Yoji
Tabata
ab,
Ikki
Horiguchi
a,
Matthias P.
Lutolf
b and
Yasuyuki
Sakai
*a
aLaboratory of Organs and Biosystems Engineering, Institute of Industrial Science, The University of Tokyo, 4-6-1 Komaba Meguro-ku, Tokyo, Japan. E-mail: sakaiyas@iis.u-tokyo.ac.jp; Fax: (+81)0354526353; Tel: (+81)0354526352
bLaboratory of Stem Cell Bioengineering, Institute of Bioengineering, Ecole Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne, Switzerland. E-mail: matthias.lutolf@epfl.ch; Fax: (+41)0216939665; Tel: (+41)0216931876
First published on 1st October 2013
Pluripotent stem cells hold great promise for many pharmaceutical and therapeutic applications. However, the lack of scalable methodologies to expand these cells to clinically relevant numbers is a major roadblock in realizing their full potential. To address this problem, we report here a scalable approach for the expansion of pluripotent stem cells within bioactive hydrogel capsules in stirred bioreactors. To achieve rapid crosslinking of cellular microenvironments with tuneable, cell-instructive functionality, we combined calcium-mediated alginate (CaAlg) complexation with crosslinking of poly(ethylene glycol) (PEG) macromers via a Michael-type addition. The resulting hybrid networks have been shown to have very good handling properties and can be readily decorated with biologically active signals such as integrin ligands or Cadherin-based motifs to influence the fate of mouse induced pluripotent stem (iPS) cells. Air-driven co-axial extrusion was used to reproducibly generate gel microcapsules in high-throughput. Furthermore, the gel capsules can be enveloped in a poly(L-lysine) shell to control swelling or molecular permeability independently of the gel composition. iPS cells entrapped within such capsules expanded with limited commitment to the endodermal lineage. Functionalization of gels with an appropriate density of Arg-Gly-Asp (RGD) ligands further increased the iPS cell expansion rate and reduced the spontaneous differentiation. Therefore, the combination of micro-scale instruction of cell fate by an engineered microenvironment and macro-scale cell manipulation in bioreactors opens up exciting opportunities for stem cell-based applications.
In vivo, stem cell fate is strongly influenced by various signals from their microenvironment, also termed niche.14 For example, stem cells are often anchored to their niche via adhesion molecules provided by various niche cells and the ECM.15 However, current stirred bioreactor culture systems lack the possibility of exposing stem cells to key niche signals which often results in limited control of stem cell fate. To tackle this problem, we propose to integrate biomimetic mechanisms of stem cell fate control into macro-scale cell culture settings.
During the last decade, a wide variety of synthetic hydrogels have been engineered as artificial niches. In contrast to naturally derived hydrogels, synthetic systems overcome the risk of pathogenic contamination and show a reduced batch-to-batch variability.16 Importantly, synthetic hydrogels such as those based on branched PEG are highly tuneable in terms of their physicochemical and biochemical properties, and as a consequence are increasingly used for in vitro stem cell culture.17 Such gels can be formed under cytocompatible conditions by choosing self-selective cross-linking chemistries.18–20 Moreover, cell-adhesive or proteolytically sensitive domains as well as desired signalling factors can be readily incorporated to rationally control the behaviour of gel-entrapped cells.19–22 On a practical level, a wide variety of PEG-based gel building blocks, such as maleimide-functionalized 4arm-PEG (4MA-PEG),23 are commercially available at reasonable cost which makes these hydrogels particularly attractive for stem cell culture.
Despite the considerable interest in PEG-based hydrogels for stem cell culture, these materials are usually utilized in formats that are not well suited for industrial-scale processes. One notable example is the recent manufacture, in a co-axial extrusion system, of hybrid hydrogel microspheres composed of PEG and calcium alginate (CaAlg).24 These materials are highly promising for stem cell manipulation in bioreactor culture, but, to the best of our knowledge, they have not yet been used for the encapsulation of mammalian cells.
Here we adapted and further developed this method for the microencapsulation and expansion culture of iPS cells in large-scale bioreactors. Using mouse iPS cells as the model system, we specifically demonstrate how the microenvironment can be tailored to achieve optimal expansion.
A co-axial extrusion mount (Fig. 1a,b) allowed stable generation of 2% (w/v) CaAlg capsules with diameters ranging from approximately 0.7 to 1.5 mm, depending on the N2 flow and on extrusion speed. 1.1 mm diameter droplets were utilized in the present study. Good uniformity in the shape and size was obtained with relative standard deviations below 3%. About 1000 capsules were generated in ca. 5 minutes during this first step. The instantaneous gelation of the CaAlg trapped 20 kDa 4MA-PEG macromers (Fig. 1c) in the alginate network. Placing these gels in a solution containing di-thiolated linear PEG (2S-PEG; mol. weight 3.4 kDa) (Fig. 1d) allowed PEG crosslinking via a Michael addition reaction (Fig. 1e), resulting in hybrid capsules composed of 2% (w/v) CaAlg/2% (w/v) PEG.
