Tracking Cell Lineages of Single Cells

From Soft-Matter
Jump to: navigation, search

Entry by Andrew Capulli, AP225 Fall 2011

Reference

"Tracking lineages of single cells in lines using a microfluidic device" Amy C. Rowat, James C. Bird, Jeremy J. Agresti, Oliver J. Rando, and David Weitz, Proc. Natl. Acad. Sci. U. S. A. 106(43), 18149–18154 (2009).

Key Words

Phenotype, Microfluidics, Progeny, Derjaguin Approximation, Disjoining Pressure, Potential (voltage), Charge Density, Dispersion Stability - in terms of cells

Introduction: Motivation

While some cell lines continue to express apparently similar phenotypes over time as they replicate, other cell populations don't replicate at all (or do so very minimally). Take the human body in say, ten years of aging: skin tissue and the cells therein are relatively conserved, meaning the skin you have at the starting point will look and will physically be very similar to your skin in 10 years. Cardiac myocytes, those that make up your heart tissue will also 'look' the same but for a different reason: for the most part, they are the same (these are very general statements but they illustrate the point that follows). Cell expression (protein expression) as a function of time and generation is a somewhat under investigated area. As generally mentioned about tissues, the behaviors of cells in bulk (tissue) can and have been observed but the behaviors of individual cells and cells lines are masked in the three dimensional "ensemble" as the authors call it. In tissues or even just unconstrained clusters of cells (say, in a petri dish of culture flask) the lineages of dividing cells become mixed in the three dimensional space so when staining and investigation into particular phenotypes (protein expression) is investigated, it may be clear which cells are expressing the phenotype but their 'age' or 'generation' is unclear... tracking if the cells are newly spawned is impossible. Therefore, in these simple set-ups it is impossible to track phenotype as it varies down one lineage (conserved genetic makeup) of cell replication (starting with one cell). Here the authors propose to address this question with a novel microfluidic device that allows them to investigate the phenotype of a single cell lineage via separate chambers that house a single cell. The chambers allow for the cell to replicate down the chamber(1 dimensionally) and consequently the authors, after a period of replications, have the consequent generations starting from a single cell assembled in a line where they can then investigate cell phenotype.

Lineage Chamber Device

Much of this paper describes the lineage chamber device as shown below in Figure 1 from the paper. A detailed account is in the paper but I'll now summarize the device for the purpose of discussion later on. Thin chambers (1 budding yeast, Saccharomyces cerevisiae, cell wide) are assembled in parallel (50 chambers in parallel per unit). The chamber can be seen in part A of Figure 1. Cells and media are injected into the device; as the authors describe, once a single cell occupies a chamber, the flow in that chamber is reduced and consequently it is less likely that another cell flows into that chamber. The flow through the device is described by the previous wiki entry on this paper that can be found by clicking: http://soft-matter.seas.harvard.edu/index.php/Tracking_lineages_of_single_cells_in_lines_using_a_microfluidic_device. Essentially, via the bypass flow paths between chambers, the authors achieve a 70% 'seeding' of a single cell per chamber which only takes about 3 minutes; the remaining 30% of the chambers may have clusters of cells or no cells. The chamber geometry is such that it allows the cell to enter upstream of media flow but does not allow the cell to exit via a constriction at the downstream exit of the chamber. The chamber dimensions are designed based on the average cell size of the yeast cell, in theory this only allows for cell replication upstream in one dimension; this means the original cell will remain downstream while subsequent generations will be spawned upstream but are not free to leave the chamber due to the incoming flow that keeps them in. Cell 1.jpg

Figure 2 from the paper is more revealing of how the device works. The arrows in Part A of Figure 2 shows the direction of flow of the cell media. Notice the single cell in the chamber on the right; the geometry of the chamber in addition to the flow constrains the cell to this position and future generations to positions in the chamber upstream. Part B shows that there remains flow within the trapping chamber with or without a trapped cell; however, once a cell is trapped in the chamber, flow increases in the bypass which, as the authors note, decreases the possibility of another cell entering that chamber. This is the novelty of the device: loading of single cells into chambers without manual manipulation increases efficiency of the study beyond previous capabilities (this can be done, as mentioned, in 3 minutes). Flow based loading, although not 100% accurate as seen in Part D of Figure 2, is quite successful (70% success, which is better efficiency than predictions made by Poisson statistics (40%)). Taking a step back for a minute, we can appreciate the previous difficulties this device has overcome: cells are very small and the manual manipulation of a single cells is rather difficult. This device loads a single cell into a chamber that is only the width of a single cell (this is what I would consider impossible or close to impossible via manual manipulation or even robotics given the soft and nonuniform nature of cells).

Cell 3.jpg

Cell 2.jpg

The thumb image on the right (from Figure 3 of the paper) shows the lineage of single cell over time up the chamber.

