199 Nuclear compartments and gene regulation Moira Cockell and Susan M Gasser* Improvements in fluorescence microscopy have allowed us to explore the three-dimensional organization of the nucleus in ways that were impossible ten years ago, revealing subdomains or compartments within the nucleus defined by their enrichments of subsets of factors. Correlations have been drawn between the silencing of a gene and its proximity to a heterochromatic compartment or to the nuclear periphery. The application of genetics and high-resolution microscopy helps examine the creation, maintenance and impact of these compartments on gene expression. Addresses Swiss Institute for Experimental Cancer Research, Ch. des Boveresses 155 CH-1066 Epalinges s/Lausanne, Switzerland *e-mail: sgasser@eliot.unil.ch Current Opinion in Genetics & Development 1999, 9:199-205 http://biomednet.com/elecref/0959437X00900199 0 Elsevier Science Ltd ISSN 0959-437X Abbreviations FISH fluorescence in situ hybridization PC-G Polycomb-group PRE polycomb response element RapI P repressor activator protein 1 SAR scaffold attachment region Sir silent information regulator TPE telomeric position effect trx-G trithorax-group Introduction Given the highly condensed state of the interphase genome, most scientists agree that gcnc regulation cannot be fully explained by linear, two-dimensional models in which sequence-specific factors bind promoter elements to either induce or block transcription. Two common examples of long-range chromatin-based control include the develop- mentally regulated ‘freezing’ of genes into heritable states of either active or inactive chromatin by the Polycomb and trithorax group proteins, and the propagation of centromer- ic heterochromatin into adjacent euchromatic sequences, which results in position effect variegation. Once estab- lished, these repressed chromatin states are heritable and supercedc the presence or absence of sequence-specific transcription factors (reviewed in [l]). Although a number of studies suggest that repressed domains have unique posi- tions within the nucleus, either with respect to the nuclear envelope or to other silent sequences [2,3], the spatial posi- tioning of genes may merely bc the consequence of regulatory events that lead to the repressed state. In short, it remains to be proven that the subnuclear localization of a gene contributes actively to the establishment and/or mitot- ic inheritance of a particular chromatin state. Testing this is complicated by the paucity of our knowledge about how the spatial arrangement of chromosomes within the nucleus is established and maintained. Indeed, apart from nuclear lamins and pore complexes, it is difficult to name a hnajZe ‘structural’ protein of the nucleus, as opposed to well characterized ‘structural’ elements of chromatin (e.g. histones and histone-binding complexes). Given the dynamic nature of chromosomes through the cell cycle [,.,.S*,6*‘] and the Brownian diffusion of specific sequences observed in inter- phase [7’], it is unlikely that either active or inactive sequences have absolutely fixed nuclear positions. Rather, the clustering of similar chromdtin states (active/inactive) is likely to provide a nuclear context, beyond that provided by flanking sequences, that influences gene expression. Several recent studies addressing these questions are reviewed below. Silencing at the yeast nuclear periphery IJnlike higher eukaryotic genomes, budding yeast chromo- somes have little repetitive noncoding DNA and very few genes that are constitutively silent. Nonetheless, yeast telomeres and HM mating type loci provide useful models for chromatin-based repression: Pol 11 genes inserted at either of these sites become repressed through the binding of three silent information regulators (Sir proteins), which form a complex that appears to propagate along nucleo- somes by binding to underacetylated amino-tcrmini of histoncs H3 and H4 (reviewed in [8’,9’]). Double in sitzl and immunofluorescence detection has shown convincing- ly that sites of Sir-mediated repression-that is, telomercs and HA4 loci-are found clustered in a zone near the yeast nuclear envelope, at least in interphase nuclei (see Figure la). ‘I-he striking concentration of Sir proteins in these foci is dependent on both the integrity of the silenc- ing machinery [l&13] and the clustering of telomeric sequences. Indeed, most mutations that derepress telom- eric silencing and delocalize Sir proteins do not disrupt the clustering of telomcric DNA itself [ll]. Yeast telomeric foci thus appear to form a compartment near the nuclear periphery, in which Sir proteins are maintained at a high local concentration. Do such compartments have functional consequences for gene expression? If these foci were to actively favor repression, one might predict the exis- tence of mutants that delocalize telomeres from perinuclear foci and derepress telomeric silencing. This was indeed found to be the case for mutations in the yeast Ku complex - a heterodimer that binds the ends of yeast chro- mosomes and helps maintain the length of the telomeric TG repeats ([14-181, reviewed in [19’]). Deletion of tither HDFI or HDFZ, the genes encoding yKu subunits, disrupts both the perinuclear positioning of yeast telomeres and telomcrc proximal silencing (also called telomeric position effect [‘I’PE] [14,16]). Conversely, the repression of internal silencer-flanked reporters is enhanced, consistent with the observation that Sirs are fully functional, but no longer sequestcrcd at telomeres ([16]; L Maillet, E Gilson, person-