Article
Bacteriophage Phi29-based whole genome amplification and high-resolution gene-dosage profiling from few hundred tumor cells
Search Medline for
Authors
Published: | March 20, 2006 |
---|
Outline
Text
Introduction: Sufficient quantities of genomic DNA can be a bottleneck in genome-wide genetic analysis of clinical glioma samples. High-throughput genomic profiling by microarray-based comparative genomic hybridization (array-CGH) requires microgram quantities of genomic DNA. A particular challenge for the translation of array-CGH methodology to clinical application is to link it to a robust, up-front technology that allows consistent and unbiased amplification of restricted sources of DNA. DNA polymerase Phi29 can be utilized for the random-primed amplification of whole genomes, although the amplification may introduce bias in gene dosage. The nature and mechanism for this misrepresentation has remained uncertain. We have examined in detail the suitability of Phi29 for the whole genome amplification of archival fresh-frozen and formalin-fixed/paraffin-embedded clinical glioma samples, and its utility for subsequent global gene dosage assessments.
Methods: The utilization of 43,000-element, cDNA microarray-based array-CGH technology provided a high-resolution means to measure genome-wide representational bias and to map gene dosage alterations in the amplified DNA on a gene-by-gene basis.
Results: Phi29 amplified DNA with similar efficiency in DNA from matched pairs of fresh-frozen and formalin-fixed/paraffin-embedded glioma. We show that the distortion in gene dosage representation in the amplified DNA is non-random and reproducibly involves distinct genomic loci. Regional amplification efficiency is significantly linked to regional GC content of the template genome. The biased gene representation in amplified glioma DNA can be effectively normalized by using amplified reference DNA.
Conclusion: We provide a conclusive mechanistic explanation for the varying gene dosage representation in Phi29-amplified DNA. We describe a reliable method to normalize biased gene representations to generate highly precise and comprehensive genomic profiles in both archival fresh-frozen and formalin-fixed/paraffin-embedded tumor tissue from a few hundred cells. This amplification method should therefore lend itself to high-throughput genetic analyses of limited sources of glioma, such as stereotactic biopsies, laser-microdissected tissue, and small, paraffin-embedded specimens.