SEARCH

SEARCH BY CITATION

Keywords:

  • Pluripotency;
  • Differentiation;
  • Tissue regeneration;
  • Technology;
  • Stem cell-microenvironment interactions;
  • Experimental models;
  • Embryonic stem cells

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

Developmental gradients of morphogens and the formation of boundaries guide the choices between self-renewal and differentiation in stem cells. Still, surprisingly little is known about gene expression signatures of differentiating stem cells at the boundaries between regions. We thus combined inducible gene expression with a microfluidic technology to pattern gene expression in murine embryonic stem cells. Regional depletion of the Nanog transcriptional regulator was achieved through the exposure of cells to microfluidic gradients of morphogens. In this way, we established pluripotency-differentiation boundaries between Nanog expressing cells (pluripotency zone) and Nanog suppressed cells (early differentiation zone) within the same cell population, with a gradient of Nanog expression across the individual cell colonies, to serve as a mimic of the developmental process. Using this system, we identified strong interactions between Nanog and its target genes by constructing a network with Nanog as the root and the measured levels of gene expression in each region. Gene expression patterns at the pluripotency-differentiation boundaries recreated in vitro were similar to those in the developing blastocyst. This approach to the study of cellular commitment at the boundaries between gene expression domains, a phenomenon critical for understanding of early development, has potential to benefit fundamental research of stem cells and their application in regenerative medicine. Stem Cells 2013;31:1806-1815


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

Cell fate choices in murine embryonic stem cells (mESCs) are regulated by orchestration of LIF-gp130-Stat3, BMP-TGF-β-Smad, and MAPK-ERK signals that control a network of transcription factors such as Oct4, Sox2, Nanog, and Tcf3 [1–8]. During embryonic development, the spatial distribution of these signals guides the arrangement of cells into organized structures that subsequently evolve into tissues and organs [9]. The precise molecular changes accompanying cell responses at the boundaries between regions are still not fully elucidated, due at least in part to the limited capability to analyze cells at tissue interfaces. In vitro recapitulation of gene expression domains and boundaries between differentiating cells that are normally established in vivo would markedly advance our understanding of gene regulation in early development.

Geometrically defined multicellular patterns have been established by laser ablation and printing of proteins [10–13]. In our previous study, the differentiation of adult human mesenchymal stem cells was modulated in a microfluidic platform by doxycycline (Dox) induction of bone morphogenic protein (BMP-2), to generate an interface between osteogenic and fibrotic tissue [14]. To date, direct patterning of developmentally important gene expression at the transcriptional level has not been attempted.

To gain new insights into the processes at the pluripotency-differentiation boundaries, we investigated if the pluripotent states of mESC can be induced in a topologically specified manner by patterning the expression of Nanog, a key factor in safeguarding the pluripotent state [15, 16]. To this end, we used microfluidic gradients of morphogens in conjunction with an inducible lentivirus-based genetic complementation system we previously established [17, 18]. In this lentiviral system, a constitutively active short hairpin RNA (shRNA) depletes endogenous Nanog mRNA, and normal levels of Nanog can be restored in a Dox-dependent manner from a shRNA “immune” mRNA. In this way, in the absence of Dox, both the rescue mRNA and the green fluorescent protein (GFP) are not expressed, whereas in the presence of Dox the exogenous shRNA-immune mRNA is expressed as well as the GFP reporter.

Different Nanog levels in the individual mESCs define a Nanog-high population that is resistant to differentiation signals and a Nanog-low population that, while pluripotent, is sensitive to differentiation-inducing signals [19]. We reasoned that patterning Nanog expression in a laminar microfluidic system could modulate cell self-renewal and early differentiation commitment in a spatially defined manner (Fig. 1).

image

Figure 1. Patterning of pluripotency and differentiation in murine embryonic stem cells (mESCs) by spatial regulation of gene expression. (A): In vivo patterning of gene expression during embryonic development, where distinct pluripotency and differentiation domains are defined by the expression of Nanog. (B): Microfluidic model for patterning gene expression in an initially uniform population of mESCs. Distinct regions of mESC pluripotency and differentiation are established by patterning the expression of the Nanog transcriptional regulator through the exposure of the cells to gradients of chemicals or morphogens. (C): Tet-on system for the doxycycline (Dox)-dependent induction of Nanog. In the presence of Dox, the exogenous Nanog gene is expressed to maintain pluripotency.

