Mats Kvarnström

Image analysis algorithms for cell contour recognition in budding yeast

Quantification of protein abundance and subcellular localization dynamics from fluorescence microscopy images is of high contemporary interest in cell and molecular biology. For large-scale studies of cell populations and for time-lapse studies, such quantitative analysis can not be performed effectively without some kind of automated image analysis tool. Here, we present fast algorithms for automatic cell contour recognition in bright field images, optimized to the model organism budding yeast (Saccharomyces cerevisiae). The cell contours can be used to effectively quantify cell morphology parameters as well as protein abundance and subcellular localization from overlaid fluorescence data.

Heterogeneous kinetics of AKT signaling in individual cells are accounted for by variable protein concentration

In most solid cancers, cells harboring oncogenic mutations represent only a sub-fraction of the entire population. Within this sub-fraction the expression level of mutated proteins can vary significantly due to cellular variability limiting the efficiency of targeted therapy. To address the causes of the heterogeneity, we performed a systematic analysis of one of the most frequently mutated pathways in cancer cells, the phosphatidylinositol 3 kinase (PI3K) signaling pathway. Among others PI3K signaling is activated by the hepatocyte growth factor (HGF) that regulates proliferation of hepatocytes during liver regeneration but also fosters tumor cell proliferation. HGF-mediated responses of PI3K signaling were monitored both at the single cell and cell population level in primary mouse hepatocytes and in the hepatoma cell line Hepa1_6. Interestingly, we observed that the HGF-mediated AKT responses at the level of individual cells is rather heterogeneous. However, the overall average behavior of the single cells strongly resembled the dynamics of AKT activation determined at the cell population level. To gain insights into the molecular cause for the observed heterogeneous behavior of individual cells, we employed dynamic mathematical modeling in a stochastic framework. Our analysis demonstrated that intrinsic noise was not sufficient to explain the observed kinetic behavior, but rather the importance of extrinsic noise has to be considered. Thus, distinct from gene expression in the examined signaling pathway fluctuations of the reaction rates has only a minor impact whereas variability in the concentration of the various signaling components even in a clonal cell population is a key determinant for the kinetic behavior.

Continuous light exposure causes cumulative stress that affects the localization oscillation dynamics of the transcription factor Msn2p.

Light exposure is a potentially powerful stress factor during in vivo optical microscopy studies. In yeast, the general transcription factor Msn2p translocates from the cytoplasm to the nucleus in response to illumination. However, previous time-lapse fluorescence microscopy studies of Msn2p have utilized a variety of discrete exposure settings, which makes it difficult to correlate stress levels and illumination parameters. We here investigate how continuous illumination with blue light, corresponding to GFP excitation wavelengths, affects the localization pattern of Msn2p-GFP in budding yeast. The localization pattern was analyzed using a novel approach that combines wavelet decomposition and change point analysis. It was found that the Msn2p nucleocytoplasmic localization trajectories for individual cells exhibit up to three distinct and successive states; i) Msn2p localizes to the cytoplasm; ii) Msn2p rapidly shuttles between the cytoplasm and the nucleus; iii) Msn2p localizes to the nucleus. Many cells pass through all states consecutively at high light intensities, while at lower light intensities most cells only reach states i) or ii). This behaviour strongly indicates that continuous light exposure gradually increases the stress level over time, presumably through continuous accumulation of toxic photoproducts, thereby forcing the cell through a bistable region corresponding to nucleocytoplasmic oscillations. We also show that the localization patterns are dependent on protein kinase A (PKA) activity, i.e. yeast cells with constantly low PKA activity showed a stronger stress response. In particular, the nucleocytoplasmic oscillation frequency was found to be significantly higher for cells with low PKA activity for all light intensities.

Identification of the three-dimensional gel microstructure from transmission electron micrographs.

Mass transport in gels depends crucially on local properties of the gel network. We propose a method for identifying the three‐dimensional (3D) gel microstructure from statistical information in transmission electron micrographs. The gel strand network is modelled as a random graph with nodes and edges (branches). The distribution of edge length, the number of edges at nodes and the angles between edges at a node are estimated from transmission electron micrographs by image analysis methods. The 3D network is simulated by Markov chain Monte Carlo, with a probability function based on the statistical information found from the micrographs. The micrographs are projections of stained gel strands in slices, and we derive a formula for estimating the thickness of the stained gel slice based on the total projected gel strand length and the number of times that gel strands enter or exit the slice.

Development of automatic image analysis algorithms for protein localization studies in budding yeast

Microscopy of fluorescently labeled proteins has become a standard technique for live cell imaging. However, it is still a challenge to systematically extract quantitative data from large sets of images in an unbiased fashion, which is particularly important in high-throughput or time-lapse studies. Here we describe the development of a software package aimed at automatic quantification of abundance and spatio-temporal dynamics of fluorescently tagged proteins in vivo in the budding yeast Saccharomyces cerevisiae, one of the most important model organisms in proteomics. The image analysis methodology is based on first identifying cell contours from bright field images, and then use this information to measure and statistically analyse protein abundance in specific cellular domains from the corresponding fluorescence images. The applicability of the procedure is exemplified for two nuclear localized GFP-tagged proteins, Mcm4p and Nrm1p.