Engineered 3D tissue models for cell-laden microfluidic channels

Young S. Song, Richard L. Lin, Grace Montesano, Naside G. Durmus, Grace Lee, Seung-Schik Yoo, Emre Kayaalp, Edward Hæggström, Ali Khademhosseini, Utkan Demirci

http://www.springerlink.com/content/550231201lq76j78/fulltext.pdf

Introduction

An important part of tissue engineering technologies in areas such as organ transplantation and tissue regeneration are the 3D scaffolds on which cells are grown. These scaffolds need to be mechanically strong enough to support tissue growth but must also be porous enough to allow for the diffusion of fresh nutrients. This paper compares 1D perfusion and 2D perfusion models on agarose and aims to find the best geometry for the microfluidic channels that introduce nutrients by diffusion for highest cell viability. Theoretical models are compared to concentrations of nutrients and cell viabilities in actual 1D and 2D perfusion agarose cell scaffolds.

Summary

Fig. 1

The authors of the paper compared a 1D perfusion model with a 2D microchannel perfusion model. For the 1D model, 3T3 cells in agarose were mixed, deposited into a Petri dish, allowed to gel and covered with 3T3 medium. For the 2D perfusion model, one model included a single 300um microchannel through the center of a tube containing agarose-cell mixture and the second model included two microchannels of different separations and radii. The microchannels were constructed by letting the hot agarose gel around a small glass capillary tube. The tube was then pulled out enough to leave 1cm of the tube inside the agarose for flowing of media.

Fig. 2

Before the actual cell culture experiments were run, numerical simulations of normalized concentration were performed. The variables examined were the radius of the single microchannel in the 1D perfusion model (0.2mm to 0.8mm), radii of both microchannels in the 2D perfusion model (0.2mm to 0.8mm) and the distance between the two microchannels of the 2D perfusion model(2mm to 8mm) to determine the nutrient concentration distribution. The nutrient concentrations were normalized using the concentration at the top of the agarose.

Fig. 3

Theoretical analyses of the distribution of nutrients was also run across 3 days. Figure 3 A-C show contours of the distribution of nutrients at day 1 for the three different perfusion models. The contours suggest that the 2D perfusion model with dual channel is the best geometry to maintain high cell viability. Figure 3 D-F show theoretical normalized nutrient concentration for day 1, 2 and 3. Again, with the dual channel 2D perfusion model, the concentration distribution is the highest, allowing for cells as far away from the nutrient source as 5mm to be viable.

Fig. 4

To validate theoretical predictions, first the 1D perfusion model was tested with an initial cell viability in the Petri dish of 89%. As seen in Fig. 4 the cell viability decreased both with longer perfusion time and distance from the hydrogel surface. These results were consistent with theory. Cell viability was measured using a Live/Dead kit on 500um increments of slices of the hydrogel.

Fig. 5

After the 1D perfusion model, both of the 2D perfusion model conditions were tested on three consecutive days of culture and the viability at increasing distances from the perfusion channel was measured on each day. The results indicate that the dual channel configuration maintains cell viability over 80% after 3 days and up to 8mm away from the nutrient source while with the single channel configuration the cell viability drops to 50% after 3 days at the same distance. These results of different 3D tissue engineering approaches provide a theoretical and experimental understanding of nutrient diffusion in tissue culture for cell viability.

Discussion

While this paper compared simple 3D culture with 1D perfusion to 3D culture with 2D perfusion, it did not go far enough. Other configurations should have been tested, especially after the authors suggested that there is an ideal distance and configuration that would give the highest cell viability. In addition, cultures should have been tested after more than 3 days since the drop in viability may be significant. Experiments involving different concentrations could have been done to eliminate that as a factor. The theoretical simulations were close to half of the paper’s content and could have been combined into one figure to allow for space for more relevant experiments.

Differentiation of Human Embryonic Stem Cells to Regional Specific Neural Precursors in Chemically Defined Medium Conditions

Slaven Erceg, Sergio Lainez, Mohammad Ronaghi, Petra Stojkovic, Maria Amparo Perez-Arago, Victoria Moreno-Manzano, Ruben Moreno-Palanques, Rosa Planells-Cases, Miodrag Stojkovic.

