By Salvatore Salamone
July 20, 2005 | Studying early-stage embryos of fruit flies can give researchers great insight into genetic-level interactions that determine the ultimate development progress of the fly. But the process of extracting the information has traditionally been time-consuming and tedious, resulting in small sample sets on which to understand the pathway interactions.
John Reinitz, a member of the faculty in the Department of Applied Mathematics and Statistics at Stony Brook University, and his team have developed a high-throughput method to measure gene expression and protein levels in fruit fly (Drosophila) embryo nuclei to vastly expand the amount of data used in developmental studies.
The method is essentially an automated pipeline that captures confocal microscope images of the embryos and performs an image analysis of the individual nuclei to derive information about protein levels. These data are then used to validate and calibrate in silico models, which in turn are used to better understand pathway interactions.
“To get the data, we had to develop our own techniques,” Reinitz says. “We developed software to do data visualization and analysis to get better data to test against an analytic model.”
|FRUIT BUZZ: Confocal microscope |
image of a fruit fly embryo shows
nuclei investigated to determine
gene expression levels in
To put the impact of the automation into perspective, Reinitz’s group has information from more than 900 embryos in its database to conduct the research. In the past, some scientists would do studies based on a handful of embryos. The difference is the data extraction — in the past it was mostly done by hand.
For example, researchers frequently used a number of manual techniques to view the nuclei, such as projecting the images on a wall or drawing nucleus-sized tick marks on paper and sticking that paper to the ground glass of a slide previewing box. The researcher would then visually estimate the image density within a nucleus (the image density is related to gene expression within the nuclei).
The analysis method used by Reinitz and his group automates the image analysis and quantifies the data, thus eliminating the need to do eyeball estimations.
Early-stage fruit fly embryos offer a great opportunity to study pathway systems involved in Drosophila development. By studying the gene expression and protein-level patterns within embryos of different ages, researchers can technically derive information about the developmental process.
“In the early stage, you can see signs of segmentation,” Reinitz says. “You can watch the genes, get some understanding of cell-to-cell communications, and look at the proteins that control transcription.”
During development, a Drosophila embryo goes through distinct stages based on its age. Each stage is characterized by what is called a rapid nuclear division. Following several of these divisions, most of the nuclei move to the surface of the embryo — so basically, the embryos are shaped like a hollow football. Studying the protein levels of these nuclei on the outer shell at various stages gives researchers information about the embryo’s development.
The process used to study the expression of segmentation genes involves staining embryos with fluorescent antibodies and then viewing them under a confocal microscope where the embryos are illuminated by three different-frequency light sources.
The confocal microscope images are captured and saved. Using VisiQuest image analysis algorithms from AccuSoft, the images are run through a process that identifies the embryo and then orients it so that the long axis of the football-shaped embryo is horizontal.
Then additional imaging techniques are iteratively applied to identify the distinct nuclei within the embryo. Essentially, the old manual process of drawing a circle around each nucleus and then estimating its image density is replaced by sophisticated image analysis. The software identifies the edges of each nucleus and then determines the pixel density within that nucleus. This information determines the expression level of segmentation genes, which is considered to be proportional to protein concentrations.
The automation of this work has resulted in a high-throughput system that gives researchers much more information about an embryo’s development than was previously available. The data are being used in in silico studies of gap gene regulation.
Much of the data is available in a public database called FlyEx (http://flyex.ams.sunysb.edu/FlyEx/). Researchers using these data have already published several papers where the data were used to calibrate and validate in silico gene circuit models.