A lack of border guards?

Comment on Bender et al, page 1395

In this issue of Blood, Bender and colleagues challenge the concept that the genome is organized into independent regulatory domains.

What protects genes or gene clusters from the effects of their neighbors' activation or repression? A popular view is that genes and their regulatory elements (promoters, enhancers, silencers, and locus control regions [LCRs]) are organized into “domains” flanked by insulator elements that anchor the domain into an isolated loop by attaching it to the nuclear scaffold.^1,2  Direct evidence for insulator or boundary elements is largely limited to Drosophila and yeast, but there is also considerable circumstantial evidence supporting the idea that similar elements exist around the globin gene clusters of higher organisms.¹  Bender and colleagues have addressed this proposal directly in the mouse β-globin cluster, and the concept has been found wanting.

Sequences thought to be boundary elements have been found flanking the β-globin locus in a wide range of vertebrates. Putative boundary elements flanking the chicken locus colocalize with clear changes in chromatin structure from repressive heterochromatin outside the domain to open, active chromatin inside. When studied in in vitro assays, these elements have been shown to act as insulators, blocking enhancer-promoter interactions and/or preventing encroachment of neighboring heterochromatin.¹  They have been sufficiently effective in these assays that researchers now frequently include such elements to protect sequences in gene therapy constructs from chromosomal “position effects.”

Despite these observations, the role of such sequences within the globin loci has not been tested. In particular, the lack of appropriate technology in chickens has prevented the removal of these elements from the endogenous chicken β-globin locus. However, Bender et al have now knocked out the proposed insulator elements (HS-62.5 and 3′HS1) from the mouse β-globin cluster and, contrary to expectation, have found no effect on β-globin gene expression of either element singly or combined. This was the case in adults and embryos and after the induction of erythroid stress.

So if the flanking sequences (HSS-62.5 and 3′HS1) are not the boundary elements for the mouse β-globin cluster or if these elements are redundant, should we be searching for other candidates? Alternatively, might it be that the general nature of the domain concept, as currently proposed, is flawed? Perhaps specific elements flanking the cluster are not needed to establish or maintain its chromatin structure. If so, does this have implications for our understanding of globin gene expression? These results certainly do not require us to throw out the idea that interaction of the LCR with the gene promoters (by looping, tracking, or linking models) is critical, but they do suggest that the significance of additional interactions with the putative boundary elements, demonstrated by chromosome conformation capture (3C) technology,³  require further evaluation and reinterpretation. Indeed, finer analysis of the mice generated by Bender et al, such as globin gene switching and the effects of loss of these elements on 3C analysis, may yet improve our understanding of the regulation of expression.

The general concept that the genome is organized into regulatory domains flanked by insulators has been given particular emphasis by work on the chicken β-globin cluster in which the role of boundary elements at the endogenous locus remains to be tested. Even if they play a critical role at the chicken locus, birds and mammals have been evolving separately for approximately 270 million years. Would it be surprising if, over this long evolutionary period, the regulation of the β-globin clusters differed between chickens and mice? ▪