Download figure to PowerPoint

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

Embryonic Stem Cell Culture and Differentiation Assays

A murine ESC line with controllable Nanog expression (NanogR) was constructed and characterized as in our previous studies [16, 17]. Briefly, NanogR cells contain two expression cassettes, a constitutively active shRNA designed to deplete the Nanog gene product and an inducible shRNA-immune expression system that rescues its depletion in order to evaluate and confirm the target-specific effect of the shRNA. This engineered lentivirus based-rescue vector contains a human H1 promoter-driven shRNA targeting the 3′ untranslated region (UTR) of Nanog and a TRE promoter driving and exogenous shRNA-immune rescue mRNA and GFP, which is translated via an IRES element. Since the shRNA targets the 3′ UTR of Nanog, the unmodified coding sequence of Nanog under the TRE promoter is suitable for rescuing the shRNA-mediated effect. This system requires an ESC line, such as the Aniv15 line expressing the TetOn rt TA transactivator [20].

The NanogR cells were maintained on primary mouse irradiated embryonic fibroblasts (MEFs) in Dulbecco's modified Eagle's medium (Gibco, Life Technologies, NY, USA; http://www.lifetechnologies.com/us/en/home.html) supplemented with 15% Fetal Bovine Serum (FBS) (Hyclone, Thermo Scientific, USA); http://www.thermoscientific.com), 100 mM MEM nonessential amino acids, 0.1 mM 2-mercaptoethanol, 1 mM l-glutamine (Invitrogen, Carlsbad, CA), 100 U/ml penicillin, 100 mg/ml streptomycin, 1 mM sodium pyruvate, and 103 units/ml of leukemia induction factor (LIF) (Chemicon, Millipore, Massachusetts, USA; http://www.millipore.com/). For experimental assays, NanogR cells were cultured on 0.1% gelatin-coated tissue culture plates without MEFs. To induce differentiation doxycycline (Dox, 1 µg/ml) was withdrawn, and cells were maintained in LIF containing, LIF free, or LIF free plus retinoic acid (RA) (5 µM) media. Cell cultures were maintained at 37°C/5% CO2.

Microfluidic System

The flow system consisted of a flow chamber, a syringe pump, and a tubing loop. The flow chamber (Fig. 3) consisted of a 1-mm thick borosilicate glass plate attached to the bottom of a groove within a Plexiglas block, overlayed with another glass slide with preseeded cells. The two plates were held together with screws through a Plexiglas lid, and separated by a uniform thickness rectangular silastic spacer (McMaster-Carr Elmhurst, IL), creating a flow chamber that was 5.2-cm long, 1.2-cm wide, and 154-µm high. Fluid enters and exits the chamber from two ports connected to cylindrical reservoirs at the inlet and outlet of the Plexiglas block (Fig. 3). Plastic tubing (soft high-temperature silicone rubber tubing, 1/16′′ ID, 1/8′′ OD, McMaster-Carr Elmhurst, IL) connects the entrance port to a digital syringe pump and the exit port to waste collection bottles.

Prior to the experiment, the flow chamber and tubing were assembled, connected to a syringe pump, and primed with proper culture medium. The flow chamber and tubing were sterilized by autoclaving. The sterile plastic slide was coated with 0.1% gelatin for 4 hours, and air-dried. NanogR cells were seeded onto the slide at a density of 5,000 cells per square centimeter in LIF and Dox (500 ng/ml) supplemented culture medium and cultured overnight in a humidified 37°C/5% CO2 incubator. The system maintained two parallel laminar microfluidic streams, each at a flow rate of 4 µl/minute (shear stress at the cell culture surface τ = 2.8 mPa, Reynolds number Re = 0.022). To generate patterned pluripotent zones, NanogR cells were exposed to laminar flow for 3 days. In dye separation experiments, cells were subjected to 30–60 minutes of medium flow supplemented with 2 µM Calcein-AM, and then to 10–30 minutes of flow of complete medium.

Alkaline Phosphatase Staining

Alkaline phosphatase (AP) activity was measured using a commercial detection kit (Sigma-Aldrich, St. Louis, MO). The cells were fixed in 4% paraformaldehyde in 1× Phosphate Buffered Saline (PBS) for 2 minutes. After a wash with mixture of Tris-Buffered Saline and Tween 20 (TBST) (20 mM Tris-HCl, pH 7.4, 0.15 M NaCl, and 0.05% Tween-20), cells were immersed into alkaline-dye mixture consisting of fast red violate, Naphthol AS-BI phosphate solution, and water in 2:1:1 ratio, for 20 minutes in the dark (or 7 minutes when staining experimental samples for flow cytometry), and were again washed in TBST. AP staining images were captured by microscopy (Olympus IX81, Center Valley/PA).

