Transcriptional Augmentation: Modulation of Gene ...

1 Transcriptional Augmentation: Modulation of Gene Expression by Scaffold/Matrix Attached Regions (S/MAR Elements) Jürgen Bode1, Craig Benham2, Angela Knopp1 and Christian Mielke3 1 GBF, National Center for Biotechnological Research, D-38124 Braunschweig, Mascheroder Weg 1 2Department of Biomathematical Sciences, Box 1023, Mount Sinai School of Medicine, 1 Gustave Levy Place, New York, NY 10029 3University of Aarhus, Dept. of Molec. and Struct. Biology, C.F. Møllers Alle 131, DK-8000 Aarhus C Denmark ABSTRACT: For a long time S/MARs could only be characterized by the assays in vitro that led to

their detection.. Only recently a number of biological activities emerged which are common to most

or all S/MARs that are detected by the classical procedures. This review will focus on the

phenomenon of transcriptional augmentation which is found for genomically anchored or episomal

genes and on a group of partially overlapping activities which are suited to maintain an episomal

status. It is further attempted to correlate properties of the S/MAR-scaffold interaction with

prominent or prototype protein binding partners.

Keywords: base-unpairing region (BUR), chromatin domains, episomal vectors, scaffold/matrix

attached regions, stress-induced duplex destabilization (SIDD), transcriptional augmentation,

unwinding elements (UE).

I Introduction: Biological Activities associated with S/MARs

The proteinaceous intranuclear framework, called either `nuclear matrixA (Berezney and

Coffey, 1974) or `nuclear scaffoldA (Mirkovitch et al., 1984), is thought to mediate the domain

organization of the eukaryotic nucleus. Branched core filaments provide a supporting structure for

the formation of DNA loops and participate in diverse matrix-supported processes such as DNA

replication, -transcription and -recombination, RNA-processing and -transport as well as signal

transduction and apoptotic events (review: Berezney et al., 1995).

The DNA elements which mediate the attachment of chromatin loops, so called

scaffold/matrix attached regions (S/MARs), have attracted considerable interest due a number of

rather distinct structure-function relationships. S/MARs of several kilobasepairs are found at the

borders of chromatin domains, and shorter elements with basically the same physicochemical

properties occur in close association with certain enhancers or in introns. Accordingly, S/MARs are

found either in nontranscribed regions or within transcription units, but rarely if ever in coding

regions.

A wide range of activities has been ascribed to S/MARs, among these an insulator function

whereby two of these elements, bracketing a transcription unit, uncouple the gene from

chromosome position effects (reviewed in Bode et al., 1998) and a function as recombination

hotspots which involves nuclear matrix functions (Strissel et al., 1998). Our review will concentrate

2 on two aspects which have already received wide acceptance: the transcriptional (`augmentingA)

activity of S/MARs and their apparent function(s) in episomes. In addition we will discuss some

characterized protein binding partners and their possible contribution to these effects.

II Characteristics of S/MAR-Scaffold Recognition

S/MARs have been operationally defined according to the protocols that lead to their

detection. There are two basic criteria: first, S/MARs constitute those endogenous DNA fragments

that co-purify with the nuclear matrix (i.e. remain bound to the nuclear matrix after chromatin

proteins and DNA in the chromatin loops have been removed) or second, S/MARs represent those

exogenously added DNA fragments that bind to a purified nuclear scaffold in the presence of

(prokaryotic) competitor DNA. While the first (halo mapping) approach may reveal some features

of the scaffold/matrix interaction in the cell (Bode et al., 1996), the latter (re-association) approach

is well suited as a means by which the physicochemical properties of the interaction can be

quantified. An obvious advantage of the reassociation methodology is the fact that affinity

parameters can be simulated by computation with the final aim to localize and classify S/MARs

directly from sequence information. It is anticipated that these concepts can be successfully

applied to study the domain organization of higher cells in the framework of ongoing eukaryotic

genome projects.

Fig. 1 underlines the complexity of recogniton features for a nuclear substructure, called

`scaffoldA, that is obtained after the extraction of cell nuclei with lithium 3,5-diiodosalicylate (LIS).

