannot cross without making energetically unfavorable moves. The system therefore remains in a local optimum with a rich alternation of cell types.

Since Steinberg suggested the differential adhesion hypothesis, measurements of the relative adhesion strengths of different cells, in homotypic and heterotypic contacts, have been made, and these adhesion strengths have been found to be predictive of sorting behavior. In a more direct test, Foty and Steinberg (2005) engineered different clones of a cell line to express different amounts of the same adhesion molecule, and showed that the cells sorted with the more adhesive in the core. This test separated differences in the amount of adhesion from differences in the type of molecule, or the type of cell, and therefore supported the focus on adhesion alone as the critical determinant of sorting behavior. There is now evidence that the precise type of adhesion molecule may actually matter much less than was originally supposed: when cells expressing different cadherin adhesion molecules at the same surface concentration are mixed, they may remain mixed, suggesting that quantitative differences in the number of molecules present may be much more important than cadherin type (Duguay et al., 2003). This conclusion is supported by direct measurements of cadherin adhesion on non-living substrates, which show that heterotypic contacts adhere about as well as homotypic ones (Shi et al., 2008).

It is important to note that, despite the strong evidence that sorting behavior can be predicted by adhesion, recent data challenge the view that the underlying mechanism can be explained in terms of simple thermodynamics. In real embryos, thick actin−myosin cables run along the boundary between epithelial cells that express different cadherins. Mutations or knockdown of myosin II prevent the actin−myosin cables forming, and cells start to mix (Monier et al., 2011). This suggests that cells may detect reduced adhesion along a particular interface and actively use actin−myosin to increase surface tension there, reducing the length of the boundary as much as possible, and thus minimizing the area of heterotypic contact. This mechanism is discussed further in Davies (2013); the main point of relevance to this chapter is that sorting is still predictable by adhesion, but uses active cellular mechanisms rather than simple minimization of free energy. In the face of this on-going debate, some authors (e.g., Davies and Cachat, 2016) have moved to using phase separation to refer to the sorting of cell types with different homo- and heterotypic interaction properties, as a term that captures the effect while remaining agnostic about the underlying mechanism.

The ability of cells to sort out by adhesion-mediated phase separation is restricted by their capacity to move, and by the system’s risk of becoming trapped in a local optimum. Once separation has begun, and clusters of cells of one type form as islands in a sea of the other type, the islands will minimize their boundary area and be relatively rounded. This will be a local optimum, in the sense that any distortion of the boundaries—for example, an invasive tendril growing out from an island—would increase the interfacial area. Without such growth, though, the islands cannot meet and coalesce to reach the global optimum of only one straight boundary (Fig. 1.2). The system is therefore trapped in its local optimum, with a rich alternation of cell types. This restriction has been exploited in the construction of synthetic biological systems that generate de novo patterns using nothing but adhesion-mediated phase separation (Cachat et al., 2016).

The restricted sorting caused by the system becoming trapped in a local optimum is probably critical to the formation of organoids from mixed progenitor cells. For example, in a kidney organoid formed from the kidney progenitor by dissociation and reaggregation, ureteric bud cells that express large quantities of E-cadherin form tight coalescences; islands in a sea of metanephrogenic mesenchyme. Each of these epithelial islands develops into a cyst, and then into a tubule that goes on to branch to make a ureteric bud treelet. Inductive signals from each treelet then organize the surrounding metanephrogenic mesenchyme to make nephrons (Fig. 1.3). An interesting question, still moot, is whether the stability of even mature tissues relies, at least in part, on the system being trapped in a local optimum.

Figure 1.3 A kidney organoid produced from human iPS cells, showing a close alternation of different cell types. Ureteric bud cells are stained for Calbdinin-D-28k (green), the proximal tubule part of nephrons are stained with LTL lectin (blue), and glomerular podocytes are stained for WT1 (red). Image credit: Mona El-Hendawy and author.

The Path to Human Organoids

For decades, almost all work and commentary on organoids was done from the point of view of basic developmental biology: organoids were tools that could inform embryologists about cellular mechanisms of development. They had the useful feature that they would allow questions to be asked in a simple system, away from the complexity of the body as a whole. For such uses, animal sources were as useful as human—more so, perhaps, because organoid data could be correlated with in vivo experiments. Occasionally, suggestions for clinically relevant applications were made. In the 1980s, for example, Lauri Saxén suggested the use of organoids as well as intact, cultured embryonic organ rudiments to explore mechanisms of teratogenesis and foetal toxicology (Saxén, 1988). The use of animal organoids for the purposes of studying toxicology would, however, run into the same issues in extrapolating animal data to human that are so problematic in standard in vivo animal experiments. For predicting human toxicity, human organoids would obviously be better. In addition, any hope of using organoids as a basis for making transplantable tissues for human clinical use would clearly require human cells as a starting point. For these reasons, interest in making human organoids has steadily grown.

Until the development of ES and iPS cells (see below), production of human organoids depended on the use of fragments of tissue from human foetuses, children or adults. Human keratinocytes, amongst the easier cells to obtain (particularly in countries that practice male circumcision), will form realistically layered epidermal organoids when cultured at an air/medium interface (Noël-Hudson et al., 1995). Mixed with fibroblasts, they organize into a skin organoid that includes a dermis-like layer (Kim et al., 1999). Simple neuronal organoids were constructed in the 1980s and 1990s from foetal brain cells (Lodin et al., 1981), for the purpose of studying the dynamics of virus infection (McCarthy et al., 1991) and neuronal physiology (Aquila-Mansilla and Barnea, 1996; Barnea and Roberts, 1999). Such organoids have also been tested for their ability to integrate into adult animal brains (Bystron et al., 2002). Similarly, human thymocyte aggregates have been used to explore T-cell immunology (Choi et al., 1997).

What made a vast difference to the interest in growing organoids for application was the development of the first human embryonic stem cell (hES cell) lines (Thomson et al., 1998) and, later, human-induced pluripotent stem cell (hiPS cell) lines that can be made from differentiated tissue (Yu et al., 2007). In principle, hiPS cells can be differentiated into any foetal cell type, typically by exposing them to the same sequence of signaling molecules that lead to cells of the same type during normal embryonic development. This created the hope that hiPS cells, perhaps specific to a particular human phenotype, could be turned into the starting material for making organoids, allowing clinically important studies on pathology, toxicology, teratology, etc., to be made on the correct species, and even the correct individual or patient subgroup. It also created the hope that such organoids may be a first step to making new organs for the purposes of transplantation.

Some work in this direction was begun using mouse ES cells, before their human counterparts were available, so that differentiation and culture strategies would be in place for human pluripotent stem cells when they had been developed. Examples of this mouse ES work include the production of gut organoids (Ishikawa et al., 2004), organoids containing kidney components (Kim and Dressler, 2005), cardiac organoids (Guo, 2006), and liver organoids (Mizumoto et al., 2008). Once human ES and iPS cells became available, human organoids began to be produced to represent neural tissue (Schwarts, 2015), prostate (Calderon-Gierszal and Prins, 2015), thyroid (Ma et al., 2015), and kidney (Morizane et al., 2015; Takasato, 2015). For at least some systems, it has been possible to avoid the ESC/iPSC stage by transdifferentiating fibroblasts into the desired organ progenitor cell type; this has, for example, been done for thymus (Bredenkamp et al.,