Introduction

Human embryonic stem cells (hESC) are pluripotent cells that can be propagated in vitro and theoretically provide a source of precursor cell that can be differentiated into any cell type. Thus derivation of neural progenitors from hESC provides a mean to study the central and peripheral nervous system, and for potential cell therapy applications to treat diseases.

Currently, hESC differentiation towards neural lineages include presence of stromal cell lines and/or conditional medium includes a multi-step process which involves formation of embryoid bodies (EBs), risking pathogen cross-transfer or contamination with non-neural cells. Countermeasures such as developing feeder free conditions for growth and controlled generation of neural progenitors in feeder and animal-free conditions avoid the formation of EBs. They devised their protocol which includes usage of animal-free components of ECM and chemically defined medium. Furthermore, the protocol allows controlled differentiation towards a specific region by exposing the rosettes to a signaling factor.

Summary

Figure 1: Schematic representation of different steps

Human ESC were cultured on human foreskin fibroblasts and ES medium (enriched natural seawater medium) was changed every second day. For controlled differentiation, hESC were passaged to plates in PBS in the chemically defined medium. After 24 hours the cells attached and exhibited typical hESC morphology and after day two there were signs of neural differentiation and even further morphological development such as

Figure 2 (B and C): Morphological features of neural lineage can be seen (black arrows pointing to rosettes)

Immunocytochemical analysis, using 1% BSA for blocking and both primary and secondary antibodies, was used to detect neural markers such as PAX6 and Nestin, either data was shown, but paper stated that they were present.

After day 7, the medium was replaced with GRM medium along with basic fibroblast growth factor (bFGF). Rapid growth of rosettes occurred and at day 28 immuncytochemical analysis showed that cells were positive for numerous neural progenitor markers.

Figure 3: Differentation of hESC to neural progenitors Day 28

The hESC derived neural progenitors gave rise to astrocytes and oligodendrocytes, marked by gfaf and o4, respectively around day 42. Furthermore presence of serotonin. glutamate, and GABA, are detected. RT-PCR analysis showed that changes in expression of hESC and neural markers during ES, day 7, and day 42 stages. A down-regulation of pluripotency markers OCT4 and MAP2 and an up-regulation of PAX6, SOX1, NCAM, and MAP2, human neuronal markers. All of which indicate differentiation of hESC to neural precursor cells.

Figure 4:
D: Oligodendrocyte and astrocytes expression profile
E: Changes in gene expression of main pluripotency markers and general neural markers

To identify the population types obtained they first determined the expression of rostral-caudal (anterior-posterior) CNS markers using RT-PCR. They analyzed transcription factors involved in dopaminergic differentiation (largely rostral and midbrain markers) and found that those of the rostral markers were upregulated while midbrain cells had weak expression. The neural progenitors were then examined for caudal characteristics using class I and class II, homeodomein proteins, all caudal markers. They found high expression but little to no differentiation towards caudal cells, suggesting that using bFGF differentiated towards rostral cells as well as activation of caudal markers only. They tried to alter this rostral-caudal relationship by exposing the cells to RA (a well known caudalizing signal). Using the same protocol and methods, the found that the rostral and midbrain markers were strongly suppressed, that the cells acquired a spinal positional identity, and the caudal markers were strongly expressed or remain unchanged. Interestingly, the cells differentiated into a motorneurons upon this RA treatment.

Figure 5
E: RT-PCR analysis of rostral markers of hESC derived neural progenitors with or without RA
F and G: RT-PCR analysis of spinal markers

They also studied the functional properties of these hESC derived neuronal precursors by patch clamp analysis. 79% of tested cells evoked at least one overshooting action potential in GRM/bFGF and there were repetitive firing in 9% of cells but TTX reversible blocked action potentials. The presence of voltage-gated Na and K channels were studied in the voltage-clamp configuration and in the presence of intracellular K an early and fast inward current was followed by a sustained outward current shown in figure 6B.