The initially supercoiled (`scA) S/MAR vector is firmly bound to the scaffold until persistent

topoisomerase activities lower its superhelical density from σ = -0.055 to - 0.040 (Kay and Bode

1995). Concomitant with topoisomerization a nicking activity in the scaffold generates a relaxed

circle (`nickedA). This circle is partially bound, but only in case a S/MAR sequence is present.

S/MAR recognition by the scaffold becomes optimal only after this nicked circle has further been

processed to a linear fragment (`linA), i.e. under the conditions of the standard reassociation

assay. These observations indicate that an efficient S/MAR interaction requires a considerable

amount of conformational flexibility or exposure of single strands due to superhelical strain.

III Anatomy of a S/MAR: UEs, CUEs and the BUR

Chemical probing studies by single-strand specific enzymes or reagents have shown that

S/MARs can undergo strand separation in vivo (reviewed in Bode et al., 1996) or under the

negative superhelicity of a plasmid in vitro (Bode et al., 1992). Interestingly, these reactivity data

can be matched by computer analysis after applying superhelical tension (Benham et al., 1997 and

Fig. 2). For these analyses the sequences in question are analyzed as part of a plasmid, i.e. under

the superhelical density as it exists for the in vitro experiment (σ = -0.055). Although prokaryotes

do not contain S/MAR-type sequences (which is most evident from the fact that prokaryotic DNA at

any excess does not interfere with a S/MAR-scaffold re-association in vitro) common bacterial

3 plasmids show two well characterized, narrow unwinding elements (UEs) that flank the ampicillin

resistance gene (Benham, 1997). Although these sites are too restricted to mediate scaffold

attachment, they serve as convenient internal standards for the quantification of the related, but

more extensive features in prototype S/MARs. If we clone a 2.2 kb S/MAR sequence, derived from

the human interferon-ß upstream domain border, into our standard pTZ18R vector, it forms an

extended base-unpairing region (BUR). This BUR consists of multiple UEs some of which compete

efficiently with the mentioned internal markers. Among the UEs there may be a prominent one, the

so called `core-unwinding elementA (CUE) and this is usually the nucleation center of unwinding

which is preferentially modified by single-strand specific reagents such as chloroacetaldehyde or

potassium permanganate (see the signal at the CUE-position in Fig. 2). The S/MAR insert, called

`IA in Mielke et al. (1990), and an 800 bp subfragment (`IVA, ibid.) which include the CUE are

prototype elements that have been used by us and others in a variety of biological test systems.

Where applicable, we will refer to these elements below.

During the analysis of multiple S/MARs both from animal and plant sources we have

observed a tight correlation between the scaffold association strength and the following

parameters: i - length of the BUR, ii - number, extension and extent of individual destabilized sites

and iii - the spacing between the UEs (Bode et al., in preparation). Individual destabilized UEs may

correspond to the `AT-patchesA described by Tsutsui (1998 ) or the `90% AT boxA observed by

Michalowski et al. (1999). While many S/MARs are AT rich per se, this is not a stringent criterion

since an even distribution of AT-patches, separated by short GC-rich spacers, is also able to

confer binding affinity.

IV Transcriptional Augmentation

S/MARs have been shown to stimulate transcription initiation rates (Schübeler et al., 1996)

in a wide variety of cell lines and sometimes also in transgenic organisms suggesting a rather

widespread transcriptional facilitator function. This action is particularly evident if an integrated

heterologous reporter gene is flanked by two S/MARs thereby forming a ̀ minidomainA. Increased

transcription is largely independent of the source of the elements, it parallels their binding strength

to a LIS-scaffold (Kay and Bode 1995) and it is also seen for artificial constructs which have been

obtained by oligomerizing a CUE up to the critical minimum extension. These findings lend support

to the idea that S/MAR functions are not strictly dependent on sequence but rather on secondary

structure(s).