qRT-PCR

For gene expression analyses, total RNA was TRIzol-extracted (Invitrogen, Carlsbad, CA) and column-purified with the RNeasy kit (Qiagen, Hilden, Germany). Total RNA (1 µg) was reverse transcribed using the high capacity reverse transcription kit (Applied Biosystems, Foster City, CA). All quantitative PCR analyses were performed using the Fast SYBR Green Master Mix (Applied Biosystems, Foster City, CA) following all from (Gibco, Life Technologies, NY, USA; http://www.lifetechnologies.com/us/en/home.html) the manufacturer's protocol on a LightCycler480 Real-Time PCR System (Roche, Applied Science Penzberg, Germany). All expression levels were normalized to β-actin. Gene-specific primers used for amplification are listed in Supporting Information Table S1.

Immunofluorescence Staining

Cells cultured at different Dox concentrations and at the respective days 0, 1, 3, and 5 were rinsed twice with 1× PBS, fixed with 4% paraformaldehyde in 1× PBS for 20 minutes at room temperature, quenched with 0.2 M glycine in 1× PBS for 10 minutes, and permeabilized with 1× PBS containing 0.5% Triton X-100 for 30 minutes. Nonspecific binding sites were blocked by incubation with 1× PBS containing 0.1% Bovine Serum Albumin (BSA) for 30 minutes at room temperature. Cells were then incubated with specific primary antibodies: 1:100 anti-mouse Oct4 (Santa Cruz, Biotechnology Dallas, TX, cat no. sc-8628), or 1:150 anti-mouse Nanog (Cosmo Bio, Carlsbad, CA, cat no. REC-RCAB0002P-F), diluted in 1× PBS containing 0.2% Triton X-100 and 0.1% of BSA for overnight incubation at 4°C. After that, cells were incubated with one of the following secondary antibodies: Alexa Fluor 546 donkey anti-goat (Life Technologies, Carlsbad, CA, cat no. A-11056) (1:1,000 dilution) and Alexa Fluor 488 donkey anti-rabbit IgG (Life Technologies, Carlsbad, CA, cat no. A-21206) (1:1,000 dilution). After washing with 1× PBS containing 0.2% Triton X-100 for three times, the samples were incubated for 5 minutes with 1:1,000 DAPI in water, followed by three washes with 0.2% Triton X-100 in 1× PBS. We confirmed the complete loss of GFP 3 days after fixation (Supporting Information Fig. S9), which was necessary to avoid GFP signal contamination in the Nanog immunofluorescence staining. Therefore, the staining was completed and fluorescence images were captured 3 days after fixation with a Nikon Eclipse TE 2000-U (Melville, NY) or Olympus BX41 fluorescent microscope (Center Valley, PA).

Mathematical Model

The chamber geometry permits the wall shear stress to be directly calculated as a function of the laminar flow rate and fluid viscosity:

  • display math(1)
  • display math(2)

where µ is the medium viscosity; ρ is the medium density; Q is flow rate; w and h are the width and height of the chamber. The hydrodynamic development length L in a channel of height h can be estimated as follows [21]:

  • display math(3)

The concentration and velocity profiles were modeled using the finite element method (COMSOL Multiphysics v3.2, Burlington, MA). The concentration gradient driven by the syringe pump was based on the three-dimensional incompressible flow Navier-Stokes equations:

  • display math(4)
  • display math(5)

and the convection-diffusion transport equation

  • display math(6)

where ρ, u, and p are the density, flow velocity and pressure of medium and C and D are the concentration and diffusivity of the molecules in the medium. At each time step, the flow velocity was first obtained independently of the concentration, which was then calculated based on fluid flow field. For the convection-diffusion equation, a constant concentration was set at the channel inlet, and no-flux conditions were imposed at the channel walls. The boundary conditions were as follows: (a) specified flow velocity at the inlet, (b) zero normal stress at the outlet, and (c) no-slip conditions on the channel walls.

Image Analysis

For quantifying the intensity of Nanog and Oct4 expression in the fluidic chamber after the exposure to different differentiation media (Fig. 3), we divided the flow chamber image (12 mm in width) into 24 equal segments (each 0.5-mm wide) and quantified the staining intensity for each segment I(i). Normalized intensity values were generated as a ratio of I(i) over the smallest intensity value in the series. N(i) = I(i)/min(I(i)). In this way, we determined how uniform was the distribution of the staining intensity across the chamber, and compared different rounds of experiments.

