Human Pluripotent Stem Cell Culture: Considerations for Maintenance, Expansion, and Therapeutics

Summary

Human pluripotent stem cells (hPSCs) provide powerful resources for application in regenerative medicine and pharmaceutical development. In the past decade, various methods have been developed for large-scale hPSC culture that rely on combined use of multiple growth components, including media containing various growth factors, extracellular matrices, three-dimensional environmental (3D) cues and modes of multicellular association. In this review, we dissect these growth components by comparing cell culture methods and identifying the benefits and pitfalls associated with each one. We further provide criteria, considerations, and suggestions to achieve optimal cell growth for hPSC expansion, differentiation, and use in future therapeutic applications.

Introduction

Human pluripotent stem cells (hPSCs), including both human embryonic stem cells (hESCs) and induced pluripotent stem cells (hiPSCs) (Takahashi et al., 2007; Thomson et al., 1998; Yu et al., 2007), represent important cell resources and hold tremendous promise for cell-based therapies, drug discovery, disease modeling, and pharmaceutical applications (Daley, 2012; Engle and Puppala, 2013). At present, more than 234 hPSC lines have been registered at the National Institutes of Health (NIH), which are eligible for use in federally funded research. Numerous discussions have taken place in recent years about standards for derivation, registries, characterization, storage, banking, distribution, and engineering of hPSCs for both research and therapy (Adewumi et al., 2007; Akopian et al., 2010; Andrews et al., 2009; Borstlap et al., 2010; Chaddah et al., 2011; Crook et al., 2010; Panchision, 2013; Rao, 2013; Stacey et al., 2013; Turner et al., 2013). The subsequent resources and platforms that have emerged have opened the door to therapeutic application for regenerative medicine and for biomedical research.

In the past decade, methods for hPSC culture have evolved rapidly in an attempt to meet the pressing needs of regenerative medicine and drug discovery (Figure 1). Despite rapid progress in developing new hPSC culture methods however, we still face many problems that need to be resolved prior to their future application. These challenges include: (i) a lack of standardized protocols for specific applications, (ii) an absence of efficient assays to monitor cellular stress induced by suboptimal growth conditions, and excessive apoptotic and spontaneous differentiation signals during cell processing, (iii) the impurity or heterogeneity of propagated cells (Serra et al., 2012; Stewart et al., 2006), (iv) genomic instability related to chromosomal abnormalities (Liang and Zhang, 2013), and (v) potential tumorigenicity (Lee et al., 2013; Malchenko et al., 2010).

Figure 1.

Timeline of the development of cell culture modules for hPSCs and future directions (Year 2014 to 2016).

In this review, we will (i) briefly describe the principles underlying hPSC expansion, (ii) discuss current cell culture modules and platforms for hPSC culture, (iii) present basic methods for hPSC maintenance, (iv) dissect new culture methods for expansion of both hPSCs and differentiated cells in bioreactors, (v) summarize practical hPSC culture methods for genetic engineering and high-throughput drug screening (Table 1), and finally (vi) elucidate current challenges and highlight future directions.

Table 1.

Summary of cell culture methods for undifferentiated and differentiated hPSCs

Methods & Applications	Advantages	Disadvantages	Applications	Key references
Colony-type Cultures (CTC)
on MEF feeder	well characterized, predicative outcomes, support various hPSC line growth, economical	xenogeneic feeder, undefined, variability in components, labor-intensive and time-consuming	SCM, RES, DA, hPSC line derivation, single-cell cloning	1. Thomson et al. 1998
on human feeders	xeno-free, clinical-grade hESC derivation	undefined components, variability in culture	SCM, RES, DA, cell-based therapeutics	2. Richards et al. 2002
on Matrigel	feeder-free	partially defined components, variability in culture	SCM, RES, DA	3. Xu et al. 2001
on vitronectin	xeno-free, chemically defined components	lower efficiency	SCM, RES, DA, cell-based therapeutics	4. Braam et al. 2008c
on laminin isoform 511	xeno-free, chemically defined components	low yield of purified laminin protein, expensive	SCM, RES, DA, cell-based therapeutics	5. Rodin et al. 2010
on synthetic surfaces	xeno-free, chemically defined substrates, economical	NA	SCM, RES, DA, cell-based therapeutics, small-scale hPSC expansion	6. Saha et al. 2010
General Comments (based on refs. 1–12)	conventional methods for hPSC culture, de novo hPSC line derivation, compatible with many growth conditions, sustain pluripotent states, high differentiation potential, not require single-cell dissociation steps, not require small molecules for passaging	Low recovery rates from cryopreservation, reported chromosomal abnormalities, low transfection rates (3 –35%), low efficiency for transduction. heterogeneity & variability; not ideal for single-cell analysis, high-throughput (HTP) assays, and hPSC expansion	hPSC maintenance, hPSC banking, research-grade experiments, lineage differentiation, non-HTP drug assays, single-cell cloning, EB derivation, pluripotent state conversion	7. Liew et al. 2007 8. Braam et al. 2008a 9. Braam et al. 2008b 10. Chen et al. 2012a 11. Mallon et al. 2006 12. Hanna et al. 2010
Non-colony Type Monolayer (NCM)
on Matrigel	high plating efficiency, robustness, high cell yield related to colony culture, high efficiency for genetic manipulation	use various small molecules (e.g., ROCKi & JAKi) for single-cell plating	SCM, RES, HTP assays, hPSC expansion, cryopreservation	13. Chen et al. 2012a
on Matrigel	use ROCKi only for initial cell passages, high pluripotency	observed in some hPSC lines	SCM, RES, HTP assays, cryopreservation	14. Kunova et al. 2013
on laminin isoform 521	without use of ROCKi or JAKi, xeno- & chemically-defined conditions	lower efficiency than ROCKi, not ideal for hPSC expansion (costly)	SCM, RES, HTP, cell-based therapeutics	15. Chen et al. unpublished
General Comments (based on refs. 13–15)	higher scalability & cell production, consistent growth rates: 3-day culture, relatively homogeneous cells, compatible with xeno-free cultures, high recovery rates after thawing cells, versatile for transfection & transduction	lower cell production than SC, reported chromosomal abnormalities, selection pressure for mutated cells?	hPSC expansion, cell banking, cell-based therapeutics, genetic manipulation
Suspension Culture (SC)
clump inoculation	feeder- & matrix-free	heterogeneity of aggregates	EB differentiation	16. Gerecht-Nir et al. 2004
clump inoculation	optional use of ROCKi	intermediate cell expansion	cell-based therapeutics	17. Amit et al. 2010
single-cell inoculation	controllable size of aggregates demonstrated high cell yield	long-term use of ROCKi	hPSC expansion cell-based therapeutics	18. Krawetz et al. 2010 19. Chen et al. 2012
single-cell inoculation	early development of SC in hPSCs, controllable size of aggregates	relative lower cell production	hPSC expansion, cell-based therapeutics	20. Steiner et al. 2010
SCMC (refs. 21–23)	adjustable growth areas, reduced shear-force damages, demonstrated high cell yield different microcarriers	variability of MC attachment require coating with MC require cell dissociation with MC	hPSC expansion cell-based therapeutics	21. Lock & Tzanakakis 2009; 22. Fernandes et al, 2009; 23. Oh et al. 2009
SCMC_ME	better designs of shear-force protection, various, economical encapusulation	compromized cell production, need to do decapsulation steps, increased complexity	hPSC expansion, cell-based therapeutics	24. Serra et al. 2011
General comments (based on refs. 16–29)	feeder- and/or matrix-free, high hPSC expansion rates, compatible with spinner flasks/bioreactors, monitoring of growth conditions, controllable autocrine and paracrine, compatible with MC or ME or both, micro-version available: microfuidic bioreactors, cellular plasticity between SC and CTC	agitation-induced shear force, altered gene expression patterns, increased heterogeneity, compromised pluripotency, cell loss after mechanical passaging, require ROCKi for single-cell or small clump passaging, variability in expansion rates, not ideal for HTP screening	hPSC expansion cell-based therapeutics cell transplantation organogenesis	25. Ungrin et al., 2008 26. Siti-lsmail et al. 2008 27. Zweigerdt et al.2011 28. Serra et al. 2012 29. Villa-Diaz et al. 2009