A S/MAR-stimulated expression (`augmentationA) is never observed in transient assays but

it requires stable integration into the genome suggesting that S/MARs exert their effect by

modulating chromatin structures (Poljak et al., 1994; reviews Bode et al., 1995, 1996, 1998). Fig. 3

demonstrates these basic facts and also indicates the importance of distance between the

promoter and the S/MAR element. For this particular experiment we have used a retroviral gene

4 transfer system which integrates a single, intact copy (the so-called provirus) into a subclass of

genomic sites. The provirus contains a long transcription unit formed by a bicistronic construct and

allows cloning of an 800 bp minimal S/MAR (element ̀ IVA) at various distances (positions a-c, 3´

and 5´3´) from the promoter which resides in the 5´ long terminal repeat (LTR). The lefthand

diagram demonstrates that all constructs are functional since they are transcribed in the transient

phase of gene expression. Here, transcriptional initiation is maximal for the S/MAR free control

(Sp) but is inhibited (20-60%) in the presence of the S/MAR. Expression is monitored by mRNA

levels as well as by the reporter genes yielding consistent data (except for SEAP expression in

construct Sp-a which is explained by multiple cryptic translation-initiation sites in the upstream

S/MAR; Schübeler et al., 1996).

These relationships are dramatically changed upon integration of the transgene, i.e. its

covalent anchoring to the host cell genome. Now transcriptional initiation rates are up-regulated by

the distal S/MAR (in Sp-c) and for the `minidomainA situation (i.e. the 5´-S/MAR-gene-3´-S/MAR

arrangement seen in Sp-5´3´). This minidomain arises as a consequence of the retroviral

integration mechanism during which the complete sequence information of the 3´LTR is copied

and transferred to the 5´-LTR. It is noted that this `double copy vectorA principle has been

successfully applied for the construction of high and stably expressing therapeutic retroviral

vectors using fragment `IVA (Auten et al., 1999).

Transcriptional augmentation, i.e. the increase of initiation rates in the integrated but not in

the pre-integrative state, is the most stringent criterion to discriminate S/MAR and enhancer effects

since a prototype enhancement would occur in both states. The inefficiency of S/MARs prior to

integration might be a consequence of the fact that transiently transfected genes are transcribed

with ease since they adopt an open if any chromatin structure. While enhancers would add

efficiency by stabilizing the transcription initiation complex, an element overcoming restrictions due

to an ordered chromatin structure appears to be unnecessary. A status with characteristics both of

a transiently expressed template and an integrated copy is the replicating episome. We will devote

a separate chapter to the function of S/MARs in episomes which appear to utilize a highly specific

set of interactions for the support of transcription and for replication functions (chapter IV.B.6).

A detailed inspection of Fig. 3 demonstrates that the augmenting function of a S/MAR

critically depends on its distance from the promoter and the direction of transcription. This

phenomenon is explained best by considering the topological changes of the DNA template

occurring during transcription and the potential role S/MARs may play in relieving superhelical

strain (chapter IVB1).

A Transcriptional competence

The hallmark of complex organisms is differential, tissue-specific gene expression.

Although the states of expression are stable and can be propagated through many cell divisions,

5 they do not generally reflect genetic differences but rather epigenetic mechanisms that rely on

DNA sequence patterns contained in nuclear non-coding DNA. Special sequences with regulatory

potential such as introns, locus-control regions (LCRs), S/MARs and repetitive elements are found

in these regions. Thus, the genome can be considered a mosaic of sequence motifs which

cooperate to determine the state of an organism. Four successive events, potentiation, initiation,

elongation and termination have been defined which represent an ordered process for balanced

transcription (review Krawetz et al., 1999).

Potentiation creates transcriptional competence, for instance by changing the local

chromatin conformation from a compact 30 nm fiber to the open 10 nm fiber. This change is a

prerequisite for gene expression as it permits the action of transcription factors that are specific for

the respective gene domain. There are models which implicate S/MARs in establishing

transcriptional competence and others considering their immediate influence on the transcriptional

level. We will use these principal modes of action to structure this paragraph well knowing that

some phenomena may fall in between these two phases .1 Domain size

While it is generally agreed that the average size of a chromatin domain in a eukaryotic cell

is around 70 kb, the natural distribution of S/MARs reveals sizes ranging between 3 and about 200

kb (Gasser and Laemmli, 1987). Generally the smaller loop sizes are assigned to genes that can

be highly transcribed under certain circumstances and prototype examples for this may be the

histone gene cluster (5 kb) which is regulated in a cell-cycle dependent fashion and the type I

interferon gene cluster (loop sizes 3-14 kb; Strissel et al., 1998) members of which are rapidly

activated following a viral infection. It is proposed that these loci are permanently potentiated as a

possible consequence of the close apposition of S/MARs.

