Eukaryotic chromosomes participate in transcription, replication, meiotic and mitotic condensation, pairing, recombination, and segregation. These pro cesses occur through specific interactions between nuclear proteins and chromosomal DNA sequences. Recognition of specific DNA sequences by proteins requires accessibility. In chromatin, nucleosome-free regions known as nuclease hypersensitive sites are believed to represent the" open windows" that allow enhanced access of crucial resident cis-acting DNA sequences to trans-acting factors (see Refs. 1-7a for earlier reviews). These accessible regions are operationally defined by their pronounced sensitivity to nuclease cleavage or chemical modification, and are typically two orders of magnitude more sensitive than other regions in bulk chromatin. Hypersensi tive sites generally represent a minor (ca. 1%), but highly selective fraction of the genome. These local regions should not be confused with long segments of the chromatin fiber extending over many kilobases (kb), associated with potentially active or actively transcribed genes, that exhibit a heightened general sensitivity to nucleases of perhaps an order of magnitude (8). Nuclease hypersensitive sites were first discovered in studies of SV40 viral chromatin structure in 1978, independently by Varshavsky and coworkers (9) and by Scott & Wigmore (10). Their presence in cellular chromatin was first recognized by Wu & Elgin in 1979 (11). We now know that these sites are fundamental elements in biology as they are ubiquitous among the eukary otes, being found in the cellular chromatin of plants, animals, and fungi as well as within viral or episomal genomes (see Table 2). The method most commonly used to map nuclease hypersensitive sites is DNase I digestion of nuclei followed by indirect end-labeling of the resulting purified double-stranded DNA (14). However, the resolution of this method has been generally overestimated, and these regions are now being elucidated by a wide variety of other enzymatic and chemical probes in conjunction with alternative mapping procedures that yield single-nucleotide resolution (see Table 1). While the sites themselves most often encompass a unit length close to that of the nucleosomal repeat (ca. 200 bp), or several multiples thereof, the fine structure of any given site often exhibits multiple" hot" and" cold" spots that vary depending upon the cleavage reagent used (see Figure 1). Sites are hypersensitive because of the absence of a canonical nucleosome, but contain cold spots (footprints) that reflect the presence of bound trans-acting factors.

Within a given cell type, DNase I-hypersensitive sites fall into two major categories: constitutive and inducible. Constitutive sites are often present in promoter regions of genes" poised" for transcriptional induction; their pres ence precedes transcriptional activation and is independent of gene expres sion. Inducible sites also can appear prior to transcription and often persist