Abbreviations:

DA, low-throughput drug assay, EB, embryoid body; HTP, High-throughput assay; JAKi, JAK inhibitor 1; MC, microcarriers; ME, microencapsulation; NA, not determined; refs: references; RES: research experiments; ROCKi, ROCK inhibitors; SC, suspension culture; SCM, stem cell maintenance

Human PSCs Require Distinct Conditions from Murine PSCs

Unlike mouse embryonic stem cells (mESCs), which rely on BMP-4 and Stat3 in the presence of leukemia inhibitory factor (LIF) for self-renewal (Ying et al., 2003), both hESCs and hiPSCs depend on cooperation between different signaling pathways that are related to basic fibroblast growth factor (FGF-2), Noggin, Activin/Nodal, and TGF-β pathways (James et al., 2005; Vallier et al., 2005; Wang et al., 2005; Xiao et al., 2006; Xu et al., 2005). Thus, the pluripotent states of mESCs and hPSCs are quite different, with mESCs and hPSCs being described as representing naïve and primed pluripotent states respectively (Nichols and Smith, 2009; Ying et al., 2008). Phenotypically, mESCs can passage and grow as single cells, whereas hPSCs have a drastic loss of viability after enzymatic dissociation as single cells. Thus, hPSCs are generally plated as clumps and grow as colonies and aggregates. Therefore, the genetic and phenotypic differences between mESCs and hPSCs have necessitated different culture modules to support hPSC growth in vitro.

Interestingly, the use of LIF with two inhibitors (i.e., GSK-3βi and ERK1/2i, abbreviated as 2i), which suppress glycogen synthase kinase-3β (GSK-3β) and extracellular signal-regulated kinases 1/2 (ERK1/2), supports naïve mESC growth under feeder-free conditions in defined medium (Ying et al., 2008). Recently, Gafni et al. reported direct conversion of primed hPSCs to the naïve state using naive human stem cell medium (NHSM) that contains LIF and 2i. These naïve hPSCs had approximately 40% increase in single-cell cloning efficiency (Gafni et al., 2013). Through a different screening strategy for small molecule inhibitors that support naïve pluripotency in hPSCs, Chan et al. identified a distinct LIF-dependent pluripotent state that also harbors a gene expression signature for native preimplantation epiblast (Chan et al., 2013), thus demonstrating another method for converting hPSCs to a naïve state. Hence, manipulation of pluripotent states through perturbation of growth factor signaling enhances single-cell plating efficiency, which has greater potential for both genetic engineering and hPSC expansion.

Key Contributing Factors in hPSC Culture

The growth of any type of mammalian cell in vitro requires growth media, extracellular matrices, and environmental factors. Here, we will discuss three key factors that influence the quality, robustness, and utility of various hPSC culture methods: (i) growth medium, (ii) extracellular matrices, and (iii) environmental cues (e.g., a growth environment in a bioreactor) (Table 1). Ideally, the information about various cell culture components described here can be used to formulate new and tailored protocols for specific uses.

Growth medium development for hPSC culture

Growth medium is one of the most critical components of hPSC culture, and has undergone a dynamic evolution since it was initially used for hESC culture (Thomson et al., 1998). The ultimate goal for therapeutic use is to develop a serum-free, xeno-free, and chemically defined medium, suitable for supporting the growth of almost all types of hPSC lines (Figure 1). The first generation of hESC medium commonly contained fetal bovine serum (FBS) and undefined/conditioned secretory components from mouse embryonic fibroblasts (MEFs). In recent years, scientists have established more standardized and better-defined medium to replace xenogeneic elements in media (Genbacev et al., 2005; Li et al., 2005). Vallier et al. used chemically defined medium with Activin A, Nodal, and FGF-2 to propagate hESCs (Vallier et al., 2005). The Knock-Out Serum Replacement (KSR) is widely used with FGF-2 to support feeder-based hPSC culture, and a defined culture medium (termed TeSR1) containing FGF-2, lithium chloride (LiCl), γ-aminobutyric acid (GABA), TGF-β, and pipeolic acid was developed by Thomson and colleagues for use in feeder-free conditions (Ludwig et al., 2006). More recently, Thomson and coworkers developed chemically defined E8 medium (E8), which is a derivative of TeSR1 containing 8 components, that lacks both serum albumin and β-mercaptoethanol. This E8 medium, combined with EDTA passaging, may be suitable for culturing a broad range of hiPSC and hESC lines, particularly to improve episomal vector-based reprogramming efficiencies as well as experimental consistency (Chen et al., 2011b; Chen et al., 2010b).