Sequential images were acquired using a phase-inverted microscope (Olympus IX81, Center Valley, PA), stored, assembled, and analyzed digitally by Photoshop CS4 program (Adobe, Mountain View, CA). For quantitative analysis of immunofluorescent staining, the assembled color images were converted to black-white images. Integrated staining intensities were quantified using Matlab R2007 software (Matlab, Natick/MA) by normalized integration of segmented images (0.5 mm), with intensity moving across the y-axis (the width) of the chamber.

Dynamic Array Analysis

For qPCR analysis, 24 samples of 10 cells each from the NanogR rescue clone harvested from each of the three regions of interest: right (R), middle (boundary) (M), and left (L) were sorted directly into a mixture of CellsDirect 2× Reaction Mix (component of CellsDirect One-Step qRT PCR Kits, Invitrogen, Carlsbad, CA), 0.2× TaqMan Assay Mix (Applied Biosystems, Foster City, CA), and SuperScript III RT/Platinum Taq Mix (Invitrogen, Carlsbad, CA). Reverse transcription (RT) and specific target amplification (STA) were performed sequentially. RT was performed for 15 minutes at 50°C. Samples were then incubated for 2 minutes at 95°C to inactivate the RT enzyme and activate Taq polymerase. Thermal cycling conditions for STA were 18 cycles of 95°C for 15 seconds and 60°C for 4 minutes. Preamplified cDNA was diluted with Tris-EDTA (TE) buffer (1:5) for real-time PCR using BioMark 48-48 Dynamic Arrays (Fluidigm, San Francisco, CA) [22]. The PCR profile included a 10 minutes, 95°C hot-start to activate the Taq polymerase, followed by 40 cycles of a two-step program: 15 seconds at 95°C (denaturation) and 60 seconds at 60°C (annealing and extension). Data were analyzed using BioMark Real-Time PCR Analysis Software v2.0 and previously established methods [22].

Data Processing

All real-time PCR values were normalized to the endogenous control by subtracting, for each sample of n = 10 sorted cells, the value of the GAPDH expression level. ΔCt values are represented in Figure 6. Additionally, relative values represented in Supporting Information Figure S8 were obtained using the perfusion medium R (LIF,Dox)/L(LIF,Dox) as reference sample.

Network

Nanog ChIP-seq targets were obtained from published data [23–28]. The network was built in the yEd Graph editor software http://www.yworks.com/en/products_yed_about.html.

Statistical Analysis

Experimental results were expressed as average ± SD of the measured and calculated values. Data were analyzed by multiway ANOVA for multiple comparisons, followed by Tukey's post hoc analysis, with the confidence level set to 0.05 for all tests, using Statistic software. p < .05 was considered significant in all cases. For gene profiling, comparisons between experimental groups were performed using permutation analysis for differential expression [29].

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

Regulation of Nanog Expression and Pluripotency by Dox, LIF, and RA

The induction parameters of Nanog in the rescue clone NanogR were characterized for a range of Dox concentrations (0–1,000 ng/ml) in the presence of the differentiation factors LIF and RA, using LIF-containing medium, LIF-free medium, and LIF-free medium supplemented with RA (Fig. 2; Supporting Information Figs. S1--S3). Specifically, following Dox titration studies and harvesting the Nanog rescue clone at three different time points (Day 1, Day 3, and Day 5), we performed three runs of differentiation: (a) titration of Dox concentration in NanogR cells in ESC media with LIF (Supporting Information Fig. S1); (b) titration of Dox concentration in NanogR cells in ESC media without LIF (Supporting Information Fig. S2); and (c) titration of Dox concentration in NanogR cells in ESC media without LIF and with RA (Supporting Information Fig. S3).

Figure 2 shows that 100 ng/ml of Dox during 3 days in media without LIF and with RA is the best condition for detecting if the pluripotency state is compromised and early cell fate commitment is occurring. Since our main focus was to study the patterns of gene expression at the boundaries where cells start their commitment to differentiation pathways, we selected day 3 as the most adequate time point for capturing the cell fate transitions in Fluidigm experiments.

image

Figure 2. Regulation of gene expression in murine embryonic stem cells by exposure to different levels of Dox in the presence of LIF and RA. (A): Alkaline phosphatase staining of NanogR cells. (B): Immunofluorescent staining of NanogR cells (blue: DAPI; green: Nanog; red: Oct4). (C): Expression levels of Nanog, Oct4, Sox2, Esrrb, Fgf5, Gata4, Gata6, and Foxa2, measured by RT-PCR. *, p < .05 compares the corresponding conditions between two analytic days at the same Dox concentration. +, p < .05 compares the corresponding condition with cultures in the absence of Dox and at the same analytic day. Data are shown for day 3 of cultivation in medium containing LIF (+LIF) as control, medium without LIF (−LIF), or without LIF and including retinoic acid (RA) (−LIF+RA). Scale bar = 100 µm. Abbreviations: LIF, leukemia induction factor; RA, retinoic acid.

