One of the first successful tissue replacement therapies was artificial skin, developed by Drs. Ioannis Yannas and John Burke in 1979. Since then, tissue engineering has blossomed into a multidisciplinary field of biomaterials science, cell biology, and translational medicine. Advances in stem cell research, in particular, have greatly progressed the field of tissue engineering and regenerative medicine. Scientists have developed effective differentiation protocols to convert pluripotent or multipotent stem cells to their favorite cell type, but selecting superior subpopulations of lineage-committed cells for optimized performance of the engineered tissue is still an ongoing field of research. The lab of Arjun Raj takes a systems approach to biology in their work to quantify cellular functions. Allison Cote, a fourth-year CPM student in the lab, along with Claire Macleod of the Mauck Lab—a lab with expertise in cartilage differentiation—spearheaded a study that will force those in the stem cell field to think twice about correlating marker gene expression with phenotype.
Allison and colleagues utilized mesenchymal stem cells (MSCs), which can undergo either osteogenic, adipogenic, or chondrogenic differentiation, to ask whether chondrogenic marker expression could predict chondrocyte potential. Chondrogenic differentiation is typically characterized by proteoglycan synthesis and matrix accumulation, along with a general increase in aggrecan gene expression over the first seven days. Single-molecule RNA fluorescence in situ hybridization (FISH) was used to visualize individual mRNA molecules within an MSC. The researchers observed that over an entire cell population, the canonical pattern of aggrecan expression held true. However, individual MSCs showed significant cell-to-cell variability in aggrecan copy number during chondrogenic differentiation. Given the inherent heterogeneity between MSCs, Allison and colleagues hypothesized that higher expression of aggrecan would yield chondrogenic cells more robustly. Interestingly, increased aggrecan mRNA was not predictive of high-performing cells—as indicated by aggrecan core protein, a central component of the extracellular matrix (low-performing cells lacked extracellular staining for aggrecan core protein). Even more remarkably, twelve hours after cell division, sister MSC cells showed vastly divergent levels of aggrecan and GAPDH expression, suggesting that the heterogeneity in marker expression is not heritable or, at least, not maintained when propagated. This is quite a surprising finding given the traditional notion of marker gene expression, but those in the field of single-cell biology would argue this is not unusual, long knowing that cellular heterogeneity is present in prokaryotes and eukaryotes.
Since expression of aggrecan ineffectually sorted MSC populations, high-throughput RNA sequencing was employed to find genes that are more predictive of cell fate. Strikingly, the analysis demonstrated no patterns of gene expression that correlated with matrix accumulation. The most predictive genes, MMP13 and aggrecan, were only loosely correlative. How can analysis of aggrecan on a population, but not on a cell-to-cell basis, show a trend in expression? The answer lies in normalizing aggrecan copy number to housekeeping genes such as GAPDH. Previous studies in the Raj lab have demonstrated that expression of these housekeeping genes correlates with cell size, and that the chondrocyte spread – cell area and volume – increased with passage number. In examining gene expression of de-differentiating chondrocytes, Allison and colleagues noted higher GAPDH levels in these cells and only minor changes in aggrecan copy number, suggesting that the process of de-differentiation was primarily due to increased cell size. As has been reported recently, transcription is a stochastic process; this paper’s findings bolster that idea and suggest that marker copy number fluctuates rapidly over a short timescale.
MSCs cultured in vitro exist as a heterogeneous population, as defined by their gene and protein expression levels. These heterogeneous populations of MSCs differentiate to produce a heterogeneous population of chondrocytes that are low-performing and therefore not suitable for cell replacement therapies. Allison Cote et al. showed that obtaining a homogeneous population of MSCs by sorting for similar gene expression still resulted in a heterogeneous population of chondrocytes upon differentiation.
If gene expression profiles cannot predict chondrogenic differentiation, then what can? Other regulatory mechanisms, such as post-translational modifications, most likely dictate a chondrocyte’s functional capacity – not every cell that translates aggrecan core protein will appropriately modify the protein for secretion. And once aggrecan is secreted, it associates with the extracellular matrix (through interactions with hyaluronic acid and collagen, among other molecules) so if there are abnormalities in that network, matrix accumulation may be inhibited. Allison remarks, “I think our paper highlights that successful differentiation of a set of stem cells really requires a functional readout, and gene expression of what would normally be considered ‘marker’ genes is often insufficient because successful differentiation requires the whole intracellular system to be in a differentiated state. I think it also highlights intrapopulation cell heterogeneity, a characteristic that is often explored in the fields of single-cell and stem cell biology, but less so in tissue engineering, particularly with mesenchymal stem cells. I think cell-to-cell variability is something that has caused difficulty in defining an optimal differentiation routine for the massive quantities of cells needed for therapeutic applications.”
Currently, Allison is working on using expansion microscopy in combination with single molecule RNA FISH to examine the structure of RNA at regions where it is densely packed within the cell, such as transcription sites and mitochondria.
A commentary on:
Cote AJ, McLeod CM, Farrell MJ, McClanahan PD, Dunagin MC, Raj A, Mauck RL. Single-cell differences in matrix gene expression do not predict matrix deposition. Nature Communications, 2016; 7(10865).
Link to the PubMed page.