Extracellular components

Extracellular components contain diverse organic matrices from animal cells, hydrogel, individual matrix proteins, synthetic surfaces, and some commercially well-defined and xenogeneic-free components. The major commercially available products include CELLstart™, which contains components only of human origin (Invitrogen Inc.), StemAdhere™, which has defined matrix with fully human proteins produced in human cells under completely defined conditions (Primorigen Biosciences Inc.), and Synthemax®-R Surface, a unique synthetic peptide acrylate-coating surface (Corning Inc.).

Matrigel

Thus far, Matrigel has been one of the most widely used extracellular components for feeder-free culture of hPSCs. It is a basement membrane matrix, rich in types I and IV collagens, laminin, entactin, heparan sulfate proteoglycan, matrix metalloproteinases, undefined growth factors, and chemical compounds (Kleinman et al., 1983; Kleinman et al., 1982; Mackay et al., 1993; Vukicevic et al., 1992). Although it is widely used for research purposes, it is important to note that Matrigel, which is a semi-chemically defined, xenogeneic substrate, does not support hPSCs for clinical therapies.

Extracellular matrix (ECM) proteins

Many ECM proteins are developmentally regulated, and some can be used to support hPSC self-renewal or lineage commitment (Braam et al., 2008c; Rodin et al., 2010; Xu et al., 2001). Recombinant vitronectin is a defined substrate that sustains hESC self-renewal through adhesion with αVβ5 integrin (Braam et al., 2008c). There is also increasing evidence showing that specific laminin isoforms (expressed in post-implantation embryos) may play an important role in sustaining long-term hPSC growth. By plating hPSC cell clumps on cell culture dishes coated with the human recombinant laminin-511 (LM-511), Tryggvason and colleagues found that this single laminin isoform maintained self-renewal of normal hPSCs for more than 20 passages (Rodin et al., 2010). This laminin isoform-based protocol is free from animal products and feeders with only a single undefined substrate (i.e., human albumin), producing homogeneous hPSCs that are suitable for future therapeutic use (Rodin et al., 2010). Unfortunately, the current use of laminin for both hPSC maintenance and expansion is limited due to the high cost of obtaining highly purified proteins. New methods that utilize synthetic surfaces are being developed to simulate the effects of ECM proteins (such as laminin) on hPSC growth.

Synthetic surfaces

It is feasible that synthetic surfaces could mimic major signal transduction pathways that are required for hPSC growth. Synthetic surfaces that modulate TGF-β signaling can influence TGF-β-related cell fate decisions (Li et al., 2011), and surface arrangement of properly condensed peptides (e.g., laminin peptides) define new 3D synthetic scaffolds that support hPSC growth (Derda et al., 2007). Moreover, modifications of cell growth surfaces with simplified techniques could facilitate the development of chemically defined conditions to expand clinically relevant hPSCs at low cost. For example, Jaenisch and coworkers modified a cell culture plastic surface by UV/ozone radiation, producing significantly improved hPSC numbers under fully defined conditions. There was a 3-fold increase in hPSC numbers compared with feeder-based culture (Saha et al., 2011). Several groups have also designed new surface substrates based on a high-affinity cyclic arginine-glycine-aspartate (RGD) peptide that contains the RGD integrin recognition sequence (Kolhar et al., 2010), synthetic peptide–acrylate surfaces (Melkoumian et al., 2010), and synthetic polymer coating with poly[2-(methacryloyloxy)ethyl dimethyl-(3-sulfopropyl)ammonium hydroxide] (PMEDSAH) (Villa-Diaz et al., 2010) for long-term maintenance of hPSCs. Thus, advanced material technology could offer fully synthetic environmental cues that sustain long-term culture of clinical-grade hPSCs.

Environmental cues

There are a number of environmental cues, including cues from both physical and physiological environments (besides the cells, growth medium components, and ECM) that encourage hPSC growth, such as temperature, humidity, osmosity, acidity, rigidity of growth surfaces, cell density, gas diffusion exchange, and modes of multicellular associations. Among these various factors, we will highlight the consumption of oxygen (O₂) and the modes of multicellular associations, which are two key factors that influence hPSC growth.

The physiological environment of early-stage mammalian embryo development is hypoxic. Human ESC culture needs to be designed to maximally mimic this in vivo condition. However, conventional hESC culture has been traditionally implemented under normoxia (i.e., 21% O₂) conditions. Low oxygen tension (~2 to 3% O₂) at physiological levels prevents spontaneous differentiation of hESCs (Ezashi et al., 2005). Moreover, physiologic oxygen (2%) also enhances hESC clonal recovery by 6-fold and reduces spontaneous chromosomal abnormalities without affecting expression of pluripotentcy markers in both WA01 (H1) and WA09 (H9) hESCs (Forsyth et al., 2006). Telomere length in some hESCs lines (e.g., WA09 cells) is more sensitive to oxidative stress than others (e.g., WA01 cells) under normoxia (Forsyth et al., 2006). Thus, optimal oxygen levels would be required to maintain genomic integrity, which should be further validated for hPSC expansion under different growth conditions.

In contrast to oxygen tension, the modes of multicellular association are primarily physical properties of intercellular interactions. However, these physical properties have a significant impact on cell density, ligand-receptor interactions, signal gradient processing, intracellular signal transduction, and the microenvironments of hPSCs. Conventionally, hESCs are grown on MEFs or on Matrigel as colonies and differentiate as embryoid body (EB) aggregates. With the discovery of single-cell death and survival mechanisms (Androutsellis-Theotokis et al., 2006; Chen et al., 2010b; Ohgushi et al., 2010; Watanabe et al., 2007), we can now grow hPSCs in different modes such as colonies (Thomson et al., 1998), single cells (Tsutsui et al., 2011), single-cell based non-colony type monolayer (NCM) (Chen et al., 2012a; Kunova et al., 2013), and suspended aggregates (Gerecht-Nir et al., 2004; Steiner et al., 2010; Ungrin et al., 2008). The choice of a specific mode for hPSC growth would depend on individual research aims, and pharmaceutical or clinical applications.

hPSC Maintenance Methods Should be Tailored for Different Uses

Maintaining high-quality hPSCs is essential as all subsequent applications would depend on the starting materials. Large-scale maintenance of these cells is usually not necessary in the research environment. In general, with respect to different usages of the cells, corresponding methods should be adjusted in order to reduce the workload and cost.