Figure 6: Neuronal excitability and study of voltage dependent channels
A: Action potentials evoked by depolarizing current steps. Both cases show that the spikes were fully blocked with TTX (red)
B: Early inward current suggests presence of voltage dependent sodium channels and the second outward component is consistent with K channels

Finally tested the neurotransmitter sensitivity to GABA, glutamate, dopamine, and acetylcholine of the neurons present in the culture and found that the majority of the cells evoked whole cell current and can be blocked by corresponding inhibitors.

Discussion

The paper demonstrated that neural progenitors can be generated in a one step approach without forming EBs and showed that there are advantages including non spontaneous differentiation of hESC and early markers for neural cells. They didn't however discuss how the method can lead differentiation towards more specific neuronal cells and they deterred from using a feeder layer for culturing but they maintained they undifferentiated hESC on a feeder layer before starting their study which seemed contradictory. I should point out that their actual experiment didn't use those cells until they were transferred to a feeder-free culture. Although they succeeded differentiating their cells to general regions of the brain, it didn't seem very specific and may be hard to evaluate.

Modulation of Proliferation and Differentiation of Human Bone Marrow Stromal Cells by Fibroblast Growth Factor 2: Implications for Tissue Engineering

Hankemeier, S. et al. 2005. Modulation of Proliferation and Differentiation of Human Bone Marrow Stromal Cells by Fibroblast Growth Factor 2: Potential Implications for Tissue Engineering. Tissue Engineering. 11(1/2): 41-49

Introduction:

A recent approach to ligament and tendon reconstruction in patients involves the use of human bone marrow stromal cells (BMSCs). BMSCs have the potential to differentiate into a number of mesenchymal cell lineages that are integral to ligament and tendon formation, such as fibroblasts and chondrocytes. In addition, the usage of BMSCs avoids the issue of transplant rejection and also avoids the ethical concerns associated with other possible stem cell therapies. In order for BMSCs to succeed as a tissue engineering material, they must be able to proliferate, properly differentiate, and have similar expression patterns to those of the desired cell type. In this paper, the effects of FGF-2, a protein of importance in tendon and ligament healing, on the properties of BMSCs were studied.

To study the effect of FGF-2 on BMSCs, the authors set up three cultures: 1) BMSCs with a low dose of FGF-2 (3 ng/mL), 2) BMSCs with a high dose of FGF-2 (30 ng/mL), and 3) BMSCs with no FGF-2. Proliferation of the cells in each culture was quantified using BrdU. Apoptosis rates were quantified using flow cytometry; apoptosis was determined by the presence of a specific fluorescent marker (annexin V). To study the gene expression of the cells in different cultures, RT-PCR was run with a number of different primers for various genes of interest integral to the formation of ligaments and tendons.

Results:

Each of the three cell cultures displayed different morphologies. The cell density in the low-dose FGF-2 culture was higher compared to the other two cultures, and the cell density in the high-dose FGF-2 culture stagnated after day 14 (Fig. 1). Compared to the control, the low-dose FGF-2 culture had a higher cell density after 28 days and it also showed a much higher degree of BrdU incorporation.



In contrast to the differences in cell proliferation, the apoptosis rates for cells in each culture were relatively similar and the difference from culture to culture was not statistically significant.
Each culture had a different pattern of gene expression (Fig. 4).



Though collagen I, collagen III, fibronectin, and alpha-SMA mRNAs were present in all cultures, each culture expressed them in different amounts. The low-dose FGF-2 culture expressed much more collagen III and vimentin than either the control or the high-dose culture. For all the genes tested, the low-dose FGF-2 culture expressed at least as much mRNA as any of the other cultures.

Discussion:

The initial proliferation in low-dose FGF-2 cells coupled with a steady increase in expression of several proteins such as collagen I and III suggests that FGF-2 acts on BMSCs in a biphasic manner: the first phase involves proliferation of cells, and the second phase involves an upregulation of certain genes that is associated with differentiation of these BMSCs into specific cell types. Furthermore, the low-dose FGF-2 cultures displayed a higher degree of cell proliferation and gene expression than the high-dose FGF-cultures, which suggests that the effects of FGF-2 on BMSCs act in a dose-dependent manner.

