Isolation of rare transcripts by representational difference analysis
Isolation of rare transcripts by representational difference analysis Michael J. O'Neill* and Andrew H. Sinclair
Department of Paediatrics and Centre for Hormone Research, The University of Melbourne, Royal Children's Hospital, Melbourne, Victoria 3052, Australia
Received January 13, 1997;Revised and Accepted May 9, 1997
ABSTRACT
Representational difference analysis (RDA) is a powerful technique for cloning the differences between genomes, and has recently been adapted for cloning differentially expressed genes. RDA, like other PCR-based differential screening methods, is prone to the production of false positives. We have identified a major source of false positives in RDA of cDNA and have introduced improvements which minimise their production. These modifications also significantly increase sensitivity, allowing for the isolation of rare differential transcripts from nanogram amounts of mRNA.
In the absence of informative genetic mutants, efforts in cloning developmentally important genes have often focused on identification and cloning of transcripts based solely on their expression profiles. However, traditional approaches to cloning differentially expressed genes, such as subtractive hybridisation, have required large amounts of RNA making them unsuitable for studies of embryonic development. In recent years several PCR-based differential screening methods (1 -3 ) have circumvented the need for large amounts of starting material. Techniques such as differential display can be performed with sub-microgram amounts of total RNA. Early optimism about the new techniques has been dampened by reports of a high incidence of false positives and poor representation of rare mRNAs (4 ,5 ). Given the apparent widespread use of these methods, surprisingly few genes have been cloned from embryonic material.
In this study we assessed the ability of representational difference analysis (RDA) of cDNA in detecting rare transcripts. RDA was performed as previously described (6 ,7 ) with the following exceptions: (i) 100-150 ng of poly A+ RNA was used in the cDNA synthesis; (ii) digested and excess adaptors were removed by washing the cDNA on Microcon 30 filters (Amicon®); and (iii) 25 [mu]g of driver cDNA was used in the hybridisations while maintaining the previously described driver:tester ratios (a detailed protocol is available from the authors).
In this experiment, four tester samples were produced using 10 [mu]g total RNA from 6.5 day chick embryo brain spiked with varying amounts of a known transcript. A polyadenylated in vitro transcript was generated from a 1.3 kb fragment of the neomycin resistance gene cloned into the pSp64-polyA+ vector (Promega®) (8 ). The in vitro transcript was added to a final concentration of: 0.5% of mRNA in sample 1; 0.05% in sample 2; 0.005% in sample 3; 0.0005% in sample 4, assuming poly A+ RNA to be 1.5% of total RNA. Poly A+ RNA was purified from spiked total RNA using oligo-dT magnetic beads (Dynal®) and cDNA was synthesised from the eluent. The resulting double stranded cDNA was digested with DpnII and ligated to R Bgl adaptors (a 24mer annealed to a 12mer) (6 ). The 1.3 kb neo fragment produces two internal DpnII fragments of 311 and 376 bp and a number of very small fragments (>80 bp). With the addition of RDA adaptors the expected fragments were 359 and 424 bp, respectively. The driver for RDA was produced from unspiked 6.5 day chick embryo brain RNA (from the same source as the testers). The only difference between the four testers and the driver is the presence of the spiked neo transcript in each tester, consequently the two DpnII fragments (359 and 424 bp) from the neo cDNA were the only end products expected after subtraction.
Figure 1 A shows the results after two rounds of subtractive hybridisation and PCR enrichment (difference product 2 or DP2), while Figure 1 B shows the results after three rounds (difference product 3 or DP3). Unexpectedly, a range of bands of varying size was detected. However, the highest band corresponds to the 359 bp band; one of the two expected. Southern analysis of the DP3 gel confirms the identity of the expected neo band (Fig. 1 C). This result shows the exquisite sensitivity of RDA in enriching for cDNA fragments unique to the tester even when those fragments are generated from a transcript present at one copy per cell (sample 4). While RDA is shown to be very sensitive in this experiment, the presence of several unexpected bands in DP2 and DP3 indicates a propensity for RDA producing false positives.
REFERENCES
1 Liang,P. and Pardee,A.B. (1992) Science, 257, 967-971.MEDLINE Abstract
