You are here: Vale Lab Home Page>Research>Core Technologies>HT Microscopy
To analyze many different samples of stained cells (obtained, for instance,
after RNAi treatment with different dsRNA molecules) automated, unattended,
microscopy combined with some form of image analysis becomes a necessity. Automated
"high-throughput" microscopes consist of a motorized xy stage, a motorized
z-drive combined with some form of automated focusing, motorized shutters, filter
wheels and software that combines these components into a machine that can acquire
hundreds or even up to many hundreds of thousands images without human intervention.
Although it is possible to build such a system from off-the-shelve components
(see for instance Roy Wollman's description),
integration into a reliable entity can be quite challenging (albeit most certainly
not impossible). A number of commercial high-throughput systems are on the market
(for instance: Cellomics, GE, BD, Evotec). Our lab has used a Beckman IC-100
(discontinued), and now has a Molecular Devices ImageXPress
Micro. The ImageXPress micro has an extremely fast, voice-coil driven xy-stage
with very high precision. The microscope also has a reflection-based ("hardware")
autofocussing mechanism. We have not yet ventured into using immersion medium
(water or oil), but have been pushing the resolution obtainable with air objectives
using a (Nikon) 40x 0.95 na objective. Since the working distance of this objective
is short (~ 150 micron), it is imperative to use glass bottom dishes with a
low "skirt" (meaning that the glass is very close to the bottom of
the plate). The standard software that runs the ImageXPress micro is stupid
enough to lift the plate into mid-air whenever it fails to find the reflecting
interface, something that happens regularly and leads to loads of unusable images.
We therefore operate the system with a set of macros ("journals" in
Molecular Devices' language) that we wrote and restrict z-movements between
sites, but that "due to the nature of the macro language provided"
slow down the machine considerably.
The system is also equipped with a Thermo CRS robotic arm and plate hotels, allowing us to image (at least theoretically) up to 45 plates without human intervention. Data are streamed directly onto our Linux-based server through Gigabit Ethernet. Analysis is then carried out using multiple other computers running matlab scripts. The first major screen that we executed using this system was a screen for genes involved in metaphase spindle organization in Drosophila S2 cells.
updated 4/9/07
back to Home Page