The authors conclude, therefore, that a low dose of FGF-2 has a positive effect on the proliferation and differentiation of BMSCs whereas a high dose of FGF-2 has an adverse effect on BMSCs. This dose-dependent behavior of FGF-2 on BMSC activity is a variable that should be taken into account either in culture on in the design of ligament/tendon reconstruction methods using BMSCs.

Comments:

Though the use of various concentrations of FGF-2 to examine its effect on BMSC makes sense, the authors never explicitly mention why they chose the concentrations they did. I would have liked to known why they chose the numbers they used, and, more importantly, I would liked to have known if the concentrations they used are physiologically relevant. The authors also do not mention how many samples they use to calculate the cell proliferation data, and it seems as if they only used a single sample for each culture. I think the data would be more reliable if the authors had calculated cell proliferation data for a number of samples.

The authors are also a bit flippant in describing the morphology of each cell culture. They write that “large flattened cells and star shaped cells were observed in smaller numbers”, but they make no effort to properly characterize these cells. They also do not explicitly compare the morphology of the two FGF-2 cultures with that of the control. Since a major part of their paper involves the differentiation of BMSCs as a result of FGF-2, I think it would be helpful if they described the morphology in more detail.

Finally, the paper presents a clear correlation between FGF-2 and proliferation and differentiation of BMSCs but makes no effort to elucidate the molecular basis of this interaction. Why, for example, does FGF-2 act in a dose-dependent manner? This paper only studies an input (different doses of FGF-2) and the corresponding products (cell proliferation, mRNA expression, etc), so further studies are necessary to examine the exact way in which BMSCs are affected by FGF-2.

ABM/P-15 modulates proliferation and mRNA synthesis of growth factors of periodontal ligament cells

P. Emecen; A.C. Akman; S.S. Hakki; E.E. Hakki; B Demiralp; T.F. Tozum; R.M. Nohutcu. “ABM/P-15 modulates proliferation and mRNA synthesis of growth factors of periodontal ligament cells.” Acta Odontologica Scandinavica, 2009, 67(2): 65-73.

Introduction
Over the past ten years, ABM/P-15 has been studied as a potential matrix for bone repair by promoting the concentration-dependent binding of dermal fibroblasts and osteoblasts on ABM. The ABM (anorganic bovine-derived bone mineral) particles are coated with a synthetic cell-binding 15-residue peptide, P-15, which mimics the cell-binding region of the α1 chain in Type 1 collagen. These particles have been shown in the literature to enhance cell-attachment in periodontal ligament fibroblasts, dermal fibroblasts, and human osteosarcoma cells and thus suggest that ABM/P-15 could be a viable matrix and alternative to bone-grafts and less biocompatible methods for bone repair. Few studies are available that evaluate the mechanisms of how ABM/P-15 affects the proliferation and mineralization of cells once they have achieved attachment to ABM. In order to expand this library of knowledge, the authors study the effect of ABM/P-15 on the proliferation and mineralization of periodontal ligament cells and elucidate how ABM/P-15 regulates growth factors necessary (i.e. biomarkers) for periodontal regeneration in vitro.

Summary – of methods, of results
The authors in this study isolated periodontal ligament (PDL) cells from the root surfaces of healthy premolar teeth by harvesting PDL tissue from the root and culturing pieces in Petri dishes at 37C for five passages total in the appropriate media. Cells from the fifth and third passage were used in subsequent experiments. The 5x104 PDL cells and 1x104 PDL cells were cultured in either 5% FBS or 5% FBS+ABM/P-15 (50mg/ml) on 24 well tissue-culture plates; all experiments were done in triplicate. Cell morphology was investigated on Day 7 using DMEM and a phase contrast microscope. The cells for the control and particles (Fig. 1A, 1B) were spindle-shaped and polarized; the particles did not seem to affect the morphology expected of these cells and the cells are seen to closely associate with the particles in an orthogonal manner (Fig. 1B). A proliferation assay was performed by staining the cells with trypan blue to check for cell viability and viable cells were counted with a hemacytometer on Days 1, 6, 8, and 10. The ABM/P-15 treated cells showed consistently higher proliferation that the control group.

























