A novel exogenous concentration-gradient collagen scaffold augments full-thickness articular cartilage repair.

Mimura T., Imai S., Kubo M., Isoya E., Ando K., Okumura N., Matsusue Y. A novel exogenous concentration-gradient collagen scaffold augments full-thickness articular cartilage repair. Osteoarthritis and Cartilage 2008;16(9):1083-91.

http://dx.doi.org/10.1016/j.joca.2008.02.003

Background

Articular cartilage has poor regenerative properties because it is avascular and aneural. Furthermore, chondrocytes, the cells responsible for matrix synthesis and maintenance, are essentially immobilized in lacunae (small spaces in the cartilage), so they are unable to migrate to the site of injury and promote healing. Thus, partial defects in the cartilage show little healing if any. However, if the injury penetrates the subchondral bone, cartilage can be regenerated through migration and differentiation of mesenchymal stem cells (MSCs) or other mesenchymal progenitor cells. Small diameter (less than 3 mm) full-thickness defects can be repaired in this manner. On the other hand, large diameter defects (larger than 5 mm) are filled with fibrous tissue, which is much weaker than the original articular cartilage. Previous studies have used collagen gels and membranes seeded with chondrocytes to treat the defect. This study uses a novel concentration gradient (CG) collagen type I gel to recruit MSCs from the host (Japanese white rabbit) to the center of the defect to enhance repair.

Summary

To create a concentration gradient (CG) collagen gel, the authors formed composites of two gels with different concentrations of collagen type I. Two composites were made: a 33% CG gel (0.18% and 0.24% collagen) and a 50% CG gel (0.18% and 0.27% collagen).

In order to demonstrate the superiority of the CG gel, the researchers created artificial defects in the cartilage of the rabbits and implanted a 33% or 50% CG gel, or a non-composite gel (0.18% collagen). Histological sections were prepared at set time points (1, 2, 3, 4, 8, 12 weeks post operation). To detect MSCs, the animals were injected with BrdU one hour prior to sacrifice. BrdU will be incorporated by proliferating cells, thus they will not stain chondrocytes.

Shown in Fig. 7 are the histological sections for week 2. An interesting result is that the 33% CG gel has more BrdU cells in the central region than the 50% CG gel does, indicating higher levels of cell migration.

At week 3, the researchers found that the 33% CG gel had significantly fewer BrdU-positive cells in the peripheral regions compared to the control and 50% CG gel. Taken together with the result from week 2, the researchers suggested that the 33% CG gel was superior at recruitment of MSCs.

The authors suggest that migration occurs via haptotaxis (migration towards higher density of adhesion sites). To study the haptotactic effect of collagen, combinations of collagen gels were studied in in vitro migration assays. Because the CG gel was eventually to be implanted into a Japanese white rabbit, MSCs were taken from those rabbits to be used in the experiment.

Interestingly, cell migration peaks at a certain CG and then drops later. The cause of this behavior may be the reduced mobility at higher binding site densities, which competes with the haptotactic effect of the collagen and decreases the amount of cell migration. This would also explain the result seen in Fig. 7, where the higher CG gel appears to induce less migration.

Articular cartilage tissue formation is also evaluated. The researchers used a specific histological grading scale to evaluate the cartilage at 4, 8, and 12 weeks post-op. While the 33% CG gel has superior characteristics over the control in the 4 and 8 week period, no significant difference was noted at the 12 week period. Furthermore, no significant difference was noted between the 33% and 50% CG gels.

Critique

What is peculiar about this work is that at the 12 week postoperative time point, there is no significant difference between a non-composite collagen gel and the two CG gels. This raises the question of whether increasing migration of MSCs into the defect zone is of any help at all in the long term. The authors claim that though there may not be significant differences in between the histological grades, the 33% CG gel leads at every time point. However, it might be possible that the grading scale they used (shown in the paper) may give unnecessary weight to certain aspects of the tissue.