Feeder-based culture of hPSC colonies is suitable for routine maintenance

MEFs are the most frequently used feeder cells because they support robust growth of all-types of ES cells as colonies (Mallon et al., 2006; Thomson et al., 1998). However, since MEFs have complex, undefined, and xenogeneic properties, various human cell types have been substituted to support hESC growth, including hESC derivatives. Bongso and colleagues initially reported that human fetal and adult fibroblast feeders support long-term undifferentiated growth of hESCs (Richards et al., 2002). Human cells from adult tissues such as fallopian tube, foreskin, marrow-derived stromal cells, and uterine endometrium have all been exploited to expand hESCs (Mallon et al., 2006). The use of feeders is suitable for routine hPSC maintenance, genetic engineering, and single-cell cloning. However, feeder cultures are not designed for clinical application, as they could be potentially problematic due to the introduction of undefined factors into cultures.

Feeder-free culture of hiPSC colonies provide well-defined conditions

Considerable effort has been made to develop feeder-free conditions with defined medium such as KSR supplemented with Activin A and FGF-2, which was found to support undifferentiated self-renewal of hESCs (Xiao et al., 2006). FGF-2 and suppression of BMP signaling sustain the pluripotent state of hESCs (Xu et al., 2005) and is often used to support undifferentiated growth of hESCs under feeder-free conditions (Wang et al., 2005). Despite the presence of animal-derived components in KSR, it was still employed to derive 6 clinical-grade hESC lines (Crook et al., 2007). It would be interesting to compare these lines to the next generation of clinical-grade hESC lines, which have been derived under completely xeno-free conditions.

De novo xeno-free culture is required for clinical application

Complete xeno-free derivation and maintenance of hESCs is an enormous task because it requires implementation of xeno-free practices at every step during the derivation process, including mechanical isolation of the ICM, use of human feeder cells, and propagation and maintenance in xeno-free defined medium. This approach is designed for clinical application and, due to its time-consuming nature, may not be suitable for some research experiments.

Overall, quality control of hPSC culture is a paramount issue because these maintained cells could potentially be used for clinical applications and for many important assays involved in drug. Alterations in cellular states and identities are certain to affect therapeutic outcomes and data interpretation. Therefore, periodic examination is recommended to verify gene expression profiles, gene copy numbers, and chromosomal integrity (reviewed in (Baker et al., 2007; Lee et al., 2013; Pistollato et al., 2012).

Suspension Cultures Enable Large-Scale Production of hPSCs

Generation of clinically relevant quantities of hPSCs, ranging from 10⁷–10¹⁰ or beyond, is essential for their clinical use. Suspension culture in bioreactors provides a promising platform to robustly manufacture hPSC products for this purpose. Generally, hPSCs expanded by suspension culture in bioreactors remain pluripotent and chromosomally stable. Among many types of bioreactors, spinner vessels and stirred-tank bioreactors are of particular interest. These bioreactors are equipped with a glass vessel having a working volume from 50 mL to 200 L and an impeller to provide a homogeneous growth environment and allow efficient gas and nutrient transfer (Serra et al., 2012; Ungrin et al., 2008). These machines can precisely control culture conditions by real-time monitoring of temperature, oxygen levels, acidity or basicity, autocrine, growth factor consumptions, and metabolite concentrations. Technically, suspension culture can be performed in several ways, including aggregated hPSC clumps, hPSCs immobilized on microcarriers, and microencapsulation.

At present, there are no consensus views on how to assess the growth rates of various expansion protocols, because they are implemented in different laboratories. To facilitate the comparison among various growth rates generated from different laboratories, we would like to introduce a simple parameter, i.e., fold increase in cell number per day (abbreviated as FIPD) for any given number of days of culture. The concept of FIPD provides a quick analysis of growth rates of propagated cells (Table 2). However, we also need to point out that there are many confounding variables involved in the regulation of cell growth rates. Furthermore, the changes of growth rates are a dynamic process, which vary with initial inoculated cell density and the modes of multicellular association. So, we also recommend that researchers examine the FIPD along with other parameters such as cell doubling times as well as the entire growth curves when comparing between different protocols.

Table 2.

Summary of cell growth rates in different hPSC culture methods

Cell culture	Inoculation	MC/ME	Bioreactors	Cells	FIPD	References
SC	single cells	NA	bioreactor	hESCs	4.2	Krawetz et al., 2010
SC	single cells	NA	spinner flask	WA09	4.2	Chen et al., 2012b
SC	clumps	NA	Erlenmeyer	hPSCs	2.5	Amit et al., 2010
SC	clumps	NA	STLV	hESC EBs	2.5	Gerecht-Nir et al., 2004
SC	single cells	NA	spinner flask	hiPSCs	1.5	Zweigerdt et al., 2011*
SC	single cells	NA	spinner flask	hiPSCs	0.9	Zweigerdt et al., 2011*
SC	single cells	NA	spinner flask	hESCs	0.7	Steiner et al., 2010
SC	clumps	NA	spinner flask	hESC EBs	0.7	Cameron et al., 2006
SC	clumps	NA	spinner flask+gbi	hESC EBs	0.6	Yirme et al., 2008*
SC	clumps	NA	spinner flask+pi	hESC EBs	0.2	Yirme et al., 2008*
SC	clumps	NA	STLV + pi	hESC EBs	0.1	Yirme et al., 2008*
SCMC	clumps	MG-polystyrene	spinner flask	hESCs	5.6	Lock & Tzanakakis, 2009*
SCMC	clumps	MG-polystyrene	spinner flask	hESCs	4.3	Lock & Tzanakakis, 2009*
SCMC	clumps	MG-DE53	spinner flask	HES-3	2.0	Chen et al., 2011a*
SCMC	clumps	MG-DE53	spinner flask	HES-2	1.0	Chen et al., 2011a*
SCMC	clumps	nc-Cytodex 3	spinner flask	hESCs	0.5	Fernandes et al., 2009
SCMC	clumps	MG-cellulose	spinner flask	HES-3	4.0	Oh et al., 2009*
SCMC	clumps	MG-Cytodex 3	ULAP	hESCs	1.3	Nie et al., 2009
SCMC_ME	single cells	plus alginate	spinner flask	hPSCs	1.0	Serra et al., 2011*
SCMC_ME	single cells	no alginate	spinner flask	hPSCs	0.4	Serra et al., 2011*
Static SC	clumps	LM-polystyrene	ULAP	HES-3	1.2	Heng et al., 2012*
Static SC	clumps	VN-polystyrene	ULAP	HES-3	1.2	Heng et al., 2012*
Static SC	clumps	MG-cellulose	NA	HES-3	1.0	Oh et al., 2009*
Static NCM	single cells	NA	NA	hPSCs	1.0	Chen et al., 2012a*
Static SC	single cells	Hillex II	ULAP	hESCs	0.6	Phillips et al., 2008
Static colonies	single cells	NA	NA	hPSCs	0.4	Chen et al., 2012a*

