In the modification of cDNA-RDA, schematised in Fig. 1B, the classical protocol is followed to generate tester and driver representations from normal and mutant brain poly(A)⁺-RNA. Following DpnII cleavage and prior to subtractive hybridization, the 5' protruding ends of the driver are filled-in by normal dNTPs using Klenow. After hybridization, the recessed ends of duplexes, containing tester fragments, are protected from degradation by Exo III by filling with α-thio-dNTPs, and Exo III is used to digest the heteroduplexes and driver homoduplexes that are unprotected by α-thio-dNTPs 4. Single stranded fragments and single stranded extensions from the ends of duplexes are degraded by MBN. Next, the mixture is subjected to the 28 cycles of amplification in the same PCR conditions as in the classical RDA protocol, to generate a difference product. In our experience, this simple modification, a variant of which was proposed before 5, is able to generate a difference product that appears as discrete bands on agarose gel after the first round of hybridization (Fig. 2).

We were interested in applying RDA to examine how the extracellular protein Reelin influences the expression of specific genes during brain development. For the subtractive hybridization, a representation from normal newborn mice brains was used as the driver. The tester derived from scrambler (mDab1-deficient) brains at the same developmental stage. Whereas the representations showed a typical smear in gel electrophoresis (Fig. 2A), the product of the first round of the modified RDA protocol had reduced complexity, with only a few prominent bands (Fig. 2B). In contrast, in a parallel subtraction carried out with the same tester and driver DNA preparations by following the classical protocol of Hubank and Schatz, a difference product showing a few discrete bands was obtained only after the second round of enrichment (DP-2), as shown in Fig. 2D. The whole products DP-1 (Fig. 2C) and DP-2 (Fig. 2D) were cloned into pCR2.1 (Invitrogen). Six and 8 different cDNA fragments were obtained respectively, from DP-1 and DP-2. None of the 8 DP-2 fragments was differentially expressed. Among the six DP-1 fragments, two were from known mouse genes that code for proteins involved in general cellular functions (alpha-tubulin (Tuba1)) and DNA-metabolism (M-phase phospho-protein); one was highly similar to a novel human gene and three had partial homology to mouse EST. Among the novel fragments cloned, a partial 530 bp long cDNA could be extended to 1146 bp by using alignment with genomic and EST databases, and verification by RT-PCR. Northern blot analysis revealed that this gene is expressed in the brain as an approximately 6.5 kb mRNA (Fig. 3). Duplicate quantitative Northern blots analysis of normal (N), reelin-deficient and mDab1-deficient (scrambler) mRNA was carried out by hybridising blots sequentially with the radiolabeled candidate fragment and with a control beta-actin probe. The signals were estimated using a Cyclone-Phosphor Imager apparatus and the OptiQuant program. As shown in Fig. 4, the expression level of this new gene is comparable in normal and reelin-deficient brains but is reduced by about 50% in Dab1deficient brains. This observation of reduced expression in the Dab1-mutant mouse brain suggests that this novel gene is a potential partner in the Reelin-signalling pathway implicated in brain development. Because the database searches revealed that this fragment did not match any known entry, it was provisionally named Nam16. The full-length message and its characterization are currently under investigation.

Normal BALB/c, reelin -/- and Scrambler (Dab1^-/-) mice were bred as described 3. All animal procedures were approved by the institutional Animal Ethics Committee. Brains from four to six newborn (P0) animals were dissected under cold anesthesia and total RNA extracted using an RNeasy RNA kit (Qiagen). Poly(A)⁺-RNA was prepared with PolyATract mRNA Isolation Systems (Promega). cDNA was synthesised by oligo(dT) priming using Superscript II as recommended by the manufacturer (Invitrogen). Double-stranded cDNA was synthesised from 5 micrograms poly(A)⁺-RNA and RDA was performed as described by Hubank and Schatz 1. Restriction reactions were performed with DpnII (New England Biolabs). PCR amplification was carried out using Taq DNA polymerase (Invitrogen). For fill-in reactions, Klenow Fragment (Promega) was added at the proportion of 1 Unit per μg DNA, together with dNTPs (Roche), at a final concentration of 40 μM, and the mixture was incubated for 30 minutes at 37°C for 30 minutes, followed by heating for 10 min at 70°C. Conditions for fill-in with α-thio-dNTP (Promega) were similar. After fill-in, the mixture was submitted to two rounds of extraction with phenol:chlorophorm (1:1) and precipitated with ethanol in the presence of 5 micrograms of glycogen. Digestion was performed for 35 min at 37°C with 300 U of Exonuclease III (Promega) and 10 U of Mung bean nuclease (New England Biolabs) in the presence of 20 mM Tris-acetate, 10 mM Mg⁺⁺-acetate, 50 mM K⁺-acetate, 1 mM DTT, 0,05 mg/ml BSA. To inactivate the enzymes 4 reaction volumes of 50 mM Tris. HCl (pH8,9) were added and the samples were incubated for 10 min at 75°C. This was followed by 28 cycles of PCR following the conditions described in 2. The final difference products were cloned into the pCR2.1 vector (Invitrogen), and sequenced with an ABI prism sequencer using AmpliTAq DNA Polymerase and a Big Dye Terminator Cycle sequencing kit (PE Biosystems, England).

2,5 μg of poly (A) RNA were electrophoresed in an agarose-formaldehyde gel and blotted to Hybond N nylon membranes (Amersham). After UV-crosslinking, membranes were hybridised with [³²P]-dCTP (ICN Biomedicals NV/SA) labelled DNA probes prepared with the RadPrime DNA Labeling System (Invitrogen). After overnight hybridization at 42°C in 50% formamide 4, the membranes were washed twice in 2 × SSC/0,1% SDS for 15 min at room temperature and then once in 1 × SSC/ 0,1% SDS for 15 min at 42°C. mRNA sizes were estimated by comparison with an RNA ladder (Invitrogen). For quantitative analysis, the signals on the membranes were quantified using a Cyclone (Packard) apparatus and the OptiQuant program (Fig. 4).