The Von Kossa staining method was used to analyze the mineralization of the cells every week for four weeks; no mineralization was observed for either the control or cells treated with particles. The mRNA expression of growth factors TGF-beta, BMP-2, IGF-I, b-FGF, PDGF, COL-1, and VEGF were studied using RT-PCR and the appropriate cDNA primers on the 3rd and 7th day. The authors claim to observe an increase in expression for BMP-2 TGF-beta as well as a decrease in b-FGF and IGF-I for the ABM/P-15 treated cells versus the control group (Fig 3). No difference was observed in COL-1 or VEGF expressions, while the PDGF seemed to first be greater in the particle-treated cells on Day 3, they decrease to a signal equivalent to the control group by Day 7.













Critique
This paper served its intended purpose in expanding our knowledge of ABM/P-15 on different cell types by investigating their effect on periodontal ligament (PDL) cells. They use the cells in the third and fifth passages to conduct their studies, without commenting on the fourth. It is not clear if they mixed the cells from both passages before seeding the cells into wells for each assay; if they did not, there may be some discrepancy in the health of the cells between each passage and this would definitely influence the results. Different concentrations of PDL cells were seeded for the proliferation and mineralization assays. The reason for this was not explained, and the fact that no mineralization was observed in the control or particle-treated group may be attributed to the lower cell concentration. The authors also claim that the cells were spindle-shaped, densely packed, and well-oriented with the ABM/P-15 particles; however, the images of the control and particle-treated group show no real difference. In fact, the particle-treated cells align themselves perpendicular to the particles; it would seem that if the cell integrins were actually binding to the particles, they would be oriented parallel instead to increase interactions. There is a statistically significant increase in cell number in the particle-treated group, though, so this suggests that cell binding is increased. A cell attachment assay should be performed to confirm how much the cells are interacting with P-15 on the particle. The mRNA results show that COL-1 did not change between Day 3 and Day 7 and it is not clear whether or not the particles are faster and just reached a plateau sooner than the control, or if there is no real difference.. They should consider performing these assays at earlier time points to verify. Also, to help with their mineralization assay, the authors should consider evaluating the expression of alkaline phosphatase (ALP), an early differentiation marker for the osteoblastic phenotype (high levels of ALP are necessary for mineralization potential).

Functional Life-Long Maintenance of Engineered Liver Tissue in Mice Following Transplantation Under The Kidney Capsule

Functional Life-Long Maintenance of Engineered Liver Tissue in Mice Following Transplantation Under The Kidney Capsule

Authors: Kazuo Ohashi, Fumikazu Koyama, Kohei Tatsumi, Midori Shima, Frank Park, Yoshiyuki Nakajima, Teruo Okano

Introduction

An area of active research in the treatment of liver diseases is the use of hepatocytes in cell-based therapies. Hepatocyte transplantation has been shown to be effective but is limited by the number of cells that can be transplanted at one time. An alternative approach is to change an individual hepatocyte into biologically active tissue capable of carrying out liver function in an ectopic site. The current challenge is to increase the lifespan of these functional hepatocytes. In this study, functional hepatocytes were engineered in a mouse kidney capsule. This liver system was monitored for roughly the life-span of a normal mouse (about 450 days). Alongside stability, liver-specific functions, drug metabolism and regenerative potential were also analyzed.

Materials/Methods

Hepatocytes were isolated from transgenic mice expressing human α-1 antitrypsin (hA1AT) under hepatocyte-specific promoter (hA1AT-FVB/N). Cells were isolated using a modified two-step collagenase perfusion method, as described in Ohashi et al. Cells were then filtered through a nylon mesh membrane and purified by centrifugation. Cells were resuspended in equal amounts of serum-free DMEM media and EHS-gel to a final ratio of 1.5 x 10⁶ cells/100 μL. Cell viability was assessed using Trypan blue exclusion. Experiments were only conducted when the viability exceeded 90%. 1.5 x 10⁶ cells were transplanted under the left kidney capsule space in female wild-type FVB/N mice. hA1AT serum samples were collected periodically and assayed through ELISA . Serum analysis showed hA1AT levels in the range of 9000-30,000 ng/mL, which suggested the engineered liver was viable throughout the 450 days (Figure 1).