Another issue that ought to have been addressed is the lack of consistency between histological sections analyses. In the paper, data for the central region of the defect is only shown for week 2, while peripheral region data is only shown for week 3. The authors did not explain why certain data was omitted or why they chose to analyze those particular regions at those specific time points.

Effects of the mechanical properties of collagen gel on the in vitro formation of microvessel networks by endothelial cells.

Tissue Engineering. 2007 Jul;13(7):1443-53.

Note: Figure captions are directly from the paper.

Introduction:

In recent years, tissue-engineered organs have become an increasingly promising source of organs for organ replacement therapy. Like those naturally found in the body, tissue-engineered organs require vascular networks to provide cells with nutrients and oxygen, as well as to remove waste products. However, the process by which these networks are constructed is not well-characterized, and thus, tissue-engineered organs cannot yet be vascularized reliably in the laboratory.

Recent studies suggest that the formation of vascular networks is heavily dependent upon the interaction between endothelial cells (ECs) and the surrounding extracellular matrix (ECM). In particular, in vitro models suggest that the mechanical properties of the ECM substrate govern many important cellular processes, including proliferation, migration, and differentiation. Although these effects have been observed, direct correlations between specific mechanical properties and cellular responses have not yet been established, and thus, these phenomena have not been understood on any quantitative level.

Summary:

General summary

The purpose of this study was to understand the effect that the stiffness of a substrate has on the formation of three-dimensional vasculature. To characterize this effect, bovine pulmonary microvascular ECs (BPMECs) were grown on collagen gels of varying stiffness in the presence of an angiogenic factor. Several different forms of microscopy (light, confocal, scanning electron, and transmission electron microscopy) were used to characterize the cellular response to the stiffness of the growth matrix.

Summary of results

Collagen gel stiffness is modulated by changing the pH of the gel

A series of collagen gels were prepared and polymerized under pH conditions between 5 and 10. The mechanical properties of these gels were characterized through uniaxial compression tests, from which stress-strain curves were derived. Based upon these curves, a relaxation modulus (a parameter for assessing the stiffness of viscoelastic materials) was calculated for each of the gels.

Figure 1 shows a plot of the relaxation modulus as a function of the pH of the gel. As indicated in the graph, the stiffness parameter appears to be linearly correlated with the pH of the gel, and there was roughly a 5-fold increase in relaxation modulus for an increase in pH from 5 to 9. In each of the subsequent cell studies, a gel at pH 5 was termed “flexible,” while gels produced at a pH of 9 were “rigid”.

Collagen gel stiffness affects the structure of cellular 3D networks

Using confocal laser scanning microscopy, the three-dimensional structures of the capillary-like networks formed by the cells were observed. Figure 3 (as numbered in the paper) shows the networks that were produced in flexible (A, B) and rigid (C, D) gels. As the paper discusses, “In the flexible gel, most cells…formed thin, dense networks,” while the rigid gels seemed to encourage the formation of “thick networks that were sparsely distributed” (Yamamura, et al., 2010). In addition, the networks formed in the flexible gel were found to be relatively shallow, primarily distributed 0-50 mm below the gel surface. By contrast, networks in the rigid gels reached significantly deeper (up to 120 mm below the gel surface).

Discussion and comments:

The purpose of this investigation was to consider the effect of collagen gel stiffness on the formation of three-dimensional vasculature by endothelial cells. This was accomplished through the use of an in vitro 3-D angiogenesis model coupled with several different forms of microscopy. In combination, these techniques indicated that BPMECs produce dramatically different vascular networks in response to differences in matrix stiffness.

Although this study suggests that matrix stiffness changes the cellular response, it provides no indication as to the molecular basis for this difference. Indeed, the discussion concludes, “This difference [in network morphology] may be a result of differences in cell-ECM interactions…” Of course, this correlation was discussed in broad terms in the introduction, and this statement suggests that no further light was shed on the cellular response through this study.