Download figure to PowerPoint

image

Figure 3. Microfluidic chamber. (A): Flow chamber and its components (main chamber: 52 mm × 12 mm × 0.154 mm). (B): Modeling of velocity and concentration profiles. Using COMSOL Multiphysics, the velocity and concentration profiles were defined for the flow rate of 4 µl/minute per channel (with the Reynolds number of Re = 0.022 and the hydrodynamic development length of 48 µm, corresponding to a fully developed laminar flow). The wall shear stress in the cell culture zone was 2.8 mPa for the fluidic chamber, well under the shear stress values of cells exposed to normal interstitial [31] or blood flow [32]. Steady-state profiles of flow velocity and concentrations of Dox, leukemia induction factor (LIF), and retinoic acid (RA) are shown for the input concentrations of 500 µg/ml Dox = 1.13 × 10−6 M; 103 units LIF/ml = 2.13 × 10−10 M, and 5 µM RA. Diffusion coefficients: DDox = 3.93 × 10−6 cm2/second, DLIF = 1.66 × 10−6 cm2/second, DRA = 2.3 × 10−6 cm2/second; flow viscosity (ν = 10−3 Pa second); fluid density (ρ = 103 kg/m3).

Download figure to PowerPoint

Exposure of cells to [mt]100 ng/ml of Dox resulted in expression of exogenous shRNA immune Nanog mRNA, in spite of the constitutive suppression of the endogenous shRNA-targeted Nanog mRNA. Intermediate concentrations of Dox (100–1,000 ng/ml) induced a graded expression of Nanog, resulting in varying degrees of mESC pluripotency states.

Exposure of cells to LIF at low Nanog levels did not induce cell differentiation, and the expression of Oct4 was sustained even when Nanog expression was completely suppressed by the removal of Dox (Fig. 2; Supporting Information Fig. S1). Therefore, in the presence of LIF, downregulation of Nanog alone was insufficient to induce differentiation, suggesting the existence of compensatory mechanisms maintaining the pluripotent state. In the absence of LIF (Fig. 2; Supporting Information Fig. S2), we observed Dox-concentration-dependent decreases in AP activity and Oct4 expression, and increase in the expression of Fgf5, indicative of ectodermal differentiation [30].

In the presence of RA without LIF, a stronger differentiation inducer, we observed robust differentiation, evidenced by the loss of AP activity and Oct4 expression (Fig. 2; Supporting Information Fig. S3). Concomitant expression of the endoderm marker, Gata4, and a lack of expression of Fgf5 suggest RA-mediated endodermal differentiation and suppression of ectodermal differentiation.

Computational Modeling of Concentration and Velocity Profiles

The flow field and the concentration profiles of Dox, LIF, and RA were analyzed in detail with the aid of mathematical modeling and COMSOL Multiphysics software (Fig. 3). The flow rates of 4 µl/minute in each channel correspond to the Reynolds number Re = 0.022 and the hydrodynamic length of 0.048 mm at which a fully developed flow is established (corresponding to ∼0.1% of the total length of 52 mm). Thus, the flow of medium over the cells could be considered as fully developed, with negligible variation in the direction transverse to the flow. The wall shear stress acting at the cells was estimated to 2.8 mPa, a level well under the physiological values of shear stress associated with the interstitial flow of fluids in tissues [31] and the flow of blood [32].

We developed a simple fluidic model to study the profiles of flow and morphogen concentrations at steady state. Different concentration profiles of Dox, LIF, and RA were modeled, to show that the bilaminar-fluidic system permits alignment of flow with different concentration profiles, and the creation of a concentration interface (Fig. 3B). The model predictions of the concentration profiles were verified in experiments with Calcein-AM (diffusion coefficient of Calcein-AM is DCalcein = 2.6 × 10−6 cm2/second) that was used as a model molecule for flow studies at the inlet concentrations of 2 µM and 0, and the flow rates of 4 µl/minute per channel. The basic modeling approaches were those developed in our previous studies [14].

