SUMMARY Recent studies recognize a vast diversity of non-coding RNAs with largely unknown functions, but few have examined interspersed repeat sequences, which constitute almost half our genome. RNA hybridization in situ using CoT-1 (highly repeated) DNA probes detects surprisingly abundant euchromatin-associated RNA comprised predominantly of repeat sequences (“CoT-1 RNA”), including LINE-1. CoT-1-hybridizing RNA strictly localizes to the interphase chromosome territory in cis, and remains stably associated with the chromosome territory following prolonged transcriptional inhibition. The CoT-1 RNA territory resists mechanical disruption and fractionates with the non-chromatin scaffold, but can be experimentally released. Loss of repeat-rich, stable nuclear RNAs from euchromatin corresponds to aberrant chromatin distribution and condensation. CoT-1 RNA has several properties similar to XIST chromosomal RNA, but is excluded from chromatin condensed by XIST. These findings impact two “black boxes” of genome science: the poorly understood diversity of non-coding RNA and the unexplained abundance of repetitive elements.

We identified four proteins in nuclear extracts from HeLa cells which specifically bind to a scaffold attachment region (SAR) element from the human genome. Of these four proteins, SAF‐A (scaffold attachment factor A), shows the highest affinity for several homologous and heterologous SAR elements from vertebrate cells. SAF‐A is an abundant nuclear protein and a constituent of the nuclear matrix and scaffold. The homogeneously purified protein is a novel double stranded DNA binding protein with an apparent molecular weight of 120 kDa. SAF‐A binds at multiple sites to the human SAR element; competition studies with synthetic polynucleotides indicate that these sites most probably reside in the multitude of A/T‐stretches which are distributed throughout this element. In addition we show by electron microscopy that the protein forms large aggregates and mediates the formation of looped DNA structures.

Interphase chromatin is arranged into topologically separated domains comprising gene expression and replication units through genomic sequence elements, so-called MAR or SAR regions (for matrix- or scaffold-associating regions). S/MAR regions are located near the boundaries of actively transcribed genes and were shown to influence their activity. We show that scaffold attachment factor B (SAF-B), which specifically binds to S/MAR regions, interacts with RNA polymerase II (RNA pol II) and a subset of serine-/arginine-rich RNA processing factors (SR proteins). SAF-B localized to the nucleus in a speckled pattern that coincided with the distribution of the SR protein SC35. Furthermore, we show that overexpressed SAF-B induced an increase of the 10S splice product using an E1A reporter gene and repressed the activity of an S/MAR flanked CAT reporter gene construct in vivo . This indicates an association of SAF-B with SR proteins and components of the transcription machinery. Our results describe the coupling of a chromatin organizing S/MAR element with transcription and pre-mRNA processing components and we propose that SAF-B serves as a molecular base to assemble a 'transcriptosome complex' in the vicinity of actively transcribed genes.

SARs (scaffold attachment regions) are candidate DNA elements for partitioning eukaryotic genomes into independent chromatin loops by attaching DNA to proteins of a nuclear scaffold or matrix. The interaction of SARs with the nuclear scaffold is evolutionarily conserved and appears to be due to specific DNA binding proteins that recognize SARs by a mechanism not yet understood. We describe a novel, evolutionarily conserved protein domain that specifically binds to SARs but is not related to SAR binding motifs of other proteins. This domain was first identified in human scaffold attachment factor A (SAF-A) and was thus designated SAF-Box. The SAF-Box is present in many different proteins ranging from yeast to human in origin and appears to be structurally related to a homeodomain. We show here that SAF-Boxes from four different origins, as well as a synthetic SAF-Box peptide, bind to natural and artificial SARs with high specificity. Specific SAR binding of the novel domain is achieved by an unusual mass binding mode, is sensitive to distamycin but not to chromomycin, and displays a clear preference for long DNA fragments. This is the first characterization of a specific SAR binding domain that is conserved throughout evolution and has DNA binding properties that closely resemble that of the unfractionated nuclear scaffold.In the eukaryotic nucleus, chromosomes occupy individual, nonoverlapping territories and reactions of DNA and RNA metabolism are confined to discrete structures in the nuclear interior (for review, see reference 40). Despite many efforts to elucidate the molecular basis for nuclear architecture, a clear conception of higher-order structures in the nucleus has not yet emerged. One much debated possibility is that structure and function of the nucleus are determined by a proteinaceous, skeletonlike entity called the nuclear scaffold or nuclear matrix and its interaction with architectural DNA

The activation of mammalian origins of replication depends so far on ill understood epigenetic events, such as binding of transcription factors, chromatin structure, and nuclear localization. Understanding these mechanisms is not only a scientific challenge but also represents a prerequisite for the rational design of nonviral episomal vectors for mammalian cells. In this paper, we demonstrate that a tetramer of a 155-bp minimal nuclear scaffold͞matrix attached region DNA module linked to an upstream transcription unit is sufficient for replication and mitotic stability of a mammalian episome in the absence of selection. Fluorescence in situ hybridization analyses, crosslinking with cis-diammineplatinum(II)-dichloride and chromatin immunoprecipitation demonstrate that this vector associates with the nuclear matrix or scaffold in vivo by means of specific interaction of the nuclear scaffold͞matrix attached region with the nuclear matrix protein SAF-A. Results presented in this paper define the minimal requirements of an episomal vector for mammalian cells on the molecular level.DNA replication ͉ mitotic stability ͉ nuclear matrix ͉ SAF-A