Fig 1. Serum hA1AT serum concentration levels measured at various timepoints during the 450 days. A (n=11) and B (n=9) are two separate experiments using different hepatocytes. C and D show the maematoxylin and eosin staining of the engineered liver tissues.

After 450 days, histological analyses were run on collected samples to test for liver regeneration stimulus and cytochrome P induction. In each case, specimens were collected from the naive liver, engineered liver and duodenum (as a positive control).

Some collected samples came from rats receiving intraperitoneal injections of Phenobarbital (PB) or 3-methylcholantree (3-MC) for 3 consecutive days before having their native and engineered liver tissues processed for histological analysis. The drugs were given to test the liver’s ability to uptake compounds and induce production of drug-metabolizing enzymes, which in this case was cytochrome P (CYP). PB is a CYP2B inducer, while 3-MC is a CYP1A inducer. Immunostaining for both showed strong responses in the engineered and naive liver samples. (Figure 2)

Fig 2. Immunohistochemical staining for CYP2B (A-C) and CYP1A (D-F).A & D show naïve mouse liver before respective treatments, B & E show them after treatment, and C & F show the engineered liver after treatment.

A 70% partial hepatectomy (PH) was performed on some naive rat livers to stimulate liver regeneration. Post surgery, 1 mg BrdU was administered to the naïve and engineered liver tissue per day for 14 days to measure hepatocyte proliferation. After 14 days, the naive liver, engineered liver and duodenum were removed for histological analysis. Analysis showed a 23±37% increase in serum hA1AT levels in PH mice relative to non-PH mice (Figure 3A). There was a significant increase in the amount of BrdU-labelling index (LI) in the PH mice than the sham-operated control, consistent with the values obtained from the naïve livers (Figure 3B). hA1AT staining showed a strong positive signal in the engineered tissue (Figure 3E, H) but not anywhere else.

Fig 3. Results from testing the regenerative capability of the engineered liver tissues. (A) hA1AT levels were measured by ELISA between day 450 and day 464. Triangles represent the control group while circles represent the PH group. # indicates p <> C-F are the controls, while F-H are PH. C & F are from duodenum, D & G from naïve liver and E & H from engineered liver tissue.

Finally, the study also assessed the ability of the engineered liver tissues to synthesize glycogen. PAS staining was done on sections from sham-operated and PH mice at day 464. Similar intensity levels were shown for both naïve and engineered livers, though the naïve liver images were not shown. Pretreatment with salivary amylase to remove the glycogen diminished the PAS staining, as expected (Figure 4)

Fig 4. PAS staining of the engineered liver tissues to assess glycogen synthesis. A & B were from non-PH mice, while B & D were from PH mice. A & C were not treated with salivary amylase, while B & D were.

Critique

The paper was very interesting, as the use of engineered organs for therapy is a hot area of research. However, this paper did have some drawbacks. First, they sometimes referred to data that was not presented in the paper. For example, they claimed that hA1AT staining was done on all organs when testing regeneration potential of the engineered liver tissue. They say staining showed that hA1AT was only being produced from the engineered liver tissues, but it would have been nice to see the justification. Figure labeling was also an issue for me. Sometimes they did not label their figures properly, such as in Figure 3B. It can be inferred that white represents naïve liver tissue and black represents engineered, but that should be clearly stated on the figure itself. Also, they only tested a few aspects of the liver’s abilities; much more testing is needed before this method can be used for therapy. Along with this, the hepatocytes will also need to be tested in other ectopic areas, ideally those closer to the conditions a normal liver would experience. The renal capsule is fine, but most likely it is a different environment than what the liver experiences sitting in the abdominal cavity. Finally, they only used female mice as the recipients for the engineered hepatocytes. Ideally, the tests would be run on equal numbers of both sexes.