Hydrogels are known to have variable volume depending on both their molecular composition and the medium in which they are incubated.25 CaAlg and CaAlg/PEG hybrid capsules suspended in culture medium exhibited limited swelling behaviour, not exceeding volumetric swelling degrees of 50% (Fig. 2a). However, after CaAlg removal by ethylenediaminetetraacetic acid (EDTA) their volume dramatically increased, surpassing 200% swelling. Pure 2% (w/v) PEG, obtained by removing the CaAlg gel phase, was robust enough to keep a round shape even after the CaAlg removal. Conversely, covering capsules with a PLL shell (step 3) significantly restrained swelling, even for the loosest PEG gels. Excess swelling is not problematic because of gel instability but may also be associated with harmful effects on encapsulated cells, exposing them to excessive stress which could reduce viability. Therefore, our coating strategy opens up new possibilities for highly swelling, soft materials in bioreactor cultures that cannot be used otherwise. Diffusion tests on FITC-labelled dextran of controlled molecular weight revealed the expected inverse relationship of molecular weight and diffusivity (Fig. 2b). Stable diffusion was observed after six hours of stirred incubation, in which molecules smaller than 4 kDa could almost freely penetrate into raw CaAlg/PEG capsules, whereas the relative concentration of molecules larger than 20 kDa remained below 70%. PLL coating substantially affected the diffusion of larger macromolecules (>70 kDa). The selective diffusion characteristics allow, for example, capsule uptake or release of nutrients, gases or cellular waste that have small molecular weight. Conversely, it may enable the confinement of larger proteins released by cells that may enhance auto- or paracrine processes. Changing the thickness of the PLL layer had however a minimal influence on the diffusivity of dextran. As alternatives, the PLL molecular weight and concentration could be considered for further controlling the molecular cut-off of the shell.26
Next, we sought to modify the biochemical capsule properties by functionalizing PEG networks with bioactive moieties (Fig. 1f–j). Thiolated biomolecules can be directly tethered to PEG macromers before crosslinking in order to achieve a nearly quantitative incorporation efficiency.18,23 As a first bioconjugation strategy, we therefore used as a model system cysteine-containing Arg-Gly-Asp (RGD) peptide to generate integrin-binding networks (Fig. 1f,i). As a second strategy based on affinity binding, Fc-tagged molecules could be immobilized as well with an intermediate PEG linker bearing both maleimide and ProteinA (Fig. 1g).27 Fc-E-cadherin/PEG linker construct (Fig. 1g) was formed and subsequently added to the gel precursor solution to form E-cadherin-binding networks (Fig. 1j). To validate this affinity-based bioconjugation strategy, capsules were generated from a series of precursor solutions containing different amounts of ProteinA linker and an excess amount of FITC-hIgG. The fluorescence intensity gradually increased as a function of ProteinA content (Fig. 2c), demonstrating that the bioconjugation is well controllable. The signal remains stable for at least one week (data not shown). Of note, we observed that PLL-coated capsules usually showed higher fluorescence intensity, most likely because of their lower swelling-related dilution of tethered proteins. Furthermore, introducing an alternative high-affinity coupling such as Biotin/NeutAvidin offers the possibility of tethering multiple biomolecules in an orthogonal manner.28
To test whether capsule bioconjugation could promote iPS cell expansion and maintenance of pluripotency, capsule formulations containing RGD and E-cadherin were used. Capsules containing 500 μM RGD increased iPS cell expansion by approximately 80% compared to inert gels; lower concentrations (e.g. 100 μM) did not show any significant effect. Furthermore, incorporation of 10 μg mL−1 E-cadherin did not induce any significant change on expansion (Fig. 3b). In order to assess proliferation in a more refined manner, glucose consumption was quantified every two days during the entire culture period. For each group, total glucose consumption gradually increased over time until almost complete depletion of the medium glucose at day 8 (Fig. 3k). Lactose production was monitored in the same way in order to evaluate the lactose to glucose ratio. Lactose to glucose ratios remained between 0.8 and 1.2 (data not shown) over the culture period. These results suggest that both aerobic and anaerobic metabolism were equally present, since the ratio is close to zero when aerobic metabolism is favoured and close to two when anaerobic metabolism dominates.29
To probe iPS cell fate changes in situ, a reporter cell line was used in which green fluorescent protein (GFP) is expressed under the control of the Nanog promoter.30 In each capsule type, the majority of colonies exhibited stable and homogeneous Nanog expression during the entire culture period (Fig. 3a). Additionally, quantitative expression of pluripotency (Nanog, Oct4, Sox2 and Rex1)31,32 as well as primitive endoderm markers (GATA4, HNF4 and AFP)33–35 was evaluated using quantitative RT-PCR (Fig. 4). Encapsulation culture generally enhanced the expression of pluripotency markers compared to conventional static culture. In particular, Rex1, known to be the most stringent pluripotency marker, was significantly up-regulated under all capsule conditions, especially in those modified with bioactive cues. Primitive endodermic GATA4 significantly decreased when cells were exposed to both 100 μM and 500 μM, while HNF4 was uniquely down-regulated in 100 μM capsule. 10 μg mL−1 E-cadherin however did not alter the expression of the latter genes, but strongly supported the expression of AFP, with a ca. 100-fold increase. These data show the possibility of exploiting bioconjugation concepts in capsule engineering in order to control pluripotency cell fate.