Patterning of Embryonic Pluripotency by Microfluidic Gradients of Dox, LIF, and RA

In microfluidic flow experiments, we generated patterned zones of mESC pluripotency and differentiation, by applying gradients of Dox, LIF, and RA across the flow chamber. The streams of culture medium with supplements selectively inducing cell pluripotency or differentiation were introduced at the right (R) and left (L) of the microfluidic chamber (Figs. 1B, 3), to establish the following four experimental groups:

Group 1: Control (no gradient): R(LIF + Dox)/L(LIF + Dox)

Group 2: Gradient of Dox: R(LIF + Dox)/L(LIF)

Group 3: Gradients of Dox and LIF: R(LIF + Dox)/L(none)

Group 4: Gradients of Dox and LIF and RA: R(LIF + Dox)/L(RA)

Uniform concentrations of Dox and LIF in the flow chamber (control, Fig. 4 first column) elicited cellular responses consistent with those observed in the corresponding static cultures. Nanog and Oct4 expression levels and the staining for AP, Nanog, Oct4, and cell nuclei were spatially uniform across the chamber and without apparent effects of flow (Fig. 4; Supporting Information Fig. S4).

image

Figure 4. Patterning of murine embryonic stem cell pluripotency by microfluidic gradients of Dox, LIF, and RA. (A): Experimental design: cultured NanogR cells were exposed to concentration gradients of Dox, LIF, and RA by two microfluidic streams: Right and Left (R/L) containing (a) R(LIF+Dox)/L(LIF+Dox), (b) R(LIF+Dox)/L(LIF), (c) R(LIF+Dox)/L(none), and (d) R(LIF+Dox)/L(RA). Concentration profiles calculated from mathematical models are shown for each of the four conditions. (B): Normalized intensities of Nanog/Oct4 immunostaining fluorescence across the flow chamber (n = 4–5). (C): Staining for AP, Nanog, and Oct4 at the middle section of the flow chamber. Arrows represent the flow direction: from top to bottom in the respective panels (C). Scale bars = 1 mm (C). Abbreviations: AP, alkaline phosphatase; LIF, leukemia induction factor; RA, retinoic acid.

Download figure to PowerPoint

Cellular responses to a gradient of Dox only (Fig. 4A, second column; Supporting Information Fig. S5) or to the combination of Dox and LIF gradients (Fig. 4A, third column; Supporting Information Fig. S6) were comparable. In both cases, patterned Nanog expression was observed within the same population of cells, along the gradient of Dox or Dox/LIF, and with spatially uniform expression levels of Oct4. Cells exhibited within a short time period patterned AP staining along the Dox or Dox/LIF concentration profiles. Most interestingly, high-magnification fluorescent images demonstrated patterned Nanog expression within the individual cell colonies (Supporting Information Fig. S7, third and fourth column). LIF gradient had no apparent effect on cellular responses (Supporting Information Figs. S5--S7). Nanog suppression, with or without LIF, affected AP activity but did not affect expression of Oct4. Suppression of Nanog and removal of LIF were thus not sufficient to induce significant mESC transition from pluripotency to differentiation.

Clearly defined zones of pluripotency and differentiation were established when the cells were exposed to the same-direction gradients of Dox and LIF superimposed with an opposing gradient of RA (Fig. 4A, fourth column; Supporting Information Fig. S7). In this group only, both Nanog and Oct4 levels decreased, in parallel with the decrease in Dox and LIF, and with the increase in RA. Measured transitions in Nanog and Oct4 expression between the pluripotency and differentiation zones were consistent with the concentration profiles of Dox, LIF, and RA (Fig. 4B--4C, fourth column; Supporting Information Fig. S7). The AP staining pattern was similar to that in the Dox and Dox/LIF groups, except that AP activity in the pluripotency region was markedly lower (Fig. 4C, fourth column), presumably due to RA diffusion or direct interactions between the pluripotency and differentiation regions. Parallel staining with DAPI showed that the observed differences in Nanog and Oct4 levels were not caused by nonuniform distributions of cells. In sum, downregulation of Nanog alone in cultured cells could not induce definitive lineage commitment, a finding consistent with the previous reports [15]. Instead, a combination of Nanog suppression with proper gradients of LIF and RA was necessary for robust differentiation commitment.

