The aim of the “Primary Screen” is to find and locate candidates or structures of interest. This is achieved with low resolution, low optical sectioning performance and with the aim of altering the sample as little as possible. On the other hand, the primary scan has to be performed at maximum scan speed so as not to miss a short event in a large number of objects. Typically, samples for screening grow in multi-well plates. These are polymer stage inserts containing a number of cavities (here named “wells”). In each well a single individuum or a small number of e.g. cells is usually growing. The grid structure of the well plate allows easy object relocation and change monitoring. To fit to varying sample sizes, well plates with 6, 24 or 96 wells, etc. are available. For example, zebra fish embryos at different stages of development are typical objects analyzed in multi-well plates (where one fish embryo is seeded into one well). The aim of Primary Screen is to get sufficient information about every well at appropriate time resolution. This information is then used to recognize and decide which samples in which wells are of interest and should be investigated further in a so called Secondary Screen (see below).
Primary screening is performed by observing large fields of view. Sometimes a camera is suitable for this task. Confocal provides better contrast, modern systems offer low magnification optics for imaging a complete well at once. Confocal imaging at large pinhole settings is sufficient (this allows out-of-focus events to be found, too). Two-channel recording is also helpful, such as a fluorescence image and a transmitted light image simultaneously. All parameters needed for a primary scan are stored in a “scan job”, which is retrievable for continuing the screening process after analyzing a rare event – or for starting a new experiment the next day.
Before starting a matrix screen experiment, the system needs to know the start coordinates and the distances of the wells in x- and y-direction. As wells are aligned in a regular Cartesian pattern, the system can then automatically move from well to well, either in a regular manner from upper left to lower right, or in a predefined sequence, where all wells are addressed after manual selection by the operator. Different scan jobs can be created and freely combined at multiple positions. For reproducibility these screens, xy coordinates as well as all other relevant imaging scan parameters are storable in a template. The advantage of storing parameters in templates is obvious: if you exchange the multi-well plate, only the start coordinate has to be readjusted.
This automation saves a significant amount of time; more experiments can be run in a shorter time in a standardized way.
During the primary screen, one of the individuals will occasionally start to modify its shape or other optical properties (e.g. fluorescence intensity, color ratio and so on). A classical example is the change from resting nuclei (G-phase) to dividing nuclei (mitosis). Here, a nuclear stain will significantly alter in shape and intensity.
One possibility to investigate details of the mitosis is to inspect all objects by watching the images on the monitor and manually indicate which well should be further investigated in greater detail. In many cases, this procedure is already a very powerful tool to identify and follow up rare events.
These days, modern software can often take the place of human recognition skills. Currently, pattern recognition software is available that can automatically identify morphologies like spheres, lines, and a huge toolbox of various complex patterns. The images from the primary screen are transferred online to an image analysis computer, parallel processed and evaluated while the screening process is continued. The pattern recognition rules can be modified at any time in order to adapt to experimental changes or to new ideas and research targets. As computers are network connected, monitoring and modification of rule sets might even be done from the home office, including remote control of the microscope.
In essence, the identification process yields a coordinate in the multi-well plate which marks the position of the rare event sought after. This coordinate is then transferred back to the system.
When an event of interest is found, the task is to record and analyze the structural and temporal changes in great detail. For this purpose, the system has to center the object of interest and perform a predefined data acquisition schedule; usually at high magnification and high resolution. Centering the object within the well that has been found to contain the rare event is called “single object tracking function”. This is a separate job which identifies the exact coordinates inside the well and moves the stage in order to center these coordinates for further data acquisition. This is also important as the objects may move during the experiment. It is also possible to center interactively and then activate the high-precision acquisition by a single mouse click. For other tasks, a series of identified positions may be stored and then recorded automatically.
The high precision recording (“secondary screen”) is done at high magnification and with high pixel resolution. Here, an optical vario-system (zoom) is very helpful, as it allows the magnification to be altered without mechanically changing the objective lens and is not prone to decentration. All parameters for the microscope are adapted automatically with the magnification change (e.g. pinhole diameter). In addition to centration in x and y, the correct z-position of the object must be found and recognized. For this purpose, an autofocus routine is implemented.
Other parameters for high-precision scanning can be pre-stored in a scan job. These parameters include, for example, scan speed, scan format, spectral properties, scan mode, repetition and so on. A very typical record at high definition includes a three-dimensional z-stack that allows analysis of structural changes and interactions during the rare event.
With the possibility to define and combine different scan jobs, a macro function is available which allows complex experiments to be performed at a single position. Additionally, the experiments can be assigned to various positions and looped for long-term high definition observation.