Real-Time Image Analysis of Cells Undergoing Mitotic
Catastrophe
Prof. Michael Mackey (mackey@engineering.uiowa.edu)
Biomedical Engineering / Radiology
5013 Seamans Center
335-6058
Dr. Fiorenza Ianzini (fiorenza-ianzini@uiowa.edu)
Department of Radiology
3970 John Pappajohn Pavilion
384-8094
Introduction
Mammalian cells exposed to physical and chemical stresses usually
exhibit transient growth delays that lead to an accumulation of cells
in S and G2 phases of the cell cycle. Under these conditions, abnormal
levels of cyclin B1, a positive regulator of cell division, accumulate
in the cell, leading to premature cell division and ultimately
resulting in mitotic catastrophe (Mackey, 1993; Swanson et a.l,
1995;Mackey et a.l, 1996; Ianzini and Mackey, 1997; Ianzini et
al. 1999). Such untimely entry into mitosis is associated with a
previously uncharacterized form of DNA damage that is thought to lead
to chromosome instability and may be an initiating event in cell
transformation (Ianzini and Mackey, 1998). Although mitotic catastrophe
is generally a lethal process, the question still remains whether a
small fraction of cells undergoing this process may survive, thus
potentially contributing to the destabilization of the genome that is
reflected in chromosomal instability. Experiments are underway to
determine the fate of cells undergoing mitotic catastrophe following
radiation exposure using the Large Scale Digital Cell Analysis System
(LSDCAS) at the newly established Real-Time Cell Analysis Facility here
at the University of Iowa. LSDCAS is a computer-controlled microscope
system that is capable of automatically generating digital movies of
over 1000 separate microscope fields over a three-week interval
following treatment. LSDCAS is a joint Biomedical Engineering /
Radiology venture, where Dr. Ianzini's lab in Radiology contains the
microscope systems (and is in fact a full cell culture and biochemistry
lab) and analysis is performed using Unix computers in Biomedical
Engineering. For this purpose, a high performance, dual processor
Compaq AlphaServer running Tru64 Unix located in Biomedical Engineering
will be useful in the analysis of images generated by LSDCAS. In
addition, four Compaq XP1000 workstations are located in Dr. Mackey's
newly renovated lab in Biomedical Engineering and are dedicated to
software development for LSDCAS.
Problem Statement
The frequency of radiation-induced chromosome instability has been
estimated to be about 1:3300 of the treated cell population after
exposure to 8 Gy of X-rays, a treatment which usually kills about 98%
of the cells in the population. Note that death in this system is
defined in terms of the ability to form a cell colony of more than 50
cells. In order for mitotic catastrophe to contribute to chromosome
instability, the surviving fraction of cells undergoing this event must
be no less than .015 per surviving cell, thus requiring that 33,000
single cells must be analyzed to reliably detect 10 cells surviving
mitotic catastrophe, under the assumption that cells undergoing this
process become unstable. This rather large number of cells would
require about 33 cells to be analyzed in each of 1,000 microscope
fields; it is expected that, on average, about two out of three fields
will yield a colony during the three-week interval following
irradiation. LSDCAS can acquire the image data for these studies, yet
the volume of data produced is overwhelmingly difficult to analyze.
This project will provide for automatic segmentation of individual
cells in the microscope field images of control and irradiated cell
cultures. First, entry of cells into mitosis must be accurately
detected. Following cell division, each daughter cell must be
identified and analyzed for subsequent cell division. The results of
these analyses must then be presented in an easily comprehended form.
Suggested Methods
Software currently exists in LSDCAS for the automatic segmentation of
individual cells in microscope fields. The need here is for an
algorithm that can determine the success of cell division for a
particular cell. It is expected that a multi-threaded approach toward
this problem will aid in the achievement of the project goals. Since
the criterion used for cell survival is the ability to form a 50-cell
colony, the number of threads that must be dispatched for the analysis
of a single cell will not be a burden on the multi-processing
AlphaServer used in this project. Thus, an algorithm can be developed
to track a single cell from one cell division through the next. Then,
if two daughter cells are produced, two new threads of execution (using
the same algorithm) can be dispatched to follow the daughter cells,
with termination of the thread associated with the previous cell
generation. Each cell generation can then be described by the results
of the analysis performed by its corresponding thread. Cell division
can be detected by simultaneously monitoring cell shape (defined as the
perimeter / area ratio) and mean cell pixel intensity, as cells round
up and become bright at cell division in the phase contrast microscope
images. Following cell division, each daughter cell can be identified
through a statistical analysis of the segmented dividing cell:
successfully dividing cells present an hourglass shape prior to
cytokinesis, at which time the two daughter cells separate and
establish their own boundaries. Cells undergoing mitotic catastrophe
typically round up, vibrate, and then spread out without undergoing
cytokinesis. Later, a destabilization of the nucleus occurs, leading to
a fragmented texture in the image. Sometimes, cells undergoing mitotic
catastrophe temporarily divide, yet later fuse, forming a
multinucleated cell. Further, cells undergoing mitotic catastrophe that
die present a dramatic cell lysis that can also be distinguished by
abrupt changes in mean pixel intensity. All of these features can be
used to distinguish between cells undergoing mitotic catastrophe, cells
dividing normally, and cells that die. The cell division data so
obtained can be organized into a cell pedigree, which is a binary tree
used to depict the fate of cell progeny following a particular
treatment.
Expected Results
It is expected that this project will provide the means towards
obtaining an accurate estimate of the probability of cell death
following a variety of toxic treatments. Further, data on the
survivability of cells undergoing mitotic catastrophe will rigorously
test the hypothesis that a small fraction of cells undergoing this
event may survive and contribute to the later appearance of chromosomal
instability, a finding that would establish the importance of this
mechanism as a potential initiator of cell transformation.
References