S/MARs repeated over a short distance might sterically interfere with a cooperative 10 to 30

nm fiber transition and thereby counteract inactivation. In accord with such a model an artificial

S/MAR-luciferase-S/MAR minidomain with a 3 kb loop was found to remain active after

transfection for more than 3 month whereas a truncated control (S/MAR-luciferase) construct, for

which the loop size is determined by the genomic site of integration, lost half its expression over a

period of 6 weeks (Bode et al., 1995). In contrast to these small, permanently open domains,

genes that are only expressed in distinct cell types or at certain stages of development are

typically embedded in larger domains which have to acquire transcriptional competence under the

respective circumstances.

2 Displacement of histone H1 by D-proteins: The active Mechanism

Are S/MARs involved in active domain opening processes? The specificity of the S/MAR-

scaffold interaction is mediated by proteins recognizing certain non-B structural features. One of

the more prominent components co-purifying with the scaffold is histone H1 (Tab. 1) which plays a

central role in the compaction of chromatin into a transcriptionally silent fiber. Nucleation of H1

6 assembly has been shown to occur on S/MARs, especially on An tracts, from where this histone

spreads in a cooperative manner to occupy flanking DNA. These A-tracts can be titrated with

distamycin whereby the cooperative interactions are broken, H1 is redistributed and transcription

from the S/MAR templates becomes de-repressed. An extensive search for proteins with a

Distamycin-like specificity led to a `D-proteinA candidate, HMG-I/Y, which is known to bind to A-

tracts longer than four AT base pairs. HMG I/Y binding in turn may be facilitated by HMG-1 which

recognizes deformed or deformable DNA without any defined sequence specificity (Käs et al.,

1993).

Thompson et al. (1994) have generated transgenic mice homozygous for a heat shock gene

which is transcribed already at the two-cell stage. During development the same gene is

expressed in a number of differentiated tissues. Transgenic lines were generated from constructs

flanked by S/MAR element ̀ IA and the respective downstream border from the human interferon ß

domain or from S/MAR-free controls. Prior Before the onset of differentiation S/MARs augmented

transgene expression in a consistent, position-independent manner. In contrast, in the

differentiated tissues these expression characteristics were lost.

These observations would agree with the the above mechanism: studies in developing mice

show that HMG-I/Y proteins are expressed at elevated levels in proliferating, relatively

undifferentiated cells whereas somatic histone H1 is not detected until the four-cell stage. This

higher ratio of HMG-I/Y to H1 in undifferentiated cells might be one parameter supporting the

augmenting function of S/MARs. After differentiation occurred, augmentation could be restored for

ear fibroblasts taken from the adult animal provided that they were kept proliferating prior to the

assay.

3 Regional Demethylation

The regulation of eukaryotic gene expression is a complicated process which involves the

interaction of a large number of transacting factors with specific cis-regulatory elements. DNA

methylation plays a role in this scheme by modulating protein-DNA interactions. Available

evidence indicates that methylation serves to fix a transcriptionally silent state, after the assembly

of repressive nucleoprotein complexes. Since the discovery of a demethylase, demethylation is

seen as an active enzymatic process, controlled by specific cis regulatory elements (Bhattachrya,

1999).

In the immune system DNA methylation plays multiple roles, most prominently in expression

and gene rearrangement, which are both controlled by enhancer-S/MAR combinations. Transgenic

models have shown that the µ- and κ chain S/MARs complement the intronic enhancers in the

demethylation reaction, presumably through some form of synergy. While the µ enhancer alone

can establish local areas of accessible chromatin, the S/MARs extend accessibility to more distal

positions. These properties held true even when the specific interactions between enhancer- and

7 promoter-bound factors were interrupted by linking µ enhancer-S/MAR fragments to sites for

bacteriophage RNA polymerases which were either close to or distal from the enhancer. The

long-range accessibility was shown to correlate with extended demethylation of the gene construct

but not necessarily with its active transcription. (Jenuwein et al., 1997)

