Looping versus linking: toward a model for long-distance gene activation

Locus control regions (LCRs) are defined by their ability, in transgenic assays, to direct high-level, tissue-specific expression of linked genes at all sites of integration examined and at moderately constant levels per gene copy. The most extensively examined LCR is that associated with the b-globin locus in mammals, where the b-globin genes reside in a linear array and are usually arranged in order of their developmental expression (Fig. 1). The b-globin LCR consists of several DNase I hypersensitive sites (HSs) spread over a region of 20–30 kb; the DNA sequence associated with each of the HSs contains numerous binding sites for erythroid-specific and ubiquitous transcription factors (for review, see Grosveld et al. 1993; Orkin 1995; Martin et al. 1996; Hardison et al. 1997). Expression of stably integrated b-globin transgenes in the absence of the LCR occurs only at low levels and varies depending upon the site of integration, and so is said to be subject to position effects. The high level of expression driven by the LCR at ectopic sites appears to be the product of at least two separable activities, namely the establishment of an ‘open’ chromatin domain and direct gene activation. Gene activation is also a property of enhancers, which are defined by their ability to direct high-level expression of linked genes in transient transfection assays. Although enhancers typically also improve the expression of transgenes integrated into the genome, such expression varies among integration sites due to negative position effects. Thus, unlike LCRs, enhancers are only capable of gene activation at a subset of genomic loci. Presumably, LCRs subsume the function of enhancers along with a more dominant chromatin ‘opening’ activity that can override negative effects from neighboring regions. In the case of the b-globin LCR, 58 HS2 is known to act as an enhancer in transient transfection assays, but elements in addition to HS2 are required for LCR activity in single-copy transgenes (Ellis et al. 1993, 1996). Evidence from one study indicates that an otherwise intact human b-globin locus translocated near centromeric heterochromatin may be subject to a position effect (Rees et al. 1994). Thus, the distinction between enhancers and LCRs may only be one of degree, but in most transgenic assays is still measurable. Given the apparent importance of LCRs in the regulation of many tissue-specific genes, the mechanism by which these elements function has been the subject of a great deal of debate. The currently predominant model for LCR function involves the establishment of an open chromatin domain encompassing the regulated genes, with subsequent gene activation accomplished by direct interactions between elements of the LCR and gene promoters by DNA looping. In this review, we will summarize models for the establishment of active chromatin domains by LCRs and describe in detail the evidence for the looping model of long-distance gene activation. Recent data, especially relating to the function of boundary elements and evidence for facilitators of enhancer and LCR function, indicate that simple LCR–promoter interactions are unlikely. Based on these data, we propose an alternative to the looping model, which we term the linking model.