Notes:

The growth rates are evaluated by fold increase (relative to the initial seeding cells) in cell number per day (i.e., FIPD). Representative studies are shown. Asterisk signs indicated pair-wise designs for comparative studies in the same study or publication. Abbreviations: clumps: hPSC clump inoculation; gbi, glass ball impeller in spinner flasks; LM-polystyrene, laminin-coated polystyrene; MG-polystyrene, Matrigel coated with polystyrene; nc-cytodex 3, not coated with the microcarrier cytodex™3 as control in the study; NA, not available; NCM, single-cell based non-colony monolayer culture; pi: paddle impeller in spinner flasks; SC, suspension culture; SCMC, suspension culture combined with microcarrier (MC); SCMC_ME: suspension culture combined with both microcarrier (MC) and microencapsulation (ME) methods; STLV, slow-turning lateral vessels; ULAP, ultra low attachment plates from Corning Inc.; VN-polystyrene, vitronectin-coated polystyrene.

Suspension culture of hPSC aggregates vs single cell inoculation

Stirred-suspension bioreactors or spinner flasks were originally used for passaging human EBs (Dang et al., 2004; Gerecht-Nir et al., 2004) and subsequently for propagating mESC aggregates (Fok and Zandstra, 2005). Recently, this method has been increasingly used for expansion of undifferentiated hPSCs (Amit et al., 2010; Steiner et al., 2010), resulting in various culture methods (Table 2). Routinely, there are two ways of seeding cells for hPSC-aggregated cultures: clump versus single-cell inoculation. One of the major problems of clump inoculation is that it is difficult to control the size of hPSC clumps, thus leading to increased apoptosis or spontaneous differentiation.

Single-cell inoculation for suspension culture appears to be one way to control the size of hPSC aggregates, using the ROCK inhibitor, Y-27632. The key steps of single-cell inoculation have been previously outlined (Steiner et al., 2010). However, this reported method had a lower expansion rate (FIPD = 0.7, based on the outcome from 1 week of cell culture compared with colony-type growth on feeders (Table 2). Moreover, suspension culture resulted in 58% cell loss after mechanical cell passaging whereas colony passaging had only 24% cell loss in the same study (Steiner et al., 2010). To improve cell expansion rates in suspension culture, Rancourt and coworkers used a modified single-cell inoculation method, by which single cells form aggregates in mTeSR medium containing ROCK inhibitor and rapamycin in bioreactors for 24 hours (Krawetz et al., 2010). The presence of rapamycin inhibits the appearance of differentiated cells and enhances cell vitality in cell aggregates. They were able to obtain 4.2-FIPD in 6 days (Krawetz et al., 2010). Similar expansion rates have been achieved using single cell inoculation of ESCs and iPSCs in different conditions (Chen et al., 2012b; Zweigerdt et al., 2011). Interestingly, the suspended cells resembled “EB-like” clusters in terms of morphology but they maintained the pluripotent state. Importantly, the epithelial properties of hPSCs were restored after replating these aggregates on feeders, indicating the existence of both flexibility and reversibility within hPSCs under different growth modes.

Benefits and pitfalls of suspension culture

In general, suspension culture with spinner flasks can significantly increase aggregated hPSC production without the use of feeders. The scalability of this culture may be crucial to generating sufficient cells for future therapeutics. Another advantage is its naturally formed multicellular microenvironment, which may enable hPSCs to retain high differentiation capacity. In contrast, hydrodynamic shear force-related cellular damage presents a major limitation of this method (Serra et al., 2012). Another drawback involves the variability of growth rates that exist in different suspension cultures, possibly due to the heterogeneous sizes of aggregates. As cell aggregates increase in size and in irregularity, they may induce apoptosis-related cell loss, cellular differentiation, and heterogeneity. To avoid the formation of large aggregates, optimal cell inoculation methods and passaging schedules should be determined. In addition, the long-term effects of ROCK inhibition on hPSCs remain to be evaluated. Finally, heterogeneity induced by spontaneous differentiation may be due to the sensitivity of some hESC lines (e.g., HES-3 and IMR90) to shear force (Leung et al., 2011). With regard to this problem, porous microcarriers might reduce or overcome shear force related problems.

Suspension culture with microcarriers controls aggregation

The demonstration that mESCs can be immobilized on microcarriers and propagated in bioreactors (Fok and Zandstra, 2005) provides a proof of principle for using microcarriers as a simple and scalable way to control cellular aggregation in hPSC suspension culture. Various porous and nonporous carriers with matrices are commercially available (Chen et al., 2011a; Nie et al., 2009). They greatly enhance the surface area for cell adherence and gas diffusion. Adjustable surface area for cell growth reduces the consumption of medium and growth factors, thus having potential for future clinical development. Microcarriers such as coated polystyrene (Heng et al., 2012; Phillips et al., 2008), large positively charged spherical Cytodex™3 (Fernandes et al., 2009; Fok and Zandstra, 2005; Nie et al., 2009), and positively charged cylindrical cellulose (i.e., DE52, DE53, and QA52) (Chen et al., 2010a, 2011a; Oh et al., 2009) have high attachment efficiency and sustain the pluripotency of hPSCs (Table 2). Gradual loss of pluripotency has been observed in hESCs grown on uncoated microcarriers for continuous passaging, thus either Matrigel or laminin coating is critical for stable expansion of undifferentiated hESCs attached on various microcarriers. Taken together, current studies indicate that individual components such as laminins can replace Matrigel for microcarrier coating. However, dissociation of the cells from the carriers is also needed. Therefore, economically compatible and degradable microcarrier beads need to be developed. Despite the protective nature of microcarriers, there is still a need to control the size of cell aggregation.

