Stephen Lockett, Anton Rutten,[1] Damir Sudar, Ramin Khavari,[2] Daniel Pinkel, Joe Gray
Resource for Molecular Cytogenetics, MS 74-157, Lawrence Berkeley National Laboratory, Berkeley, CA 94720.
Fluorescence in situ hybridization (FISH) is a technique for labeling specific nucleic acid sequences inside cells and has wide spread use for analyzing the genetics of tumors. When FISH is applied to solid tumors, it is important to preserve the cellular organization of the tissue so that genetic abnormalities in one cell can be compared to neighboring cells. The standard approach for undertaking such studies is to prepare thin (~4 mm) tissue sections, perform FISH and observe the labeled sequences with an epi-fluorescence microscope. However, this approach has the drawback that nearly all cells and nuclei have been truncated by the sectioning process and so it is impossible to accurately determine the number of copies of the sequences in each cell. The aim of this project is to develop FISH protocols, 3D (confocal) microscopy and 3D image analysis for studying 20 mm thick sections where most of the cells are intact.
At this meeting last year we presented interactive algorithms for rapid, visual-based enumeration of punctate FISH signals in each intact cell (Lockett et al). Since then, we have tested these algorithms, begun automation of the enumeration procedures and developed a technique for correcting images for autofluorescence (Szollosi et al).
The interactive algorithms were tested using control specimens of normal skin where two FISH signals per intact nucleus were expected. We used FISH probes to specific chromosome centromeres and to the 20ql3 region (a region commonly amplified in breast cancer) and detected over 80% of the targeted sequences in the intact nuclei.
We began automation of the enumeration procedure by developing an algorithm for detecting fluorescence stained nuclei from 3D images. The algorithm first calculated threshold intensities that were used for dividing the images into bright regions representing nuclei and the dark background. Next, the size and shape of each nuclear region was measured. Regions that were too large or irregular in shape to represent individual nuclei were split into smaller regions using erosion operations. The algorithm detected 75% of nuclei in thick, normal skin of which over 99% were deemed correctly segmented based on visual comparison with the acquired images.
Autofluorescence severely limits sensitivity when detecting FISH signals in tissue sections. An algorithm was written which calculated the autofluorescence component of images of FISH signals using images of only the autofluorescence in the same microscopic scene. After subtracting the autofluorescence component from the FISH images, signals barely detectable before correction were clear visible after correction. Use of the algorithm proved essential for the detection of locus specific probes (erbB2 and 20ql3) in tissue sections.
In this project, we developed (and will continue to develop) technology that for the first time can analyze the genetics of small, premalignant lesions, and directly correlate this molecular cytogenetic information with tissue histology. A major application of this technology will be to follow the evolution of genetic instability during the progression of breast cancer by enumerating nucleic acid sequences in histologically defined regions of tumors that presumably represent different stages of the disease.
This work was funded by the US DOE contract DEAC0376SF00098 and a grant from the Whitaker Foundation.
[1] Collaborating author, Delft University of Technology, The Netherlands.
[2] Collaborating author, The University of California, San Francisco.
References
Lockett, S.J., Thompson, C., Sudar, D.. Mullikin, J.. Hyun, B., Kharvari, R., Pinkel, D. and Gray, J. (1995)
Interactive Algorithms for Rapid Chromosome Copy Number Enumeration of Individual Whole Cell Nuclei Inside Intact Tissue Specimens. Proc. SPIE.
Szollosi, J., Lockett. S.J., Balazs, M. and Waldman, F.M. 1995. Autofluorescence Correction for Fluorescence in Situ Hybridization. Cytometry. Accepted