The authors of this study appear to have produced a novel 3-D angiogenesis model for studying the formation of vascular networks. In particular, this investigation shows that the mechanical properties of a collagen gel can be modulated by varying the pH conditions under which the gel is produced. It is unclear as to whether this is a standard protocol, but it does not appear to be a precise one. In particular, the standard deviations in relaxation modulus at each of the pH values (Figure 1) are quite significant, even averaged over nearly thirty gels. This result suggests that the stiffness of the gel cannot be controlled reproducibly to produce quantitative results. Thus, a different model may be required to study the effect of matrix stiffness on the morphology of vascular networks.

Environmentally Controlled Invasion of Cancer Cells by Engineered Bacteria

Summary

The use of bacteria as live drug delivery agents is an emerging application of synthetic biology. Toward this end, efforts have gone to develop controlled interactions between mammalian cells and bacteria. This paper details the development of a mammalian cell invading device that allows bacteria, specifically Escherichia Coli, to invade mammalian cells upon environmental signals. E. Coli have been engineered to invade various types of cancer-derived cells such as HeLa, HepG2, and U2OS cell lines. The expression of the invading device is induced by various environmental signals. The fdhF promoter and the araBAD promoter are induced by hypoxia and arabinose, respectively. When bacterial cells reach a density of greater than 10^8 bacteria/ml or are in the presence of 0.02% arabinose, activation of the invasion device occurs. The actual gene that induces the invading is the inv gene from the organism Yersinia pseudotuberculosis, which when expressed in E. Coli, will bind tightly to beta-1 integrins on mammalian cells. The binding of beta-1 integrins on the surface of mammalian cells stimulates Rac-1 and induces uptake of the bacteria into the cell. The invasin device effectively couples environmental inputs with the invasion as the output. Tuning of the input to output strength was done using a combinatorial strategy of constructing ribosome binding libraries and genetic selection. The use of invasin can then be coupled to the delivery of therapeutic proteins and plasmids into the cell.

Results

Assays were done to test for the nature of invasin. The ability of invasin to invade mammalian cells were tested by putting the inv gene under the constitutive tet promoter and allowing the MC1061 E. coli strain to interact with HeLa cells. The readout for the assay was the fraction of bacteria recovered from lysis, which is ~8% for inv+ E. coli and negligible for inv- E. coli. To determine the role that P pili plays in the adhesion and uptake of E. coli to mammalian cells, a fim deficient strain (CAMC600) was tested, and the strain retained its invasion abilities. Similar experiments with other cancer cell lines such as HepG2 and U2OS show that invasin is capable of invading cancer cells from different origins. It should be noted, however, that the level of invasion probably depends on the amount of beta-1 integrins expressed and their accessibility. Also, invasin only acts on cancer cells that are actively expressing beta-1 integrins. These results are shown in the figure below. The white bars indicate percent bacteria recovered from lysis with the inv+ system. The grey bar indicates the percent recovered with the inv+ CAMC600 system. The black bar represents the inv­- system. An asterisk indicates no bacteria recovered.

The inducibility of the invasin system was then tested with arabinose induction. The araBAD promoter was fused to the inv gene, and the expression of invasin upon induction of arabinose was tested. It was noted that the basal expression of the inv gene due to a leaky promoter was enough to produce invasin and invade mammalian cells. Hence, an RBS library was made where the RBS sequence was randomized and the clones that did not invade cells without arabinose induction were chosen. In the figure below, the white bar represents without induction. The grey bar is induction with arabinose, and the black is anaerobic induction.

Invasin is further linked to hypoxia by switching the promoter to the fdhF promoter which results in the expression of the inv gene in only anaerobic conditions. The expression levels were tuned by an RBS library. It was shown that after 2hours in anaerobic conditions, about 10% of added bacteria were recovered, using the same gentimycin assay. The inv gene is further put under the control of luxR promoter, which is part of the quorum sensing genetic circuit that is activated upon increase in cell density. Under quorum control, the inv gene is only detectable under high cell densities.