Microencapsulation protects hPSCs in culture and during cryopreservation

Microencapsulation of cells in hydrogels has also been developed to avoid the shear-force microenvironment and excessive clump agglomeration in suspension culture (Murua et al., 2008). Important considerations for designing this microenvironment include the biocompatibility and biosafety of a specific material (e.g. calcium alginate and hyaluronic acid) and the ability to mimic in vivo embryonic niches. Microencapsulation of hESCs in 1.1% calcium alginate capsules enabled prolonged feeder-free expansion and maintenance of the pluripotent state (more than 8 months) in static culture (Siti-Ismail et al., 2008). This method is being considered as a tool for integrating expansion with cryopreservation of hPSCs (Siti-Ismail et al., 2008). Indeed, hPSCs immobilized on microcarriers and encapsulated in alginate in stirred tank bioreactors exhibited high cell recovery rates (>70%) after cryopreservation, which were three-fold higher than non-encapsulated cells (Serra et al., 2011). A 19-fold increase in cell concentration was found in encapsulated hPSCs, equivalent to 1.0-FIPD in 20-day culture, whereas only a 7.5-fold increase (FIPD = 0.4) was produced in the cells without alginate microencapsulation (Serra et al., 2011). Clearly, this method shares many advantages with porous microcarriers in terms of cell protection, enhanced surface areas, scalability and reproducibility. Notably, the diffusion of cell mass and gas and monitoring of cell growth inside the capsule are limited by the physical properties of encapsulation. In addition, a decapsulation process is required to harvest the cells.

In summary, the benefits of culturing hPSCs by stirred-suspension bioreactors could be substantial. Problems frequently encountered with suspension cultures, include the properties of microcarriers, the microenvironmental setting of bioreactors, cell passage methods, the use of small molecules, growth rate control, and differentiation pressure. Many issues need to be resolved prior to further clinical application. In this regard, it is necessary to compare suspension culture methods with other emerging and complementary methods.

Cell Culture Platforms for Assaying hPSCs

Non-colony type monolayer (NCM) expansion enables downstream analysis

Owing to several major limitations (e.g., induced heterogeneity and suspended growth), neither colony-type nor suspension culture is suitable for drug screening or single-cell analysis. In response to this unment need, Scientists, have developed a single-cell based non-colony type monolayer (NCM) culture (Chen et al., 2012a; Kunova et al., 2013). The key step in this method comprises seeding dissociated cells (single cells) at high density (i.e., ~1.4 × 10⁵/cm²) on Matrigel-coated polystyrene plates in the presence of ROCK inhibitor (Y-27632) or JAK Inhibitor 1 to facilitate the initial 24-hour single-cell plating efficiency and to prevent the formation of colonies (Chen et al., 2012a). Alternatively, hPSCs can be cultivated as NCMs on human recombinant laminin-521-coated polystyrene surfaces without the use of ROCK inhibitors. Generally, hPSCs under these growth conditions remain chromosomally normal, pluripotent, and differentiable into adult tissues of the three germ layers (Chen et al., 2012a).

The advantages of this culture method include: feeder-free, controllable growth rates, generation of homogeneous hPSCs, robust cell production (i.e., ~1.0-FIPD in 4-day culture), and rapid (2- to 4-day) cell recovery from cryopreservation compared with frozen cells from colony-type culture, which usually takes 1 to 3 weeks to recover (Chen et al., 2012a). Notably, hPSCs grown as NCMs are very efficient at forming teratomas and are reversible to colony-type culture when the cells are plated as clumps on MEF feeders (Kunova et al., 2013). After adaptation to NCM culture in the presence of ROCK inhibitor (Y-27632) during initial passages, some hPSC lines could be passaged as NCM without the inhibitor (Kunova et al., 2013). This method is also easy to manage and particularly suitable for genetic engineering and high-throughput drug screening (Chen et al., 2012a). However, long-term dissociated culture might select variant cells that sustain single-cell growth. Further investigation of the genomic stability of hPSCs at chromosomal and subchromosomal levels should be carried out to compare cells cultured as colonies, in suspension, and as NCMs.

Cell Culture Methods for Improved Genetic Engineering

Both transfection and transduction of cells with genetic materials of interest via various expression systems are efficient ways to investigate the functionality of RNAs and/or proteins. However, hPSCs from conventional colony-type culture are difficult to transfect or transduce, resulting in a greater variability in transfection/transduction efficiencies (Braam et al., 2008a; Braam et al., 2008b; Chen et al., 2012a; Liew et al., 2007). Notably, transfection efficiency was greatly improved when the cells were enzymatically dissociated (Braam et al., 2008a). Furthermore, when cells were dissociated and replated as a high-density monolayer, shRNAs, microRNAs, oligonucleotides, plasmid DNAs, and lentivirus were successfully introduced into hPSCs with high efficiency (Chen et al., 2012a; Padmanabhan et al., 2012). High density single-cell plating in the presence of the ROCK inhibitor Y-27632 enables the hPSCs to form loosely packed cell clusters within 24 hours, which might facilitate the uptake of DNAs, RNAs, and lentiviruses (Chen et al., 2012a).

Microfluidic Bioreactors Allow Precise Manipulation of Environmental Cues

Microfluidic bioreactors (known as micro-bioreactors, biochips, and cell-chips) are a micro-scale version of conventional bioreactors (macro-bioreactors). Microfluidic bioreactors integrate many monitoring and control features used by macro-bioreactors, which include fluidic, hydrodynamic shear, electrical signals, and optical components (Cimetta et al., 2009). Microfluidic bioreactors have shown their potential to establish highly controllable microenvironmental cues. The 3D niche likely mimics the in vivo microenvironments, including the spatial orientations of cells and ECM, temporal control of concentration gradient of soluble factors, and the availability of both oxygen and carbon dioxide (Cimetta et al., 2009).

Microfluidic bioreactors have been applied to study stem cell behaviors in the 3D microenvironment in real time, to analyze single hESCs derived colonies (Villa-Diaz et al., 2009), and to quantitatively control signaling trajectories contributed by both autocrine and paracrine in individual hPSCs (Moledina et al., 2012). Scientists have also created 3D vascular structures to examine the effects of drugs and to investigate the interactions between different types of vascular cells (van der Meer et al., 2013). Obviously, cellular and tissue model systems developed by microfluidic bioreactors may have broad applications in drug screening, cellular assay, and tissue engineering.

