Normalization of cDNA Sequence Representation by Molecular Selection

Consider the situation in which one would like to clone sequences that are differentially expressed between two developmental stages. An experiment could be set up in which cDNA from one developmental stage is taken as reference and cDNA from a later developmental stage is taken as target. The objective of the experiment is to isolate sequences that switch on or are upregulated during the developmental process. If the experiment is carried out and the differentially expressed sequences isolated by subtractive cloning are identified by sequencing, it will generally turn out that a high proportion of the sequences are trivially interesting, highly expressed sequences, such as globin, actin, or tubulin. This can be a frustrating experience and is not an artifact of subtractive cloning since it can be shown that the isolated sequences are effectively present at higher levels in the target mRNA. Rather, highly expressed sequences are preferentially cloned because the frequency with which a clone is retrieved by subtractive cloning is correlated to the fractional representation of the clone in the initial mRNA population. Consequently, if nothing is done to compensate for the heterogeneity in sequence representation before subtractive cloning, highly expressed sequences will represent an unacceptably high proportion of clones in the subtraction product. The objective of sequence normalization is to circumvent this unfavorable situation by reducing the differences in sequence representation between cDNA species and thereby provide an opportunity to investigate both high and low expression sequences. Normalization should be envisaged in all situations where the high fractional representation of certain sequences will impede the access or exploitation of sequences expressed at lower levels (differential screening, catalog libraries, total probes, and so forth).