To further assess their potency, single cells recovered from CaAlg/PEG hybrid capsules after eight days of stirred culture were sub-cultured on either feeder layers in the presence of LIF or tissue-culture plates without LIF for three additional days (Fig. 4b,c). These experiments showed that cells had kept the ability of expanding (as evidenced by strong expression of Nanog-GFP) in conventional 2D pluripotent stem cell culture settings. Some populations nevertheless acquired an epiblast-like morphology (Fig. 4b).36 In contrast, cells cultured under non-iPS-maintaining conditions quickly lost Nanog expression and showed neuronal37 or hepatocyte-like38 morphologies (Fig. 4b). Therefore, iPS cells cultured in hydrogel capsules maintain their potency and the recovered cells can be utilised for subsequent applications.
Encapsulation cultures offer interesting advantages over other methods. For example, colonies do not clump together which is often an issue in both suspension and microcarrier culture.13,40 Secondly, cells are protected from excessive hydrodynamic stresses that may affect both cell viability40 and potency.41 Finally and perhaps most importantly, cells are embedded in 3D microenvironments that can be rationally engineered to instruct stem cell fate.17 Indeed, our study achieved similar cell expansion levels compared to other microencapsulation cultures,42 but offered the possibility of enhancing the maintenance of pluripotency by providing a favourable microenvironment. For example, incorporating 500 μM RGD allowed increasing the expansion yield by about 80% compared to inert capsules, while the pluripotency was better maintained with significantly higher expression of pluripotency markers such as Rex1 and lower expression of primitive endoderm markers such as GATA4. The RGD domain is present in many natural ECM components and culture substrates including laminin, vitronectin or matrigel that are commonly employed to maintain pluripotent stem cells.1,43,44 Indeed, the proliferative-enhancing effect of RGD in 3D encapsulation has been reported in other studies.45 Conversely, E-cadherin did not support cell proliferation despite its important expression in early embryonic development.46 Several research groups reported that preventing pluripotent stem cells from E-cadherin-mediated aggregation by exposing them to competitive E-cadherin ligands47 or by genetic knock-down,48 supports self-renewal and allows cells to grow at a single cell level, increasing access to nutrients, gases or other signals as well as to enough space for efficient proliferation. We believe that in a physically confined 3D context, E-cadherin might not exert its positive effect on pluripotency.
Mouse pluripotent stem cells are routinely cultured with LIF supplement. LIF/STAT signalling inhibits differentiation proceeding from the primitive to visceral endoderm, which is known to be the primary commitment of suspended pluripotent spheroids.49,50 In the present encapsulation culture, the primitive endoderm makers GATA4 and vHNF4 were notably suppressed, suggesting that spontaneous endodermic differentiation might be controlled at an earlier stage than LIF signalling intervenes. Indeed, several studies suggested that stem cell fate decision might be influenced by the visco-elasticity and cell-adhesive characteristics of a substrate.51,52 These factors could have played a role in our 3D culture system. Moreover, the enclosed environment obtained by the PLL-coating might play a role in confining and concentrating certain para- and/or autocrine factors. For example, pluripotent stem cells cultured in microchambers favouring the accumulation of factors produced by cells themselves such as BMP4 showed an increased expression of pluripotency markers.53 The dimension and shape of multicellular spheroids are also known to play a role in stem cell fate decision54 by prompting heterogenization of the environment within the spheroid, in terms of cell–cell or cell–matrix interaction molecules55 as well as oxygen, nutrient or protein distribution.56 It is noteworthy that the colonies we obtain in 3D gels have a flattened, oval or disk-like shape. This unusual colony shape may display different profiles of factors mentioned above, compared to spherical spheroids such as embryoid bodies (EB) that are readily committed.57,58 Colony development and commitment are highly dependent on the stiffness of the substrate. Our hybrid gels supporting iPS cell expansion have an estimated initial Young's modulus, largely determined by the CaAlg gel phase, that surpasses 10 kPa. The stiffness of the gels can nevertheless be readily modulated by the alginate content, allowing further optimization of the substrate.
