Scaffold Matrix Attachment Regions SMARs DNA at the Scaffold

Obviously, S/MARs map to non-random locations in the genome. They occur at the flanks of transcribed regions, in 5'-introns and telomeres, and also at gene breakpoint cluster regions (review: Bode et al. 1995). S/MARs are association points for common nuclear structural proteins (review: Bode et al. 2000b and below) and proved to be required for authentic and efficient chromosomal replication and transcription, for recombination and chromosome condensation. These are the levels of their firmly established biological activities:

• Transcriptional level o S/MARs augment gene expression by increasing transcription initiation rates; they are not active in transient expression systems as they require incorporation into authentic (replicated) chromatin structures (review: Bode et al. 1995).

o They provide long-term stability as they contribute to the assembly of the histone acetylation apparatus (review: Bode et al. 2003).

• Transcriptional competence o S/MARs either cooperate with genomic insulators or they function as insulators themselves (Antes et al. 2001; Goetze et al. 2005).

o They enable the topological separation of independently regulated transcription units (Bode et al. 1992; review: Bode et al.1996).

• Origin-of-replication (ORI) support o Eukaryotic ORIs are consistently associated with S/MAR elements (review: Bode et al. 2001) where these provide ARS-like functions not only in yeast (AK and Benham 2005), but also in mammalia (Nehlsen et al. 2006).

• Recombination hotspots o S/MAR-associated DNA structures are involved in the generation of breakpoint-cluster regions (BCRs; review Bode et al. 2000a). They also guide the integration of retroviral genomes (Mielke et al. 1996).

There are more recent indications for an additional role in interphase chromatid cohesion and/or separation (Mesner et al. 2003). S/MARs do not have an obvious consensus sequence. Although prototype elements consist of AT-rich regions several hundred base pairs in length, the overall base composition is definitely not the primary determinant of their activity. Instead, binding and (biological) activity appears to require a pattern of "AT patches" that confer the propensity for local strand unpairing under torsional strain (Bode et al. 2006). Both chemical and enzymatic probes have originally been applied to show that this strand separation potential is utilized in the living cell for anchoring a chromatin domain to the matrix and that DNA accessibility is modulated at times of transcriptional activity (review: Bode 1995). Subsequent bioinformatic approaches support the idea that, by these properties, S/MARs not only topologically separate each domain from its neighbours (Bode et al. 1992), but also provide platforms for the assembly of factors supporting transcriptional events within a given domain (Bode et al. 2003b).

The strand separation potential of a S/MAR is commonly displayed in the form of a stress-induced duplex destabilization (SIDD) profile, which predicts the free energy G(x) needed to effect separation of the base pair at each position x along the DNA sequence, at a certain level of torsional tension (review: Winkelmann et al. 2006). The energy stored in a base-unpaired region (BUR) can serve the formation of nearby cruciforms or slippage structures. These alternate structures, as well as single-stranded bubbles, are recognizable features for DNAses, topoisomerases, poly(ADP-ribosyl) polymerases and related enzymes (see below).

Originally, matrix-attachment elements (MARs) were characterized by their specific (re-)association with the nuclear matrix (i.e. the remnants of a salt-extraction protocol), whereas scaffold-attachment elements (SARs) were mostly characterized by their (re-)association with nuclear scaffolds [i.e. the remnants of a lithium 3,5-diiodosalycilate (LIS, a mild detergent) extraction procedure in the presence of a vast excess of bacterial competitor DNA; Kay and Bode 1995]. The observation that the elements recovered by the reassociation methods are identical or closely related has led to the consensus-term "S/MAR". Moreover, the outcome of the LIS procedure does not depend on the source of the nuclear scaffolds, as there is cross-competition between S/MARs from plants and mammals (Mielke et al. 1990), and it can be simulated by computer-assisted routines.

The binding of various forms of DNA to the nuclear scaffold has been extensively characterized (Kay and Bode 1994). As a whole, the scaffold has a strong tendency to bind single-stranded (ss) as well as supercoiled (sc) DNA. Recognition of scDNA has been ascribed to topoisomerases, since LIS-extracted scaffolds retain a pronounced nicking-closing activity. This activity appears to occur at a distinct subset of sites, as externally added S/MAR sequences and ssDNA do not interfere with the process. In contrast, there is a competition between ssDNA binding and prototype S/MAR binding on some scaffold-associated proteins, but not on others (Kay and Bode 1994; Mielke et al. 1996). In retrospect, competition patterns have proven valuable as they can be applied to reveal specific binding modes. This criterion, for example, has served to identify a novel class of S/MARs in an extended non-coding region where we detected a striking periodicity of narrow SIDD minima, which obey a periodicity of roughly 2,500 bp (Goetze et al. 2003a). A functional comparison revealed that these elements, in contrast to prototype S/MARs, have transcriptional augmentation, but no insulation activity, hinting at the existence of distinguishable classes of S/MARs (Goetze et al. 2005). While the uniform register of these elements might indicate an involvement in upper levels of chromatin organization, it is also possible that these signals serve regulatory functions, for instance in the expression of transcripts of unknown function (TUFs) that are present across large sections of the human genome.

S/MARs have been classified as either being constitutive (demarcating permanent domain boundaries in all cell types) or facultative (cell type- and activity-related) depending on their dynamic properties (details in Fig. 1 and Sect. 3.2). In the first case, the elements are marked by a constitutive DNAse I hypersensitive site in all tissues (Bode et al. 1995), which typically coincides with the preferred cleavage sites for endogenous topoisomerase II (topo II) in living cells (Iarovaia et al.1995). In the second case, hypersensitivity is correlated with either the potentiated state or active transcription (Heng et al. 2004). S/MARs partition the genome into 50-200-kb regions demarcating chromosomal (sub-)domains and/or replicons, and the number of elements that attach the ends of a (sub-)loop to the scaffold approximates 64,000 (see Fig. 1). There are likely an additional 10,000 S/MARs supporting replication foci. Apart from the technique used to derive such a conclusion, the time of observation and the cell type will also determine what subset of the total number of the estimated 74,000 S/MARs is detected (Linnemann et al. 2007). In 2006 still only a minor fraction of S/MARs (i.e. 559 for all eukaryotes) had met the standard criteria for an annotation in the S/MARt database (Liebich et al. 2000; http://sS/MARtdb.bioinf. med.uni-goettingen.de/).

Figure 1 summarizes the criteria by which the nuclear matrix/scaffold can be functionally integrated into the CT/ICD architecture.

0 0

Post a comment