Notably, high magnification images revealed patterning of the expression of Nanog and Oct4 within single cell colonies located at the boundary between the pluripotency and differentiation regions (Fig. 5). Perpendicular to the boundary, we observed gradual changes from aggregated cell populations, and clear boundaries with diffuse populations of flattened cells (Fig. 5D).

image

Figure 5. Patterning of murine embryonic stem cell pluripotency within a single-cell colony. Microfluidic gradients shown in Figure 4 for the entire cell field were also established within the individual cell colonies. Panels (A--D) show the distribution of the cells (blue, cell nuclei stained by DAPI), Nanog (green), and Oct4 (red) under experimental condition: cultured NanogR cells were exposed to concentration gradients of Dox, LIF, and RA by two microfluidic streams: (Ai) top(LIF+Dox)/bottom(LIF+Dox), (B) top(LIF+Dox)/bottom(LIF), (C) top(LIF+Dox)/bottom (none), and (D) top(LIF+Dox)/bottom(RA). Fourth column, normalized intensities of DAPI/Nanog/Oct4 immunostaining fluorescence across the interface (n = 3, four selected zones from the same image). Arrows represent the flow direction: from left to right in the respective panels. Scale bars = 100 µm. Abbreviations: LIF, leukemia induction factor; RA, retinoic acid.

Download figure to PowerPoint

image

Figure 6. Spatial patterning of gene expression in murine embryonic stem cells by microfluidic gradients. (A): Scatter dot plots of expression levels for 21 genes across the microfluidic chamber (right [R], middle [boundary] [M], and left [L]), measured using the BioMark HD System for multiplex quantitative PCR. The data are ΔCt to the reference gene GAPDH. Raw Ct data for all genes assayed are in Supporting Information data. (B): Molecular signature of Nanog targets (pluripotency markers, chromatin modifiers, and differentiation markers) in the right (R), middle (boundary) (M), and left (L) regions of the chamber. The color of each node represents the average expression level for each gene at each location (R, M, and L). The arrows represent protein-DNA interactions (Supporting Information Methods). Data are shown for the group R(LIF+Dox)/L(RA) (Fig. 4, fourth column).

Download figure to PowerPoint

Gene Expression Signatures of Differentiating Stem Cells at the Pluripotency/Differentiation Boundaries

The R(LIF + Dox)/L(RA) group (Fig. 4, fourth column; Supporting Information Fig. S7) that showed clear patterns of pluripotency and differentiation was analyzed in depth for molecular signatures in the region of pluripotency, at the boundary, and in the region of differentiation (Fig. 4 - right, boundary, and left, respectively). The cells were harvested from each region as shown in Supporting Information Figure S8 to assess the expression of genes in cells at the interface between the two regions, following cell exposure to the gradients of morphogens.

A total of 21 genes from three groups: (a) pluripotency markers, (b) chromatin modifiers, and (c) differentiation markers were analyzed by microfluidic quantitative PCR (Fluidigm) [22] in cells from each region of interest. Cell exposure to microfluidic gradients resulted in significant downregulation of pluripotency mRNAs, including: Tbx3, Klf4, Esrrb, Stat3, Lin28, Bmp4, and Dppa4 along with downregulation of Nanog (Fig. 6A; Supporting Information Fig. S8B, S8C). In parallel, differentiation marker mRNAs such as: Gata4, Gata6, Foxa2, and Olig1 were upregulated in the differentiation region of the flow chamber. This is consistent with the known roles of Gata4 and Gata6 as crucial regulators of primitive endoderm [33–35].

The switch toward differentiation in the cells within the –LIF-DOX+RA region was associated with the downregulation of important chromatin modifiers: Wdr5, Smarcc1, Jarid2, and Rai, known to play important roles in maintaining the pluripotency. Our observations support the importance of these chromatin modifiers for pluripotency [2, 36–39] and suggest that their downregulation could promote expression of primitive endoderm markers. Strong interactions between Nanog and its target genes were identified by constructing a network with Nanog as the root and the measured levels of gene expression in each region (Fig. 6B).

Withdrawal of Dox downregulated Nanog and suppressed the expression of Nanog-activated targets (Nanog, Oct4, Sox2, Bmp4, Lin28, Stat3, Esrrb, Dppa4, Tbx3, Myc, Klf4, Rest, Wdr5, Smarcc1, Jarid2, and Rai). Nanog was found to bind and repress Foxa2, preventing differentiation [3, 4]. Consistent with this previous finding, we observed gradual activation of Foxa2 outside the pluripotency region.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

During morphogenesis, cells undergo massive changes in gene expression programs, with combinations of spatial and temporal gradients of morphogens coordinating tissue development. Our strategy was to pattern in vitro the expression of key master regulator genes and thereby mimic some of the in vivo mechanisms in which molecular signals orchestrate early development from multicellular patterns to increasingly complex structures. The mechanistic relationships and the transcriptional circuitry between developmental markers and cellular responses remain largely unknown. Transcriptional patterning in vitro could provide important insights, by spatiotemporal control of molecular signaling and by establishing boundaries between gene expression domains, such as those between the wing and the notum in Drosophila melanogaster [40], foregut and hindgut [41] and between somites [42–44] in vertebrate embryos.

