Quantitative single-molecule microscopy reveals that CENP-ACnp1 deposition occurs during G2 in fission yeast

The inheritance of the histone H3 variant CENP-A in nucleosomes at centromeres following DNA replication is mediated by an epigenetic mechanism. To understand the process of epigenetic inheritance, or propagation of histones and histone variants, as nucleosomes are disassembled and reassembled in living eukaryotic cells, we have explored the feasibility of exploiting photo-activated localization microscopy (PALM). PALM of single molecules in living cells has the potential to reveal new concepts in cell biology, providing insights into stochastic variation in cellular states. However, thus far, its use has been limited to studies in bacteria or to processes occurring near the surface of eukaryotic cells. With PALM, one literally observes and ‘counts’ individual molecules in cells one-by-one and this allows the recording of images with a resolution higher than that determined by the diffraction of light (the so-called super-resolution microscopy). Here, we investigate the use of different fluorophores and develop procedures to count the centromere-specific histone H3 variant CENP-ACnp1 with single-molecule sensitivity in fission yeast (Schizosaccharomyces pombe). The results obtained are validated by and compared with ChIP-seq analyses. Using this approach, CENP-ACnp1 levels at fission yeast (S. pombe) centromeres were followed as they change during the cell cycle. Our measurements show that CENP-ACnp1 is deposited solely during the G2 phase of the cell cycle.

a single mEos2 fluorophore generates on average 1.98 localizations due to blinking (see The ability to detect proteins in single-molecule PALM imaging depends on various experimental factors. We assume that all molecules of mEos2-CENP-A Cnp1 carry a functional fluorescent protein and that the photo-conversion of mEos2 is almost quantitative (13).
However, we cannot exclude photo-activation or photo-bleaching of a subset of mEos2 prior to imaging, although care was taken to minimize the exposure of the samples to light.
During the imaging experiment, we applied low intensity UV laser irradiation to avoid photoactivation of more than one fluorophore per diffraction-limited area. In the later stages of an experiment, we made sure that we had photo-activated and recorded all the mEos2 fluorophores in a cell by gradually increasing the UV laser intensity and imaging for a sufficiently long time (Supplementary Figure S3D).
In control experiments with wild type S. pombe, lacking any mEos2 protein, we determined that the number of background single-molecule localizations per square micron was less than 2 2 (Supplementary Figure S3B), which translates to a contribution of less than 0.5 singlemolecule localizations per CENP-A Cnp1 cluster (assuming a maximum area of 0.5 x 0.5 µm 2 observed in our experiments). We also determined that the number of cellular mEos2-CENP-A Cnp1 single-molecule localizations per square micron (outside the clusters) was less than 3 (Supplementary Figure S3E), which contributes less than 0.75 localizations per cluster.

B) Post data image analysis
All calculations and analysis of cell length, cluster position, number of localizations/molecules etc., were carried out using ImageJ, Origin (OriginLab) and Microsoft Excel software.

Cell cycle experiments
Cell cycle block-release was performed using a temperature sensitive cdc25-22 mutant. Cells were grown in complete media (YES) at (permissive temperature) and the cell cycle block at G2/M phase was imposed by incubating the culture at 36°C (restrictive temperature) for 3 hours. This block was released by shifting the culture to 25°C. Samples were taken at 20 minute intervals, and the septation index was determined prior to fixation with 1% formaldehyde. Samples were then prepared for ChIP.

Quantitative Real-Time PCR
Quantitative PCR (qPCR) was performed using a LightCycler® 480 Real-Time PCR System (Roche) according to the manufacturers instructions with the LightCycler® 480 SYBR Green I Master mix (Roche). Primers sequences are described in Supplementary Table S2. All measurements were replicated 2-4 times. [28].
Note that the C-terminal tagged cnp1-GFP (S65T) strain displays slower growth at elevated temperatures, as previously reported [28]. To identify single molecule localizations each frame of the imaging movie is analysed separately. Initially, the frame is smoothed and the localization candidates are identified by a non-maximum suppression routine, then stored in a candidate list, which is sorted by photon intensity (1,2,3...n). Candidates (starting from 1) are then fitted using the original, non-blurred 8 image with a Gaussian. The fit is checked by verification parameters (intensity threshold, symmetry of gaussian, sigma etc.). If accepted, the fit is stored in a localization list and the next candidate is analysed using this fitting and verification routine. As low intensity candidates are more likely to fail the routine (35), we apply a x thres step to the routine. If more than x thres (normally set to 10) consecutive candidates do not pass the fit and verification routine, the remaining candidates on the list (containing lower intensities) are discarded and the next frame is analysed. Once all the frames of a movie are analysed a tracking filter is