RNA fingerprinting by molecular indexing.

Class IIS restriction enzymes, a subgroup of class II, cleave DNA at a precise location outside their recognition sites, and produce overhangs of unknown sequences (1). Molecular indexing is a series of techniques designed to characterize DNA fragments by these unknown sequences (2–4). I applied this principle for description of the total mRNA population using a 3′ end cDNA fragment generated by class IIS restiction enzymes (5). The method is based on the finding that Escherichia coli DNA ligase discriminates three nucleotides adjacent to the joining site. Fragments are discriminated by a library of 64 adaptors for all possible overhangs, and selected fragments are PCR-amplified using an adaptor–primer and an anchored oligo-dT primer. They are separated and displayed by a denaturing polyacrylamide gel electrophoresis. Comparing electropherograms from various sources of RNA, differentially expressed genes can be easily identified. This method has several advantages over display techniques based on arbitrarily primed PCR (6,7). In particular, the method can display most genes with low redundancy. However, amplified fragments correspond to 3′ ends of mRNA, and there is not much chance of them to containing coding regions. Additional experiments are required to obtain encoded protein sequences. Here, I describe an alternative protocol for amplifying fragments from upstream regions. The outline of the technique is as follows. The cDNA was at first digested by a class II restriction enzyme that generates an overhang of a defined sequence, and an adaptor cohesive to the end was ligated to it. The 5′ end of the cohesive end of the adaptor must be phosphorylated so that the adaptor sequence attaches the PCR template afterwards. The cDNA was digested by a class IIS restriction enzyme that produces a four nucleotide 5′ overhang. A total of 64 biotinylated adaptors were prepared for all possible overhangs. Each adaptor had a 5′ four nucleotide overhang, of which the outermost base was a mixture of A, C, G and T, and the three inner bases were one of all possible sequences. The cDNA was ligated to one of the 64 adaptors. Repeating the ligation with all the adaptors, restriction fragments that had ends created by both of the enzymes were divided into 64 subpopulations. After recovery with streptavidin-coated paramagnetic beads, the cDNA was treated with a dilute alkali. PCR amplification proceeded with this cDNA sample, using the adaptor–primers. The products were separated by denaturing polyacrylamide gel electrophoresis and the sizes of fragments were automatically recorded by a 373A sequencer (Perkin-Elmer), then analyzed by the 672 GeneScan software (Perkin-Elmer). Comparing electropherograms with different sources of RNA, specific expressed genes were easily Figure 1. Outline of the technique. This figure shows a situation where EcoRI and FokI were used as restriction enzymes.