Models for prediction and recognition of eukaryotic promoters

The enormous impact of various genome sequencing projects is bringing the importance of computer-assisted nucleotide sequence analysis to the attention of a constantly increasing number of scientists. Gene prediction is undoubtedly leading the list of important tasks in this context. It is also generally accepted that this task consists of the recognition of the exon/intron structure of the coding region as well as prediction of the corresponding promoter. However, there is much less agreement about the exact sequences that should be called a promoter. In general, the promoter is an integral part of the gene and often makes sense only in the context of its own gene, especially if important parts of the regulation are determined outside of the promoter (for example, by an intron enhancer; Stamatoyannopoulos et al. 1997). The function of a promoter is to mediate and control initiation of transcription of that part of a gene that is located immediately downstream of the promoter (38). This can be achieved either in an unregulated permanent manner (constitutive transcription) or in a highly regulated fashion by which transcription is subjected to the control of various extracellular and intracellular signals (regulated transcription). The DNA region required to fulfill this function can be determined by assays for promoter function in a heterologous context. Unfortunately, this simple scheme becomes blurred in the case of highly regulated promoters. Often complex regulation involves many more features than just the promoter; for example, enhancers, locus control regions (LCRs), and/or scaffold/matrix attachment regions (S/MARs, reviewed in Boulikas 1996). If any of these units, which are functionally completely different from promoters, happens to be located adjacent to the promoter, delineation of the promoter becomes difficult. This may be one of the reasons why promoter prediction programs almost exclusively focus on proximal promoter regions or even just on the core promoter. Therefore, I will refer to a promoter mainly as the region that is necessary to achieve transcriptional initiation, although this region may not be sufficient to determine the complete regulation of a gene. What is the basic aim of computer-assisted promoter recognition? The most obvious answer is location of an important part of the regulatory region of a gene. However, promoter prediction can also be very useful in the context of gene prediction. The promoter by definition marks the beginning of the first exon of a gene, which is often difficult to predict, especially if the first exon is not translated or very short (sometimes even more than one promoter/first exon exists). The promoter regions also contain information complementary to the exons and introns because transcriptional regulation—which can play an important part in gene function— cannot be deduced from the predicted amino acid sequence. The promoter, once understood, may even yield first clues towards the function of a completely anonymous protein, for example, if the promoter is known to be tissue or cell specific. Prediction of the functionality of a promoter would also be welcome for gene therapy approaches to improve expression of newly created vector constructs.