Using the microfluidic device (Figs. 1, 3) we patterned multicelular structures by applying localized gradients of soluble morphogens. The application of –LIF/-DOX/+RA gradients generated a boundary between the pluripotency state and the primitive endoderm state, with significant changes in gene expression of pluripotency markers, chromatin modifiers and early differentiation markers in the cells located at different positions down the morphogen gradients.

Our study established a direct link between the gradients of morphogens, gene expression profiles, and the lineage fate of differentiating stem cells. The results we collected suggest possible restrictions in the differentiation potential of stem cells to commit to specific lineage fates as a function of position along the gradient and the type of morphogen. Our findings also indicate that the cell exposure to gradients can provide patterning of cell differentiation cues. Finally, we identified the gene expression levels of Nanog targets that initiate dynamic processes leading to cell fate decisions at the boundaries.

In principle, pluripotent ESCs can be distinguished from differentiated cells by the specialized chromatin state required to epigenetically regulate the ESC phenotype. Recent studies show that in addition to the factors specific to pluripotency, chromatin remodeling enzymes play major roles in regulating chromatin in ESCs and the capacity of ESCs to self-renew and differentiate. Here, we show the efficient downregulation of some histone modifiers like Wdr5, Rai1, Smarcc1, and Jarid2 resulting in lineage commitment. This evidence supports the importance of chromatin dynamics for early lineage decisions, in line with the previously reported roles of histone modifiers in maintaining ESCs [36–38, 45].

Moreover, in this study, we successfully show how the cells condition their gene expression programs in response to microenvironment signals. Nanog signaling circuitry is tightly regulated by feed-forward regulatory loops that maintain the pluripotent state [23–28]. The identification of the gene expression levels of Nanog targets at the boundaries enables quantitative insights into the dynamic processes involved in cell fate decisions.

An important result is that by application of the –LIF/-DOX/+RA gradient one can generate a boundary populated by the cells with low transcriptional state of Nanog that are poised for differentiation [15]. In this case, the microenvironment alone determines the fate commitment. In the absence of differentiation factors, these Nanog-low cells can be maintained in their pluripotent state. In contrast, the presentation of strong differentiation signals like RA would make these same cells commit and differentiate preferentially into visceral endoderm [46].

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

This study combined inducible gene expression with a microfluidic technology for patterning gene expression in mESCs, in order to investigate effects of gradients of morphogen on cell fate decisions. We demonstrate that the boundaries between pluripotency and differentiation domains can be established within an initially uniform population of mESCs by directly patterning of Nanog expression by applying combinations of gradients of Dox, LIF, and RA. This innovative approach uses a simple and highly controllable in vitro system to modulate cellular function at the transcriptional level and engineer discrete cellular domains mimicking those present in a developing blastocyst. Further studies of the regulatory circuitries involved in cell fate decisions during early development would provide a more comprehensive view of how cells orchestrate morphogenetic processes, and help develop a roadmap for directing stem cell fates in regenerative therapies.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

We gratefully acknowledge the funding support of this work that was provided by NIH (GM078465 to I.R.L., EB002520 and EB015888 to G.V.N.) and NYSTEM (C024410 to I.R.L., C026449 to G.V.N.). We also thank to W. Leong, D. O. Freytes, F. Pereira, S. Ghaffari, and A. Ma'ayan for scientific advice and helpful discussions.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Disclosure of Potential Conflict of Interest
  10. References
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
stem1468-sup-0001-suppfig1.tiff4106KSupporting Information Figure 1
stem1468-sup-0002-suppfig2.tiff4061KSupporting Information Figure 2
stem1468-sup-0003-suppfig3.tiff3832KSupporting Information Figure 3
stem1468-sup-0004-suppfig4.tiff5703KSupporting Information Figure 4
stem1468-sup-0005-suppfig5.tiff5945KSupporting Information Figure 5
stem1468-sup-0006-suppfig6.tiff5704KSupporting Information Figure 6
stem1468-sup-0007-suppfig7.tiff5358KSupporting Information Figure 7
stem1468-sup-0008-suppfig8.tiff3712KSupporting Information Figure 8
stem1468-sup-0009-suppfig9.tiff1704KSupporting Information Figure 9
stem1468-sup-0010-suppinfo.doc106KSupporting Information
stem1468-sup-0011-suppinfo.doc810KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.