High-Throughput In Situ Hybridization
All steps required for the ISH proper are carried out in an automated fashion. Cryostat sections are placed on a standard microscope slide that is subsequently incorporated into a flow-through chamber (Fig. 1A). The chamber is placed into a temperature-controlled rack and solutions required for pre-hybridization, hybridization, and signal detection reactions are added to the flow-through chamber with an automated solvent delivery system (Fig. 1B). Because of such automation, up to 192 slides can be processed in parallel allowing a daily throughput of hundreds of sections. Although this equipment can be used for hybridization with radioactive probes its power comes to the fore when hybridization is carried out with epitope-tagged riboprobes (e.g. digoxygenin) that are detected with a biotin-tyramine-based amplification step (e.g . Kerstens et al.,1995). This detection method exhibits sensitivity comparable to that obtained with radioactive riboprobes but has the benefit of providing cellular resolution and increased speed (Fig. 1D, E and G).
Figure 1. Equipment developed for automated in situ hybridization and illustration of typical results.
Flow-through hybridization chamber (200 µl volume) composed of a microscope slide loaded with tissue sections, a pair of two 80-µm thick spacers, and a 5-mm thick glass plate into which a depression has been cut. Slide, spacers and glass plate are fastened into a metal frame with a bracket. Depression and slide form a well into which solutions are delivered by a pipetting robot. Solutions initially flow through the chamber but once the well is drained, solutions are retained in the narrow slit of the chamber as a result of capillary forces.Hybridization and detection chemistry are performed on a Tecan Genesis pipetting robot platform equipped with racks for the hybridization chambers (1), reagent containers (2), and a system for controlling the temperature of the racks (3). Solutions are added with a set of 8 parallel pipettes (4).A Leica microscope equipped with a motorized stage and a CCD camera.Sagittal brain section of an adult mouse showing the expression of the tyrosine hydroxylasegene which is expressed at numerous sites including olfactory bulb (OB), striatum (ST), cortex (CX), substantia nigra (SN), and cerebellum (CB). To visualize expression, digoxygenin (DIG)-tagged riboprobes are first detected with a antiDIG antibody to which peroxidase is coupled. This enzyme is used to activate a tyramine-biotin conjugate which thereby gets covalently attached to proteins in the vicinity of the antiDIG antibody. Subsequently, biotin is detected with a steptavidine-alkaline phosphatase-based color reaction. The grid defines 32 images individually captured and then assembled into the composite. Each image was collected at 50 fold magnification and has a resolution of 3.3 µm/pixel.Blow-up of area boxed in D; this image has the same resolution as that in D.Expression of secreted frizzled related protein 2 (sFRP2) in the midbrain (MB) detected with a radioactive probe.Expression of sFRP2 in the midbrain (MB) detected with a digoxygenin-tagged probe. Note the size of silver grain clusters on top of cells in F is broader than the size of a cell as defined by the color precipitate in G.
In addition to consistently superb image quality, the data produced by this process will lend well to automation and quantitative interpretation.
Figure 2. Automated image analysis and quantitation of gene expression
Top panel: Left: sample expression pattern of Ly-6/neurotoxin homolog (Lynx1) in a sagittal section of adult mouse brain. Right: automated intensity analysis of the expression pattern, cells expressing high levels of Lynx1 are shown in red, medium expressing cells in blue and low expressing cells in yellow. Lower panel: Quantification of differences in Lynx1 expression strength between wildtype (wt) and barrelless (brl) mice in the hippocampus. Left: hippocampus CA2 region marked in yellow, CA3 region in purple. Middle: close up view of the Lynx1 expression pattern in the hippocampus in wt and brl mice and below the respective intensity analysis. Right: plot of percentage of cells expressing Lynx1 at high (red), medium (blue) and low levels (yellow). The automated intensity analysis was performed over three different sagittal sections from wt and brl mouse brains. The data clearly indicate a decrease of expression strength in the hippocampus of brl mice compared to wt.
Non-radioactive ISH produces a blue-colored precipitate ( Fig. 1D, E and G) that is digitally imaged in a bright-field microscope at magnification appropriate for detecting single cells. At a magnification necessary for a CCD camera to acquire data at single cell resolution, the diameter of the object field of a compound microscope is but a fraction of the dimension of e.g. a mouse brain or a mid gestation mouse embryo section. This necessitates that, with the help of a motorized stage, multiple images are collected from the same section (gray boxes in Fig. 1D). Individual images are stitched together to produce a mosaic representing the entire section.
To distribute data to member laboratories of the BCM-IDDRC, the core unit will use the efficient method of a database which was recently established. Currently this database allows searches for gene names and gene sequence. In the near future, searches with anatomical terms will also be possible. Initially, the expression data are deposited on a non-public sector of the database (accessible by a password only) but upon publication of the study for which expression data were generated, all data will become accessible to the public. This is important because papers often show merely a small fraction of the data produced. In other words, the core will analyze serial sections of brain tissue, the most relevant of which will be present in the resulting publication(s). The rest of the image data, which normally would not be accessible to the scientific community, can then be viewed on www.genepaint.org.