Undifferentiated miPS cells were isolated from SNL feeder cells and re-suspended in a gel precursor solution with a cell density adjusted to 5 × 105 cells mL−1. The cell suspension was gently homogenized with a pipette and loaded into an air-driven syringe pump droplet generator, in which both the extrusion and air flow rates were adjustable. A 0.2 μm filter was integrated into the N2 flow path. We used here 100 μL min−1 and 2 L min−1 respectively. Diameters of the outer (gas) and inner (gel solution) outlets were 0.260 and 1.194 mm respectively. The mixture was dropwise dispensed into the first gelation bath containing 150 mM CaCl2 (Wako) and 0.01% polyoxyethylene (20) sorbitan monolaurate (Tween20) (Wako) at room temperature. All solutions were sterile filtered. Instruments, including the dispensing nozzle and tubing, were autoclaved before each use. Extrusion was performed under the laminar flow cabinet. Spherical CaAlg capsules were instantaneously formed upon contact of the droplet with the bath solution via ionotropic gelation. Capsules were then immediately transferred to the second bath solution containing 3.4 kDa linear PEG dithiol (2SH-PEG) dissolved in DMEM, where they were gently stirred for 30 minutes at 37 °C. This solution also contained 30 mM CaCl2 for completing the CaAlg gelation. PEG macromers were covalently cross-linked to each other by forming stable and irreversible carbon–sulfur bonds, resulting in CaAlg/PEG hybrid capsules. The optimal cross-linking for such a system is achieved when the stoichiometric ratio of the reacting groups is close to one.18 In the actual case, because the mixing of the two PEG components was gradually performed by diffusion and thanks to the high reaction velocity, excessive amount, but slightly concentrated 2S-PEG solution was employed. Especially, the maleimide to thiol molar ratio was kept at 1:0.5 while the total amount ratio was at 1:2.5. Capsules were washed twice with PBS and subsequently incubated in a 0.05% (w/v) 15–30 kDa PLL hydrobromide (Sigma) solution for 10 minutes and then in a 0.075% (w/v) NaAlg solution for 5 minutes for covering up the capsules with a PLL shell. Capsules were optionally treated either with 10 mM EDTA (Sigma) for 10 minutes or with 100 U mL−1 alginate lyase (Sigma) for a few hours to dissolve CaAlg gel resulting in pure PEG capsules. Capsules were finally washed twice with DMEM and transferred into iPS-maintaining medium supplied with 1000 U mL−1 LIF. Encapsulated cells were cultured for eight days in an ultra-low attachment 6-well plate (Corning) placed on an orbital shaker stirring at 120 rpm.
Capsule permeability to dissolved molecules was analysed by studying the diffusion of fluorescent molecules. Raw and PLL-coated (20 or 40 μm-thick) capsules were generated and stored in PBS for two days at 37 °C until swelling was completed. A thicker PLL layer was obtained by incubating capsules for a longer time in the PLL solution. Capsules of each type were then incubated in 120 rpm stirred culture medium containing 0.5 mL mg−1 of 4, 20, 70 or 150 kDa FITC-labelled dextran standards (Sigma). The fluorescence within the capsules was measured and normalized to the bulk signal, assumed constant, to estimate the capsule permeability. The fluorescence profile was almost stabilized after 6 hours incubation. 10 randomly chosen capsules were analysed using confocal microscopy.
To determine the incorporation efficiency of Fc-tagged protein via the intermediate heterobifunctional PEG linker, hIgG (Invitrogen) was labelled with FITC using a protein labelling kit (Pierce) following the manufacturers’ instructions. FITC-hIgG and 10% molar excess PEG linker were premixed and added to the gel precursor solution at different final concentrations, 0, 25, 50, 100 and 200 μg mL−1. Capsules were then generated and extensively washed with PBS to remove unreacted molecules and incubated in culture medium for 24 hours. The fluorescence profile of 10 randomly chosen capsules of each group was evaluated using confocal microscopy.
This journal is © The Royal Society of Chemistry 2014 |