The intriguing observations by Spellman and Rubin pose a number of challenges about how chromosomal domains are created and maintained, why the genome contains such large clusters of similarly regulated genes, and the nature of transcriptional control. "It raises a lot of questions," says microarray aficionado Brian Oliver (NIH, Bethesda), referring to the Spellman and Rubin paper as a "call to exploration" and predicting a flood of papers exploring these domains. "Is control at the level of individual genes or whole domains?" asks Versteeg. "That's the most important question, but it's too early to say and it might take a long time to answer."
Mapping the transcriptome back onto the genome may help to link what is known about the fine-detail and large-scale regulation of transcription. In the good old days (before genome sequences and chips) the detection of quantitative changes in the expression of an individual gene (usually by northern blot analysis) was followed by a systematic and laborious characterization of its promoter and nearby enhancer sequences that act as a switch to determine whether a gene is on or off. This led to exquisite models of transcriptional regulation controlled by a precise network of sequence motifs and cis-regulatory modules.
More recently, enhancers have been found capable of regulating genes from quite substantial distances. Additional complexity has been revealed by studies of chromatin structure: different conformations of chromatin can regulate transcription and the accessibility to transcription factors by creating physical domains that are effectively 'open' or 'closed' for protein-DNA interactions.
Spellman and Rubin found that there is often a predominant gene within each chromosomal territory that is most strongly expressed or repressed and they suggest that the behavior of neighboring genes might reflect a general 'sloppiness' in transcriptional control. "We don't have a mechanism," admits Spellman, "but I think the most likely explanation is regulation at the level of chromatin." Open chromatin conformations may be created to drive the expression of a certain key gene in the domain with the rest of the nearby genes "in effect being carried along for a ride" [1
]. As long as their expression is not harmful to the cell, the changes in transcription of most genes may not be too important. "The regulation of transcription may be precise when it is needed and sloppy when it is not important," write Spellman and Rubin [1
]. Versteeg strongly rejects the notion of sloppiness in gene control, however, citing the catastrophic consequences of trisomy in humans.
Biologists will be keen to understand how the territories are established. "My gut feeling is that it's driven by boundary elements" says Spellman. Church agrees that defining the nature of the domain boundaries is an important challenge. "If we have enough examples it might be possible to search using motif alignment tools," he says, but he predicts that this will be harder than it was for promoter motifs.
It might be some time before the mechanisms involved and the biological consequences are clearly understood. "It's possible that the expression domains are regulated by the three-dimensional structure of the nucleus and the 'nuclear address' of specific chromosome regions," speculates Oliver. Versteeg proposes that "genes that are highly expressed might be clustered together to facilitate post-transcriptional functions such as splicing and RNA processing," citing the existence of nuclear speckles - sites of splicing and RNA metabolism. Experiments with directed transgene insertions may help to address some of these issues. Comparison with similar studies in other organisms, and correlations with regions of conserved synteny within the genome, are likely to provide insights. And evolution may give us hints about what's going on and about biological relevance.
The paper from Spellman and Rubin [1
] represents a delicious taste of what's to come in the post-genomic era, as extensive genome and transcriptome datasets become available. One result of the work is that microarray analysts might henceforth choose to map their gene-expression profiles to the relevant genomic location, before they construct elaborate theories about specific transcriptional programs on the basis of which genes are turned on and off. We clearly have a lot to learn about chromosomal territories and boundaries within the fly genome and perhaps in the genomes of the worm, the weed, mouse and man.