mRNA detection in budding yeast with single fluorophores

Abstract

Quantitative measurement of mRNA levels in single cells is necessary to understand phenotypic variability within an otherwise isogenic population of cells. Single-molecule mRNA Fluorescence In Situ Hybridization (FISH) has been established as the standard method for this purpose, but current protocols require a long region of mRNA to be targeted by multiple DNA probes. Here, we introduce a new single-probe FISH protocol termed sFISH for budding yeast, Saccharomyces cerevisiae using a single DNA probe labeled with a single fluorophore. In sFISH, we markedly improved probe specificity and signal-to-background ratio by using methanol fixation and inclined laser illumination. We show that sFISH reports mRNA changes that correspond to protein levels and gene copy number. Using this new FISH protocol, we can detect >50% of the total target mRNA. We also demonstrate the versatility of sFISH using FRET detection and mRNA isoform profiling as examples. Our FISH protocol with single-fluorophore sensitivity significantly reduces cost and time compared to the conventional FISH protocols and opens up new opportunities to investigate small changes in RNA at the single cell level.

Figure 1.

Comparison between single-probe FISH (sFISH) and multi-probe FISH (mFISH). yEVenus mRNA, which is 717-nt long, is probed by sFISH and mFISH. (A) Probe configurations are shown from left to right for 30 probe mFISH, sFISH and sFISH with FRET. For mFISH, we use a set of thirty Quasar 670 end-labeled probes. In sFISH, we use a single short Cy5-labeled DNA oligo probe. For FRET experiments the first Cy5-labeled probe is used in conjunction with a Cy3-labeled probe. (B) For mFISH, we use a conventional epi-fluorescence microscope setup (top). In epi-illumination, the beam is aligned along the optical axis and illuminates cells across their entire height. The difference in the beam intensity profile between the two geometries is highlighted by varying shades of red. For sFISH, we use highly inclined illumination geometry (bottom), which markedly increases the signal-to-noise ratio (SNR). This light sheet also travels with the objective, which allows imaging different planes along the vertical axis (z-sectioning). (C) The images shown are of the same field of view taken with epi-illumination (top) and then subsequently with inclined illumination (bottom) using the same laser power. The bottom image taken with inclined illumination exhibits more intense spots and lower background. (D) Comparison of spot signal-to-noise ratio (SNR) between epi- and inclined illumination. The SNR values measured with inclined illumination is plotted against those measured with epi-illumination. Most spots are found above the red line y = x, which indicates inclined illumination produces higher SNR than epi-illumination. The increase in signal to noise ratio is a factor of 2.15 on average.

Figure 2.

Correlation between sFISH spots and protein expression level. (A) Fluorescence images of single yeast cells expressing YFP (top row) and sFISH signals from Cy5-labeled probes targeting YFP mRNA (bottom row). Shown from left to right are fluorescence images of the negative control (no YFP expression), low YFP expression, and high YFP expression. Fluorescence intensities in the YFP channel and Cy5 channel are represented by false yellow and red colors, respectively. YFP images are from formaldehyde fixed cells, and Cy5 sFISH images are from methanol fixed cells. (B) Correlation plot. The mean number of FISH spots is plotted versus the mean yEVenus expression level. The error bars are measures of the standard deviation. (C) sFISH spots versus ploidy. sFISH was performed on yeast strains with four different ploidies (1n, 2n, 3n, 4n). The error bars show the standard deviation of the data. The number of spots detected per cell increases monotonically with the number of copies.

Figure 3.

Comparison of spot quality between formaldehyde (white) treated samples and methanol (blue) treated samples. (A) sFISH spots detected from the negative control strain. On average, there is ∼0.3 spots per cell in the methanol treated cells (blue) compared to ∼3.1 spots per cell in the formaldehyde treated sample (white). (B) Comparison of signal-to-noise ratio (SNR) of single probes. SNR of a single Cy5 was calculated from fluorescence time traces that captured single-step photobleaching events. Signal is obtained from the single-step drop in fluorescence intensity upon photobleaching, and the noise is calculated as the standard deviation of the signal prior to photobleaching. The histogram shows that the spots from methanol-treated cells (blue) have ∼2-fold higher SNR than those from formaldehyde-treated cells (white). (C) Comparison of Cy5 stability. The population decay curves show that sFISH spots in formaldehyde treated cells photobleach faster than those in methanol-treated cells. (D) Comparison of probe number per spot. The number of probes per spot was determined by counting the number of photobleaching steps in the fluorescence time trace. When a single probe was used, most spots photobleached in a single step regardless of the fixative of choice (left). In comparison, when five (middle) or 30 (right) probes targeting the same mRNA were used, more probes were detected from spots in methanol-treated cells than in formaldehyde-treated cells. For the methanol samples treated with multiple probes (middle and right panels), binomial distribution fits are shown in red. For the five-probe experiment (5-probe FISH), Cy5-labeled probes and inclined illumination are used; whereas, for the thirty-probe experiment, Quasar-labeled dyes and epi-illumination are used.

Figure 4.

Estimation of the hybridization efficiency of single probes. (A) The effect of varying the number of probes. The histograms of the number of spots detected per cell are plotted for 1-probe (blue) and 5-probe (white) FISH. (B) Spot intensity vs. probe number. The mean spot intensity increases linearly with the number of probes as expected from the binomial distribution. (C) Spot number versus probe number. The mean number of spots detected per cell (y) increases with the number of probes (x). The fit model is y = N(1 − (1 − p)^x) where N is the true copy number, and p is the hybridization rate for a single probe. p is extracted to be 53%.

Figure 5.

Applications of sFISH. (A) Demonstration of FRET-FISH in yeast. Fluorescence image acquired under 532-nm excitation was split into two half images based on the emission wavelength. In each image, the green half on the left is from the Cy3 emission channel, and the red half on the right is from the Cy5 emission channel. The images shown represent cells treated with both Cy3- and Cy5-probes (left), Cy3-probe only (middle), and Cy5-probe only (right). Bright, punctate spots were observed in the Cy5 channel only when cells were treated with both probes (left). (B) sFISH for mRNA isoform detection. The schematic on the left depicts alternative transcription initiation sites (arrows) at the RGL1 locus, which lead to mRNA isoforms with different lengths. Transcription from the first site produces a full-length mRNA, while from the second site produces a truncated isoform. Using sFISH with two separate probes, the relative fractions of these isoforms can be measured. Probe 1 targets the longer isoform only, whereas probe 2 targets both. The bar plot on the right shows the ratio of sFISH signals with probe 2 to probe 1 measured with glucose (left) or galactose (right) growth media. Here, the mean total fluorescence intensity per cell was used as a proxy for sFISH signal because transcription level was too high to count individual spots.