Directing hepatic differentiation of embryonic stem cells with protein microarray-based co-cultures

All figure captions are from the Revzin group authors.

Introduction

In order to treat many degenerative liver diseases, recent research has turned to directed differentiation of embryonic stem cells (ESCs) into hepatocytes, either in vitro or in vivo, due to the shortage of livers for transplantation and other alternative therapies. Unfortunately, the conversion of ESCs into mature hepatocytes has encountered very low efficiencies, making the use of ESCs for liver disease therapies impracticable. However, research groups have experienced successes utilizing in vitro differentiation methods, in which a local microenvironment or niche does not have to be solely engineered; rather, by using mature hepatocytes in co-culture with ESCs, researchers can exploit the natural microenvironment of hepatocytes without having to determine each factor alone. Microenvironmental concerns include secreted cell signals such as growth factors, as well as the extracellular matrix. This co-culture method of differentiating ESCs in the same tissue culture plate as mature hepatocytes has been widely employed. However, the method presents two obvious drawbacks which this paper seeks to remedy: (1) random co-culture of ESCs with primary cells allows infinite and uncontrollable cell-cell interactions and (2) isolation of differentiated ESCs for further analysis proves difficult in a random co-culture. These problems have often been solved with micropatterning, which A.Revzin’s group also proposes in this paper. Printed arrays of collagen and fibronectin were used to localize hepatocytes and ESCs in a known spatial orientation. This method also allowed ESCs to be easily removed via laser catapulting for RT-PCR analysis of hepatocyte-related genes. The group ultimately determined that micropatterning allowed for effective co-culture of ESCs with primary hepatocytes and efficient removal of ESCs for further study.

Results

Revzin’s group micropatterned ECM arrays on glass slides as in Figure 1. The micropatterning technique allowed the researchers to localize cell populations and to strictly control distances between cell populations. The cells were then seeded onto their respective ECM proteins (mouse ESCs on fibronectin, hepatocytes on collagen I).



Fig. 1 Schematic description of the assembly of mESC–hepatic cell co-cultures on protein microarrays. Step 1: columns of fibronectin (blue) and collagen (I) (red) spots are printed onto a silane-modified (gray) glass slide. Step 2: after incubation for 1 h and removal of unattached cells, mESCs remained adherent exclusively on the fibronectin spots. Step 3: hepatic cells (HepG2) were seeded on the same surface 24 h after mESC seeding. Hepatic cells attached on collagen I) spots. Step 4: after cultivation for the desired period of time mESCs were extracted from the surface using laser catapulting and were analyzed with RT-PCR.




As seen in Figure 2, alkaline phosphatase staining was conducted on the micropatterned slides to ensure proper cell seeding, as AP stains stem cells but not hepatocytes. Over time, the cell populations clearly expanded out past the original attachment sites.

Fig. 2 Creating micropatterned co-cultures of mESCs and hepatic cells on protein arrays. (A) Seeding mESCs onto arrays comprised of alternating 300 m diameter spots of fibronectin and collagen (I) resulted in selective attachment of stem cells on fibronectin. FITC-labeled collagen (I) was used to highlight that collagen islands were free of stem cells and were available for hepatocyte attachment. (B) Staining co-cultures for alkaline phosphatase—a stem cell marker—revealed a strong signal from the stem cell-containing spot (right) and no signal from the hepatic cell cluster (left). (C) A low magnification (4×) view of the micropatterned co-cultures shows a 14 × 10 array of cell clusters. Clusters of mESCs appear more three dimensional and scatter light, while hepatic cells appear planar and darker. (D–E) Stem cells in mono- or co-cultures were able to expand out of the original attachment sites over time in culture. The images shown here were taken on day 8. Mouse ESCs adherent on fibronectin spots and cultured in hepatic differentiation media (D) were compared to mESC-hepatic cells co-cultures (E). (F) A higher magnification view of the stem cell cluster after 8 d in co-culture shows endodermal-looking cells spreading/migrating out of the cluster.