Significance

The engineering of an invasin gene under various promoters allows it to invade mammalian cells selectively at high cell density, in anaerobic growth, or chemical induction. This scheme can be used to invade cancer cells of diverse origins. The engineered invasin can also be a target for protein engineering to alter its properties such that the expression of invasin results in binding but not invasion of cells. A very significant result of engineering an inv+ E. coli is that other devices can be added to the system to make a therapeutic bacterium. The inv+ system lays the groundwork for applications such as live vaccines and gene delivery vectors. In particular, invasive bacteria is being engineered with additional genetic devices such a self-lysis and drug delivery system that would allow bacteria to invade cancer cells and release drugs to kill the cell.

Critique

One problem with the system is that bacteria will only invade mammalian cells that are actively expressing beta-1 integrins, so that cancer cells that do not express that surface protein will not be detected by the bacteria. Moreover, cancer cells are not the only cells expressing beta-1 integrins as surface proteins, so a possibility of invasion of other cells is present. Another major problem is the anti-LPS response from the mammalian immune system. Modifications to the lipopolyssacharide structure of the E. coli will have to be done for the bacteria to survive in the body.

Anderson, J.C., Clarke E.J., Arkin, A.P., and Voigt, C.A. (2005). “Environmentally Controlled Invasion of Cancer Cells by Engineered Bacteria.” Journal of Molecular Biology. 355, 619-27

Long-term maintenance of human hepatocytes in oxygen-permeable membrane bioreactor

0. Source and Links

Cohen S, Dvir-Ginzberg M, Gamlieli-Bonshtein I, et al. “Long-term maintenance of human hepatocytes in oxygen-permeable membrane bioreactor” Biomaterials. 2006 September; Volume 27 Issue 27: 4794-4803.

http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6TWB-4K48M84-1&_user=10&_coverDate=09%2F30%2F2006&_rdoc=1&_fmt=high&_orig=search&_sort=d&_docanchor=&view=c&_searchStrId=1273475599&_rerunOrigin=google&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=08a6aae39f6c2641d9e139b1c58cca23

1. Introduction

The main goal of studying hepatocytes (liver cells) in bioreactors is to adapt that method into a liver-support device. Previous studies used flat membrane bioreactors to mimic real liver function and structure. This study used flat, O2 permeable polymer membranes to grow their cells. The O2 permeability enables oxygenation of cells attached to membrane and of the medium which stimulates sinusoidal organization (similar to in vivo morphology). The idea is that this will translate into improved function and viability. The end goal of the study is to evaluate in-vitro performance of the new O2 permeable membrane bioreactor in maintaining and differentiating hepatocytes (human).

2. Summary of Methods and Results

The authors began by preparing human hepatocyte culture in hepatocyte culture medium and feeding it growth factors and various nutrients. They established a positive control: a culture on a plastic petri dish. They then seeded the hepatocytes on a gas-permeable membrane under serum-free conditions for the entire culture time.

Figure 1: the experimental bioreactor set up

The oxygen permeable bioreactor was connected to perfusion system and fresh media was sent through the set-up continuously at 0.2 mL/min. According to a rotating schedule, therapeutic molecules were sent through the bioreactor to gather information on their effects. The molecules used were Interleukin 6 (IL-6) which activates certain genes for liver repair and Diclofenac (DIC), an analgesic which can affect metabloic functions (it is digested by proper liver function).

Figure 2: Top: cell morphology throughout the experiment run time Bottom Left: therapeutic molecule schedule and urea synthesis Bottom Right: Albumin synthesis correlated to therapeutic molecules present.

The liver-specific function was measured by gel electrophoresis and quantified – albumin, urea, secretion of total protein. Total protein analysis was carried out by performing SDS-Page on the proteins collected from the outflowing media, which were then stained with Coomassie Blue. The resulting gel was analyzed with Image Quant TL software to estimate protein molecular weights. Albumin content was measured via ELISA method; several biochemical assays were run on samples from inside the bioreactor. Liver function was also related to the degree of diclofenac digestion (measured by HPLC, a chromatography process): functioning liver cells will digest diclofenac into its metabolites 4’ OH diclofenac and 5' OH diclofenac.