Phenotype As a Function Generation: Paper Results

As discussed above, the inquiry into how cell phenotype or expression varies as a function of generation within a single cell lineage is poorly understood because of the complex nature of addressing the question: a single cell needs to be isolated and allowed to replicate and its progeny need to be organized so we can essentially identify "who's who" or which is the original cell and which are the progeny and how old they are. The authors, as summarized in the previous section, have developed a device for such a set up via the use of microfluidics and what I've termed "flow loading" of individual cell chambers. The expression level (phenotype) of three proteins was investigated. pPho84-GFP expression (a high-affinity phosphate transporter) was found to be maintained among the generations; as the authors note, this protein is found to be 'switched' either fully on or off in a cell and consequently switched on or off in progeny. The authors found this to be the case; in generations of a single cell, all expressed or none expressed this protein. However, the heat shock protein Hsp12-GFP was found to be expressed at various levels among the generations which suggests that the phenotype of the protein is not conserved among the generations and may be dictated by other factors. The "housekeeping" protein as the authors call it, Rps8b-GFP was found to be completely conserved and expressed throughout the generations; this is a ribosomal protein needed for the cell to function so these results are to be expected. The results of the three proteins are seen in Figure 4b below:

Cell 4.jpg Find the paper at: http://www.pnas.org/content/106/43/18149.full.pdf+html for a clearer image. The system can also be used dynamically as the authors describe toward the end of the paper. Results from this use of the device (via the use of MATLAB programing and dynamic analysis of the cell phenotypes is shown and discussed in the previous wiki entry on this paper: http://soft-matter.seas.harvard.edu/index.php/Tracking_lineages_of_single_cells_in_lines_using_a_microfluidic_device.

Connection to Soft Matter

Great! but how great...? As discussed above, the inquiry into how cell phenotype changes among the generations spawned from a single cell (so all have the same genetic makeup minus chance mutations) is a very interesting but complicated question to ask. The authors address this problem with their microfluidic device and via tracking the expression of three different proteins among cell generations, showed that some expressions are constant (proteins pPho84 and protein Rsp8b) while others fluctuate (protein Hsp12). As the authors point out, "Importantly, we observe that clusters form at all positions along the chambers, and that expressing cells may be adjacent to or upstream from non-expressing cells; if cell-cell communication by soluble factors determined protein expression patterns, cells downstream from or adjacent to expressing cells would consistently exhibit similar protein levels." So, the result of the Hsp12 seemingly random expression throughout the generations is further intriguing because many downstream cells did not express the protein suggesting the expression was not determined by communication but rather a generational phenomenon. This all sounds well and good but here's where things get hairy; while I beleive this device to be very clever and useful, any claims made about protein expression and phenotype need to be taken with some caution:

In class we discussed disjoining pressure and at the cellular level, this becomes significant. The chamber walls are constructed out of glass and PDMS; the glass certainly has a negative charge while the PDMS may have a negative or a positive charge (irrelevant really, so long as some portion of the chamber wall carries a charge). Within the chamber, there is the highly complex cell which carries a charge on its surface due to the numerous trans-membrane integrins and proteins. Since the chamber fits exactly one cell across, this cell and the chamber walls are very close. Using the Derjaguin approximation, we can consider the cell membrane and chamber wall as two flat plates approaching and if the we consider the glass which we know to be negative and the cell membrane to be negative, we have two surfaces with similar and constant charge densities approaching each other. Since the charge on the glass and the charge on the cell are fixed in the flat approximation, as the surfaces approach at a constant charge density the repulsion (disjoining pressure) goes to infinity as the surfaces get closer and closer. Since the chamber fits the cell so precisely, the distance between the two surfaces (cell-chamber wall) is very small resulting in a disjoining pressure which is very high. This pressure would squeeze the cell and cause elongation up the chamber. Perhaps there is a positive charge on the cell surface, there would then be strong attraction between the cell and the chamber wall causing potential tension within the cell. Either way we have a strong mechanical stress on the cell and as any tissue engineering lab would tell you, that's going to potentially influence the behavior (protein expression... phenotype!) of the cell. The authors present a great method of isolating a single cell and organizing its progeny to study phenotype but in doing so have created an environment that may cause substantial pressures on the cell. As I mentioned, this is assuming the surfaces (cell-chamber wall) approach each other at a constant charge density; if this isn't the case, and the surfaces approach at a constant potential, although the resulting pressure isn't as strong, it is still present and increasing as the surfaces approach. I wonder if these pressures influence the protein expression of the proteins examined by the authors. For example, the authors noted that the expression of Hsp12 seemed to vary among the generations. Was protein expression simply a generational variable or perhaps a function of the disjoining pressure experienced by the particular cell membrane and the surface of the chamber? If the chambers were larger, would protein expression be the same? Then again, if the chambers were larger, the purpose of the device (keeping track of the progeny) is lost. We always make assumptions so this may be one of them; this device provides a means of studying phenotype as a function of generation but, we must note that we assume the pressure experienced by the cells does not influence protein expression (ie that the phenotype observed are a function only generation).