Notes on measuring protein expression using microscopy

Context: for the project on understanding how B. subtilis modulates growth rate, I measured the expression of a ribosomal protein (RplL) fused to superfolder GFP (a.k.a. sfGFP) as we varied the growth rate of B. subtilis, either through genetic perturbations or by changing the media the cells were growing in.

The protein expression, or the amount of protein that a cell has, gives important information about the phenotypic state of a cell. Protein expression can also vary over time or from cell to cell, and in order to capture the cell-to-cell or temporal heterogeneity, people often turn to microscopy: a) attach a fluorescent protein to your protein of interest and b) use microscopy to quantify how much fluorescent protein a cell has to infer how much of your protein of interest there is in the cell.

There are some great reviews about performing quantitative microscopy that far exceed what I am covering here, but I want to highlight two corrections that I find that people often fail to do that are applicable for measuring protein expression.

Flat-field correction

Source material

The fluorescence intensity that you measure being emitted from your fluorescent proteins will be some background plus some signal, where the signal is proportional to the density of fluorescent proteins and the intensity of excitation illumination light hitting the fluorescent proteins. Because what we care about is the density of fluorescent proteins, we need to account for both the uneven illumination across the field of view of the microscope and the background.

Although aligning the microscope optics can reduce the unevenness of the illumination, empirically even well aligned microscopes will still have uneven illumination. This can be corrected for computationally after acquisition as long as you have a good reference image (as described here).

The illumination pattern on the microscope that I did most of my PhD with (Nikon TI microscope, 60x objective). I measuring the illumination pattern by taking images of a saturated dye (fluorescein) solution.

Even when there is low background from the sample, there will always be some background in the measured image because most cameras are set with an offset value so that even when noise is introduced, the measured signal is positive. This is further described here.

Thus, in order to calculate what the density of fluorescent proteins ($I(x, y)$) in the sample is, we need our acquired image ($I_{image}$), a reference image ($I_{ref}$), and a background image ($I_{bkg}$) and compute the following:

$I(x, y) = \frac{I_{image}(x, y) - I_{bkg}(x, y)}{I_{ref}(x, y) - I_{bkg}(x, y)} \cdot \frac{\Sigma_{x, y}(I_{ref}(x, y) - I_{bkg}(x, y))}{xy}$

where $\frac{\Sigma_{x, y}(I_{ref}(x, y) - I_{bkg}(x, y))}{x y}$ is a scaling factor to ensure that $I(x, y)$ has similar values to $I_{image}$.

Maturation kinetics of the fluorescent protein

Source material

After being translated, fluorescent proteins (FPs) must undergo a stochastic maturation step in order to actually become fluorescent. The maturation for most FPs can be modeled with a single exponential and a characteristic half-time $t_{50}$ on the order of ~10 of minutes. For fast growing cells (e.g. bacteria) which have a doubling time on the order of 10s of minutes, the maturation kinetics end up mattering a lot.

Assuming that the FP is in balanced growth (i.e. in a given condition, the production of the FP is proportional to the size of the cell), that means that an appreciable fraction of the proteins will have been synthesized in the last $t_{50}$ minutes and many proteins will not be fluorescent. What’s more, if we want to compare the protein expression while we also modulate the doubling time (as I did when wanting to understand how B. subtilis modulates growth rate), then the fraction of mature fluorescent proteins will vary as a function of the doubling time. Specifically, the fraction of fluorescent proteins that are mature is given by:

$F_{mat} = \frac{1}{1 + \frac{t_{50}}{t_{gr}}}$

where $t_{gr}$ is the growth rate of the cells.

Luckily, a few years ago a lab down the hall from me published a paper where they characterized the maturation kinetics of most commonly used fluorescent proteins in living cells, which means that all we have to measure protein expression is measure the growth rate of the cells.

The interpretation of our experiments is qualitatively very different depending on whether or not we account for the maturation kinetics of the fluorescent protein, highlighting the importance of this correction if we want to compare protein expression in different conditions that have different growth rates.