Figure 3 illustrates the efficiency of the laser catapulting technique employed by Revzin’s group that allowed them to remove a cell cluster independent of another. Laser catapulting offers the ability to analyze and link gene expression of a given cell clump to its location on the slide. Figure 3 also demonstrates the decrease in pluripotent gene expression in ESCs, indicating differentiation. However, the RT-PCR comparisons also indicate some subset of the ESC population remains undifferentiated. The paper states that this perhaps results from the fact that the centrally located ESCs were primarily undifferentiated, while the border cells closest to the hepatocyte co-culture did differentiate. This result indicates the importance of cellular interactions for the induction of hepatocyte cells. Similarly, they tested for expression of genes unique to the three germ layers, discovering the increase of mesodermal genes in differentiated ESCs.

Fig. 3 RT-PCR analysis of pluripotency and germ layer gene expression in stem cell micropatterns. Stem cells were cultured in hepatic differentiation media either alone or together with hepatic cells (HepG2). (A) Mouse ESCs were selectively retrieved from the co-culture using laser catapulting and were immediately used in RT-PCR studies. (B) Expression of pluripotency gene Nanog decreased over time for both mono- and co-cultivated mESCs; however, the co-culture induced a more pronounced and more rapid decrease in Nanog expression. (C) Expression levels of germ layer genes at day 1 and day 8. The endodermal marker, Sox17, was up-regulated, while other germ layer markers remained unchanged or decreased.







Additionally, researchers tested liver-related genes directly, finding that liver-specific gene expression appeared over time in the differentiated ESCs. Figure 4 illustrates the appearance of early hepatic genes. Finally, researchers tested the ESCs for mature hepatocyte markers, as seen in Figure 5. While the mature markers were not present early on, they appeared later in the co-culture stages and at the periphery of the cell population, indicating successful differentiation to hepatocytes and the importance of utilizing mature hepatocytes as a co-culture mechanism.


Fig. 4 Expression of early hepatic phenotype in micropatterned mESC cultures. (A) Expression of Afp, TTR and Alb-genes associated with early hepatic phenotype-occurred earlier and was more pronounced in mESC co-cultures compared to mono-cultures. (B) Immuno-fluorescent staining was used to corroborate the presence of AFP protein in mESC-derived cells after 5 d in culture.









Fig. 5 Expression of genes associated with mature hepatic phenotype. (A) In order to investigate location-specific differences in gene expression, cells were laser-catapulted from the center of the stem cell cluster as well as from the periphery of the cluster (edge). (B) Cells from different locations were catapulted into distinct centrifuge tubes and were analyzed using RT-PCR. These studies revealed considerable heterogeneity in G6p and Ggt gene expression with expression levels being higher at the edge (stem cell–hepatocyte interface) compared to the center of stem cell spot. It should be noted that no G6p and Ggt expression was observed in mono-cultured stem cells.









Critique

Though the cell populations seeded onto the micropatterned slides were localized, they were not immobilized. As seen in Figure 2, many of the populations expanded past their printed ECM borders, perhaps contributing to population mixing. This problem could perhaps be alleviated by immobilizing the cells on the ECM by adding an additional ECM protein. However, the paper states that a silane coating on the slides purposely allowed for the expansion of cell populations, which facilitated cell-cell interactions at the borders. While the paper argues that the populations remained distinct at the borders, cell mixing may still occur and steps could be taken to avoid contamination.

The researchers also utilized a culture media containing insulin for the co-culture incubation. While this addition perhaps assisted the differentiation process, it also somewhat defeats the purpose of co-culture, as the media now helps differentiate ESCs, rather than allowing the hepatocytes to secrete all necessary signals to direct differentiation. Additionally, an ESC population subjected to hepatocyte differentiation media was used as a control to compare media-differentiation and co-culture differentiation. While this method displayed good results (co-culture being more successful in differentiating ESCs), the media used cannot be wholly reliable as not all factors for directed differentiation can be determined.

Finally, this paper admitted problems with RT-PCR results, as the cycle to threshold number was difficult to determine despite multiple experiments. The paper did not state how the cell population was homogenized before starting RT-PCR, so a possibility for future studies would be to re-analyze the differentiated ESCs and control ESCs with a homogenized cell sample.