Suspension Cultures Enable Expansion of Differentiated Cells

The clinical application of hPSCs (e.g., cell-based replacement and drug sceening) will rely on the ability to obtain sufficient numbers of functional mature cells. Therefore, expansion of such large amounts of desired cells is of utmost importance. Scaled-up expansion could be made possible in suspension culture or in NCM formats. Currently, functionally differentiated cells can be obtained through multiple intermediate or precursor stages. Frequently, hiPSCs are expanded as EBs or precursors at the three germ-layer stage. Occasionally, these cells are expanded to generate terminally differentiated cells. In an early report, hESC EBs were cultivated in a slowly turning lateral vessel (STLV) bioreactor, resulting in a 70-fold increase in cell concentrations in 28-day culture (Gerecht-Nir et al., 2004). To optimize growth conditions, Itskovitz-Eldor and colleagues found that a spinner flask with the Glass Ball Impeller had a higher EB yield, more homogenous shape, and faster growth rate than a spinner flask with a paddle impeller and STLV bioreactor (Yirme et al., 2008). This simple comparison highlights the importance of mechanical stirring-force distribution, an easily ignored problem, in influencing the proliferation of differentiated EBs.

With respect to ectodermal lineage expansion, Reubinoff and coworkers reported direct conversion of hESC suspension culture into neural precursor spheres (Steiner et al., 2010). Furthermore, these hESC precursors could be differentiated in suspension into a highly enriched cell population for generating neurons that express β-III tubulin, tyrosine hydroxylase (TH), γ-aminobutyric acid (GABA), and glutamate, astrocytes that express glial fibrillary acidic protein (GFAP), and NG2-expressing oligodendrocyte progenitors (Steiner et al., 2010).

To expand the cells toward mesodermal lineages, hESC EBs (2–3 × 10⁵ cells/mL) propagated in spinner flasks produced a 15-fold increase in EB-derived cells in 21 days, which produced CD34⁺/CD31⁺ hematopoietic progenitors (5–6%) on day 14, comparable to the differentiation capacity of EB-derived cells under static culture (Cameron et al., 2006). Using a controlled fed-batch media dilution approach, Csaszar et al. demonstrated that they could control inhibitory feedback signaling, thus rapidly amplifying human cord blood cells ex vivo, producing 11-fold increase in functional HSCs (in 12-day culture, FIPD = 0.92). The ex vivo expanded HSCs exhibited the default capacities such as self-renewing and multilineage differentiation (Csaszar et al., 2012). This study offers strategies to expand primary cell culture ex vivo and also demonstrates the feasibility of multistep expansion of such hematopoietic lineages de novo from hPSC-derived progenitor cells in such a controlled system.

For cardiac differentiation, Lecina et al. reported that hESCs grown on laminin-coated TOSOH-10 in ultra-low attachment plates generated more than 80% beating aggregates, approximately 60% cardiomyocyte conversion related to the initial numbers of seeded hESCs (Lecina et al., 2010). To expand cardiomyocytes on a large scale, Zandstra and colleagues employed a micro-printing strategy to generate size-specified aggregates in an oxygen-controlled bioreactor (Niebruegge et al., 2009). This integrated approach enabled them to reduce the heterogeneity of cell aggregates and facilitated stirred bioreactor expansion of hPSCs as well as mesodermal differentiation. Under hypoxia (4% O₂), approximately 48.8% of cardiomyocytes were generated, as opposed to 19% contracting EBs under conventional dish culture (Niebruegge et al., 2009).

Concerning expansion of endodermal progenitors, it is also feasible to expand and differentiate hESCs to endoderm progeny on Matrigel-coated polystyrene microcarriers in spinner flasks. This method yielded 34- to 45-fold in hESC numbers in 8 days (FIPD = 4.3 to 5.6), with a definitive endoderm efficiency greater than 80% as determined by coexpression of SOX17 and FOXA2 (Lock and Tzanakakis, 2009).

Emerging Coculture Systems for Tissue Morphogenesis

Optimal hPSC bioprocesses should, together with hPSC expansion, differentiation, and cryopreservation, eventually direct differentiation toward organogenesis. To recapitulate the development of early vertebrate embryos, protocols call for selective suppression and/or activation of key signaling pathways with corresponding growth factors and small molecules under defined culture conditions. One of the first efforts to direct differentiation, with the aim of producing cardiomyocytes, was carried out by coculture of HES-2 hESCs with visceral-endoderm-like cells (i.e., mouse END-2 cells), which generated substantial numbers of ventricular-like cells (Mummery et al., 2003). Coculture of monolayer ESCs with primary hepatocytes produced homogeneous definitive endoderm-like cells, which were converted to large amounts of endocrine cells (~70%) on Matrigel under retinoid induction and hedgehog inhibition. Further, coculturing the endocrine cells with endothelial cells produced approximately 60% Insulin 1 expressing cells (Banerjee et al., 2011). However, the exact mechanism underlying pancreatic cell maturation under these coculture conditions is not clear. It is possible that adjacent heterogeneous cells during embryonic development may provide suitable environmental cues to modulate differentiation efficiency.

Indeed, hepatic morphogenesis depends on signal interactions between endodermal (epithelial), mesenchymal, and endothelial progenitors. To recapitulate early liver organogenesis, Takebe et al. generated hepatic endodermal cells from human iPSCs (iPSC-HEs) by directed differentiation (Takebe et al., 2013). These endodermal cells were cocultured with stromal cells, human umbilical vein endothelial cells, and human mesenchymal stem cells. These human iPSC-HEs were able to self-organize into functional 3D liver buds (Takebe et al., 2013). This study provides a developmental basis for establishing efficient 3D-coculture protocols.

To closely mimic the in vivo 3D ECM microenvironment for cell maturation, Christman and colleagues developed a simple method to generate tissue-specific extracellular matrix (ECM) coatings by decellularizing skeletal and cardiac tissues (DeQuach et al., 2010). These decellularized matrices have been shown to facilitate the maturation of committed skeletal myoblast progenitors and hESC-derived cardiomyocytes in 3D culture (DeQuach et al., 2010). Thus, decellularized organ scaffolds provide structural cues and ECM components for 3D tissue morphogenesis. No doubt, these native scaffolds could be replaced by compatible biomaterials in the future.