Figure 3: total protein analysis using SDS-Page

The results from the study were many-fold: the researchers not only compared their bioreactor performance with a petri dish, but they also compared the effects of the therapeutic molecules they had treated the cells with. With regard to morphology, their cells reached confluency of plates by day 4 out of a total culture time of 32 days. By day 11 the cells were polyhedral with pericellular zones similar to in vivo. With regard to protein production, total protein secretion was compared between collagen-plated and the O2 permeable method and found that liver function and albumin secretion was improved in the O2 plated method. With regard to the effects of the therapeutic molecules, the Urea synthesis decreased with DIC and recovered when it was removed; Albumin expression worsened when exposed to DIC + IL-6 but levels did recover when the drugs were removed.

Figure 4: HPLC results for diclofenac digestion. Samples from inflow and outflow were compared to determine degree of diclofenac breakdown.

Figure 5: Heptocyte functionality with respect to therapeutic molecule influence. These charts compare the protein secretion of the cells under the influence of various combinations of diclofenac and IL-6. The time schedule references the particular therapeutic molecule combinations present that day (see Figure 2 bottom left).

3. Discussion:

In this experimental set up the cells grew on the border between gas and liquid phase (the medium didn’t act as diffusion barrier so the cells were more able to organize themselves in an in vivo fashion). O2 transport is theoretically improved in this set up because O2 is usually poorly solubilized in medium. The fluid dynamics of device was highly characterized so system was controlled. Hepatocytes grown in new device were morphologically similar to those in vivo and functionally similar, as they expressed proteins for an extended period of time It is suspected IL-6 mediates the effects of diclofenac (DIC). DIC is thought to deplete the cell’s supply of ATP and thus interfere with normal protein production levels. This would explain why the damage was not permanent and after the drugs were removed, protein levels recovered. Prolonged exposure to DIC did seem to reduce the metabolic efficiency of the liver cells (better at 14 days than 29 days). The metabolites of DIC are thought to damage cytochrome P450 used in the microsomal hepatic system.

4. Conclusions:

The experimental set up was an improvement over petri dish culture. The bioreactor hepatocytes can be maintained with good functionality preserved over the long-term. The O2 permeable bioreactor has the potential to serve as a good model for drug behavior. The study also found that IL-6 mediates DIC (reduces its harmful effects).

5. Criticisms:

My first criticism is the order in which the researchers ran their therapeutic molecules. The study ran IL-6, IL-6+DIC, DIC in that particular order. They did not consider if there might be any residual effects from running the therapeutic molecules in that order. For example, they don’t know how DIC -> Il-6 would have affected their results.

My second criticism is the study’s control. The study didn’t have a control reactor which ran IL-6 only or DIC only, etc. to compare their results with. They ran all experiments in the same reactor (granted it allowed them to see recovery from previous processes). They shouldn’t have been able to claim that one drug had a particular effect over another since they did not isolate that interaction in its own bioreactor (insufficient controls). Comparing it to the collagen plate wouldn’t be as good as comparing it to an identical reactor under different conditions.

My third criticism is that there was no direct comparison between the O2 permeable method and the rival method mentioned at the beginning of the paper. The motivation for other scientists to care about the author’s work is that the work shows that their novel bioreactor is an appreciable improvement over other methods. Since the authors don’t make this comparison, the paper loses value to its readers.

My fourth criticism relates to the lack of explanation regarding the therapeutic molecules used in the study. The authors mention IL-6 reduces effect of DIC, but IL-6 alone also decreases protein expression of hepatocytes (as seen Figure 2 bottom right): this is not explained. This could mean that IL-6 has some properties that are damaging to protein production and this topic could be further explored.