Real-time RNA profiling within a single bacterium

From Soft-Matter
Jump to: navigation, search

Original entry: Zach Wissner-Gross, APPHY 226, Spring 2009

Information

Real-time RNA profiling within a single bacterium

Thuc T. Le, Sebastien Harlepp, Calin C. Guet, Kimberly Dittmar, Thierry Emonet, Tao Pan, Philippe Cluzel

PNAS, 2005, 102, 9160-9164

Soft matter keywords

Diffusion, fluorescence correlation spectroscopy

Summary

Figure 1: Varying RNA levels in a single cell (A) and in a number of cells (B). When these levels are averaged (C), the cell cycle-dependent variations disappear, and the RNA levels seem to be constant in time.

While every cell in an organism has essentially the same genetic code, cell morphology and behavior vary due to which genes are expressed. Studying intracellular RNA or protein levels represents a common method for studying changes in cell behavior. However, classical methods for for observing temporal changes in these intracellular concentrations require lysing cells at different time points.

Cluzel and coworkers offer a different approach that both yields single-cell resolution and is real-time (the authors estimate their acquisition time to be 2 seconds), without noticeably affecting the cell during the observation. Using a high-NA objective, the authors focus a blue laser beam to a diffraction-limited spot inside the bacterium, and then perform fluorescence correlation spectroscopy (FCS), which I explain below in more detail.

In short, the authors first show that their FCS data matches that of the current gold standard technique using 260 nm UV absorption (Figure 2B), after which they go on to demonstrate a few interesting findings. For example, they found that while the RNA levels of individual cells vary significantly over time, the averaging of several cells reduces this noise (Figure 1), explaining why such variation has gone largely unnoticed within the scientific community.

Soft matter example

Figure 2: Of note here are A and D, which are fits of the autocorrelation fit function to in vitro (A) and in vivo (D) data. The three plots within A and the three plots within D are self-consistent in their use of single values of <math>\tau_{\text{free}}</math> and <math>\tau_{\text{bound}}</math>, but the two graphs do not share the same values. Also shown is the excellent correspondence between the FCS technique described here and UV sprectrophotometry in measuring RNA levels.

While this paper had no mention of wetting, surfaces, or capillary forces, the technique of fluorescence correlation spectroscopy is probably worth mentioning, as it is employed specifically in soft matter experiments. FCS is typically used to measure diffusion constants of fluorescent molecules in solution.

The number of fluorescent molecules in the focal spot will follow a Poisson distribution, and the relative change in fluorescence over time with therefore vary both with the number of fluorescent particles (which determines the amplitude of the autocorrelation function) and the rate of diffusion of these particles (which determines the decay rate in the autocorrelation function). So if particles diffuse more slowly, then the fluorescence changes more slowly, and the correlation in fluorescence will apply over longer periods of time.

The authors apply this principle of FCS to a two-state system: the MS2-GFP protein, where MS2 is the domain that will bind to a specific RNA target, and GFP is green fluorescent protein, a fluorescent marker. When the MS2-GFP protein binds to RNA(or in a living bacterium, to RNA that is associated with the 70S ribosomal subunit), the GFP is now attached to a bulkier set of molecules, and its diffusion rate drops. Thus, by carefully analyzing the autocorrelation function, the authors are able to determine what fraction of MS2-GFP proteins are bound to RNA (assuming an excess of the protein, so that there's no kinetics involved).

We could stop the story here, but I'll quickly just introduce the function to which the authors fit their autocorrelation data and include a figure showing how nicely everything matches. The expression they use for the autocorrelation function <math>G(t)</math> is:

<math>G(t)=\frac{1}{N}\frac{1}{1+y^2}\left(\frac{1-y}{1+t/\tau_{\text{free}}}+\frac{4y}{1+t/\tau_{\text{bound}}}\right) </math>

where <math>N</math> is the number of fluorescent particles in the focal spot, <math>\tau_{\text{free}}</math> and <math>\tau_{\text{bound}}</math> are the characteristic decay times (in the autocorrelation function) for free and bound MS2-GFP, respectively, and <math>y</math> is the fraction of MS2-GFP that is bound. The authors then interpret the relative RNA level as being <math>Ny</math>. (Note: The factor of 4 in the numerator of the second term may seem surprising, but relates to the fact that RNA has four MS2-GFP binding sites.) See Figure 2 for examples of how this function fits data for multiple values of <math>y</math>.