Similarly, the intronic enhancer together with its associated 3´S/MAR is necessary for

demethylating the κ locus during B-cell differentiation. Replacing the κ-S/MAR by element ̀ IA or by

a plant S/MAR demonstrated that any S/MAR sequence in that position can take over this role,

while tissue specificity is mediated by NF-κB sequences within the intronic enhancer. While

enhancers and S/MARs appear to operate as inducers of chromatin accessibility and stimulators of

transcription, demethylation seems not to be a secondary result of local promoter activity as it

occurs also in the absence of nearby promoter sequences. These results suggest that the

enhancer-S/MAR complex has two related but basically separate functions: to induce regional

demethylation and to permit RNA synthesis (Kirillov et al., 1996).

Several lines of evidence suggest that DNA demethylation also participates in

recombination which is a prelude to proper expression. It is likely that the same combination of

elements first allows access to the recombination machinery - perhaps concomitant with germline

Jκ transcription. Following rearrangement, the same sequences are used a second time for

demethylation and activating the newly juxtaposed Vκ-promoter.

Is there any evidence that the natural loci behave differently from transgenes? Shulman

and coworkers have described a gene-targeting approach by which the endogenous IgH locus can

be modified in hybridoma cells. For the resulting cell lines expression of the IgH locus was found to

depend strongly on the S/MAR-Eµ-S/MAR segment. Using this system, Wiersma et al. (1999; see

also references therein) described the initially unexpected observation that expression of the

endogenous µ-gene persists at half the normal level if either the enhancer or the S/MARs are

retained. These results suggest that the roles of Eµ and S/MAR in initiating expression might be

different from their role in maintaining expression. While both elements might be required to effect

demethylation during initiation, either one may suffice to protect from a re-methylation event.

Which are the domain-opening events that are triggered by regional demethylation? The

possibility that histone H1 binds preferentially to DNA containing 5-methylcytosine in the

dinucleotide CpG and would cause domain opening after its release by demethylation (IVA2) is

appealing but is without experimental support (Campoy et al., 1995). An alternative model is based

on methylation-specific interactions with methyl CpG binding proteins (MeCPs) that are used to

repress certain genes. Two of these proteins, MeCP2 and MDBP-2 contain putative DNA binding

motifs similar to those found in H1 and/or HMG-I/Y and thereby MeCPs may be specialized

histone H1-like proteins. Interestingly, cloning an `attachment region binding proteinA (ARBP)

revealed that this prototype S/MAR binder is homologous to MeCP2 and thereby seems to be a

8 multifunctional factor with roles in looped domain organization, the structure of pericentromeric

heterochromatin and DNA methylation (Weitzel et al., 1997 and references therein). Occurrence of

a protein that can recognize a single CpG but also associates with a wide range of S/MARs

(especially if these contain a GC-rich core flanked by AT rich sequences) might suggest a

reciprocal binding to various types of DNA. An opening mechanism could therefore involve

capturing of ARBP by adjacent S/MARs due to rearrangements within a previously quiescent

domain. This would expose CpGs to the action of demethylases, which would in turn remove an

epigenetic mark typical for the inactive state.

B Transcriptional Level

1 S/MARs as Topological Sinks

Fig. 2 has demonstrated that S/MARs are composed from multiple destabilized sites. We

will describe below the ways by which a BUR can be recognized by a variety of proteins, among

these components with single-strand recognition capacity. S/MARs unwind as a consequence of

stress which might be imposed by supercoiling or by the association of a protein factor. The

observation that S/MARs can be reacted with single-strand specific agents in the living cell

(review: Bode et al., 1998) suggests that they have acquired status `CA in Fig. 4. This state is

induced by the association of single-strand-binders causing transient superhelical tension (B)

which after a while is resolved by topoisomerases. In state C the S/MAR functions as a repository

of underwound DNA which can be utilized in a dual fashion by the transcriptional machinery. If the

element is localized in the positively supercoiled part of the twin domain and the superhelical strain

is strong enough, part or all interactions with the ssDNA binding protein components are broken,

the superhelicity is immediately relieved and ongoing transcription as well as re-initiation are

facilitated (E). If the S/MARs occur downstream, but not far enough, from the transcription initiation

site, the positive superhelicity is inadequate to break these contacts and transcription is stalled as

for constructs Sp-a and Sp-b in Fig. 3. The same restriction does not appear to exist for a

promoter-proximal S/MAR in the negatively supercoiled part of the twin-domain (cf. construct Sp-

5´3´). Here initiation may generally profit from the underwound structure stored in the S/MAR.