To generate various neuroectodermal tissues, Sasai, Knoblich, and their colleagues have demonstrated the feasibility to derive complicated tissue patterns (termed cerebral organoids) through a process involving dynamic patterning and structure self-organization in 3D hPSC culture (Eiraku et al., 2008; Lancaster et al., 2013; Nakano et al., 2012). By this approach, they have generated distinct cortical neurons (Eiraku et al., 2008), a 3D optic cup structure of the neural retina that contains both rods and cones (Nakano et al., 2012), and various brain regions containing progenitor zone with plentiful radial glial stem cells (Lancaster et al., 2013). Thus, these studies have paved the way to mini-organogenesis that may be used for therapy of patients with neuronal disorders. The mini-organoids also offer in vitro model to understand the complexity of brain function at a culture dish. In addition, mini-organoids may have many implications in sciences and technologies beyond stem cell research (Sasai, 2013).

Clearly, multi-dimensional coculture and mini-organoids represent a powerful approach to achieve tissue and organ morphogenesis in vitro. With the development of efficient protocols for hPSC expansion, differentiation, maturation, and 3D-coculture scaffold, tissue morphogenesis and organogenesis are an emerging resource for future tissue replacement. Spatial and temporal control of the dynamics of intercellular interactions in multidimensional environments would open new era of contemporary medicine. The caveats of the above integrated approach for tissue engineering and organogenesis may be somewhat complicated at the current stage, which requires a comprehensive understanding of spatial-temporal signaling networks to engineer optimal 3D and/or 4D environmental cues that efficiently regulate stem cell fate commitments.

Current Challenges and Future Directions

Pluripotent stem cell based therapy certainly faces many challenges. As a treatment per se, safety is of course the most important issue. It may be related to the quality control of cell production under Good Manufacturing Practices (GMP) protocols, and optimization and standardization of both cell growth media and culture methods. Without proper quality control and standards, we may be confronted with the impurity of the cell resource and potential tumorigenicity of the cells during the course of cell engineering (Lee et al., 2013). In addition, production of large numbers of functionally mature cells with high purity at a reasonable cost is also an important issue. Understanding of these potential problems would provide us future directions to address these difficulties (Figure 1).

GMP-compatible protocols

GMP provides quality control to ensure that optimal procedures are commonly implemented in the manufacturing industry. It emphasizes the requirements of a manufacturing process as standards for testing the final products (Crook et al., 2007). Generation of six clinical-grade hESC lines was initially reported in 2007 (Crook et al., 2007). GMP-grade hPSCs should be distinguished from clinical-grade cells. Implementing GMP would not necessarily guarantee that the final cell products are of clinical-grade quality. However, it would be ideal to derive clinical-grade cells under GMP protocols, thus ensuring reproducible production of cells suitable for clinical application. Various protocols and growth modules could be combined with xeno-free culture methods to create clinically compatible protocols.

Optimization and standardization of growth conditions

Despite the existence of various formulations of cell culture media, now, we do have common platforms emerging as standards for guiding the formulation of the medium suitable for a precise pluripotent or cellular state. These standards are based on the understanding of core signaling pathways that sustain optimal stem cell growth and direct uniform differentiation (Hanna et al., 2010; Nichols and Smith, 2009). The potential challenges are: (i) we need to determine the long-term effects of several major media on hPSCs and their differentiated progenies; (ii) we also need to vigorously verify preclinical safety and efficacy of final cell products derived in vitro or ex vivo, and (iii) the final cell products derived from cell culture need to show anticipated safety and desired functionality in patients. Importantly, clinical information from patients would feedback to confirm or revise our culture medium formulations as well as growth modes and environmental cues. Without the above information in hand, we believe that we have only achieved certain capacity to culture and differentiate hPSCs at a preclinical level, but not yet in a clinical setting.

Economic considerations of hPSC resources and processing

The economics of cell processing and of obtaining desired cell resources are likely to have considerable impact on hPSC-based therapies. Obtaining clinically relevant numbers of functionally mature cells is a big challenge in the field of stem cell engineering. Practically, we need to seriously consider the cost at every stage from cell derivation to clinical trials. In the following sections, we will discuss some general considerations for cell expansion and scaling-up culture of differentiated cells.

General considerations for cell expansion

With regard to cell processing, any approach that reduces cost, whether in terms of cell culture modules or labor costs, will be relevant to the industrial scale production of undifferentiated cells and their differentiated progeny. For example, scale up in bioreactors may save on costly substrates but equipment may be more expensive to obtain and maintain. It is also important to note that the scale-up method should not deviate significantly from that used in the laboratories as culture methods influence the differentiation potential of the cells. Additionally, time taken to reach target numbers, incubator space and any special parameters required (such as low oxygen) should also be considered. It is also important to consider media and culture optimization to improve time required to reach target cell numbers. In terms of cost, it would seem likely that a synthetic product may be easier to produce compared with the purification of a complicated macromolecule (e.g., growth factors). This is relevant both in terms of small molecules versus growth factors or a synthetic substrate versus a purified ECM component.

Concerning scaling-up cultures with differentiation protocols

Manufacturing of large numbers of functionally differentiated cell types with homogeneous cellular states (at reasonable cost) represents a limiting step to both the research and therapeutic use of hPSCs. However, the mature cells are usually difficult to amplify due to low proliferation rates. If the desired terminally differentiated cell type has a corresponding proliferative intermediate, then it may not be necessary to scale up the pluripotent population and the onus shifts to scale up the differentiated precursors. However, many desired cell types, such as osteoblasts and chondroblasts do not have a suitable proliferative intermediate. In this case, we may need to start with a large number of hPSCs and then couple these pluripotent cells with differentiation protocols when a desired therapy is scheduled. We believe that linking cell expansion strategies to the proliferation features of differentiated cells would have a significant impact on saving unnecessary costs.

Conclusions

Various methods based on colony culture, non-colony types of cell growth, and aggregated suspension culture can be used to culture hPSCs. These methods maintain the epithelial characteristics of hPSCs under defined feeder, feeder-free, and xeno-free conditions. Many novel culture protocols appear to be robust and scalable, making them potential candidates to generate clinical-grade hPSCs and their derivative tissues to serve the purposes of regenerative medicine. With further modifications of the existing methods, efficient production of clinically relevant quantities of hPSCs could be attainable for both stem cell-based therapies and high-throughput drug discovery.