Microfluidic High-Throughput Imaging and Automated Lineage Tracking of
Yeast
D. Falconnet*, M. Ricicova*, A. Quiring*, A. Niemistö**, R.J. Taylor*
,
**, T. Galitski**, I.
Shmulevich**, C.L. Hansen*
,
**
*Department of Physics and Astronomy and the Center for High-Throughput Biology, University of
British Columbia, Vancouver, BC V6T 1Z4
**Institute for Systems Biology, Seattle, WA 98103
Heterogeneity in cellular response is present across all organisms, from E. coli to stem cells,
and is a critical challenge to deciphering biological responses. Even cells with identical genotype
and cell history respond differently to stimuli. This variability is obscured in many conventional
biochemical assays which necessarily average the response of large ensembles of individual cells.
A particularly interesting source of cellular variation is epigenetic differences or heritable cell
states, possibly arising from correlated protein and mRNA levels between cells and their progeny.
Quantitative analysis of large numbers of single cells is critical to investigating these effects and
is well-suited to quantitative time-lapse microscopy. Technical challenges in conventional systems
include the need to track single cells through time, difficulty in reconstructing lineage relationships,
and crude temporal control over the chemical microenvironment. Here we describe a microfluidic-
based imaging platform and related image analysis pipeline optimized for such high-throughput
experiments with yeast. Using this system we investigate heritable sources of noise in the mitogen-
activated protein kinase (MAPK) cascade which governs pheromone response in Saccharomyces
Cerevisiae.
Microfluidic devices can be used to improve long-term cell imaging through confinement, to
precisely control chemical conditions in space and time, and to execute multiple parallel experiments
in a compact high-throughput format [1][2]. Our microfluidic system (Fig. 1) is constructed from
multiple layers of PDMS and contains 128 chambers, each capable of holding up to 1000 cells,
arranged in eight columns and sixteen rows [3]. Each column contains a single genotype and each
row a single chemical condition. The cells are immobilized in agarose gel (Fig. 2), allowing the
acquisition of high-resolution fluorescent images in three colors and the tracking of cells over up to
16 hours with time resolution of 20 minutes. The flow of cells and exchange of media is automated
using integrated micromechanical valves, enabling the programmable and dynamic modulation of
the microenvironment. Yeast strains are transformed with fluorescent proteins to allow for tracking
of transcriptional response and noise while further facilitating the robust association of daughters
with mothers. Custom MATLAB image analysis software is used to process over 60,000 images per
experiment in order to segment cells, track cells through time, construct lineages, and correlate this
to activity of the pheromone response pathway.
Preliminary investigations with this system reveal that a switch-like morphological response in yeast
pheromone signaling is correlated with the gene expression and that these responses are influenced
by lineage-dependent effects. Our system's precise chemical control, high throughput, and ability to
1044
doi:10.1017/S1431927610057466
Microsc. Microanal. 16 (Suppl 2), 2010
© Microscopy Society of America 2010
https://doi.org/10.1017/S1431927610057466
Downloaded from https://www.cambridge.org/core. IP address: 54.162.69.248, on 13 Jun 2020 at 14:36:57, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.