During an induction of the human interferon-ß gene we could observe disappearance of CAA-

reactivity from the S/MAR region and simultaneous gain of reactivity in the transcription unit (Bode

et al., 1995). In line with such a `topological sinkA function of S/MARs their stimulatory effect can

be blocked if they are separated by GC rich fragments from the promoter (Poljak et al., 1994).

2 S/MARs as Repositories for Specific Transcription/Replication Factors

Boulikas (1995) has studied regulatory regions within several gene domains and found that

S/MARs are frequently a mosaic of transcription factor binding sites with a local overrepresentation

of some sites. Especially ATTA and ATTTA motifs are abundant in major classes of origins of

replication (ORIs) and S/MARs and this is typical for elements controlled by homeodomain

9 proteins. In line with the fact that a fraction of homeodomain proteins is part of the nuclear matrix it

has been suggested that this fraction is involved in the switching of genes during development by

changing the attachment points of chromatin loops. A role of the nuclear matrix in gene switching

during development is also supported by changes its the general protein composition (review:

Stein et al., 1999).

It is remarkable that there appears to exist a class of factors which interferes with

productive S/MAR-scaffold associations. This is exemplified by NFµNR, a factor that silences the

immunoglobulin heavy chain (IgH) enhancer in non-B cells (Zong and Scheuermann, 1995) and

SATB1 which is a transcriptional suppressor (Kohwi-Shigematsu et al., 1997) with an apparent

function in T-lymphocytes.

3 SAF-B, a Platform to Establish Transcription and Processing Complexes

Scaffold attachment factors (SAFs) are those nuclear proteins that interact with S/MARs

and can hence be detected by Southwestern blotting methods. A protein, designated SAF-B, is an

abundant component of chromatin, and is expressed in all tissues. SAF-B was found to interact

with the C-terminal domain of RNA polymerase II and with subset of serine/arginine-rich RNA

processing factors (SR-proteins). Since SAF-B overexpression changes alternative splicing

patterns and the activity of S/MAR- flanked reporter genes in vivo it has been proposed that the

factor serves as a platform to assemble a 'transcriptosome complex' in the vicinity of actively

transcribed genes and also participates in signal transduction pathways which influence splice-site

selection (Nayler et al. 1998).

A S/MAR-mediated transcriptosome assembly is an intriguing feature that may explain

certain aspects of the augmentation phenomenon. However, SAF-B itself is not a component of

the nuclear matrix itself and anchoring to this substructure has to occur by other components of

the transcriptosome complex such as RNA Pol II or members of the hnRNP protein family.

Whether this productive association implies major rearrangements of some constitutive S/MAR

matrix contacts is an open question.

5 An Alternative: S/MARs as Targeting Elements in Gene Transfer Experiments

The activity of cis-acting sequences is traditionally investigated by the concept of ̀ reverse

geneticsA. To this end constructs with or without the element in question are transfected in parallel

experiments and the level of transcription is monitored, for S/MAR-experiments necessarily under

stable expression conditions. Is this a reliable procedure for the study of anchoring elements?

S/MARs are not only recombinogenic sequences, which under certain circumstances yield multiple

copy integration for the associated construct (Bode et al., 1996) but they might also direct the

transgene to the nuclear matrix and cause integration in to regions of the genome with preexisting

scaffold-association potential. In this case the outcome of the result would also be determined by

the integration site in addition to the intrinsic differences of the constructs that are compared. A

10 final answer to this question will only come after a careful characterization of many integration sites

for various constructs and by investigating both classes of constructs at predefined genomic

integration sites. Rather than using homologous recombination for the alteration of genes in their

native environments (Wiersma et al., 1999) we are in the process of using site-specific

recombinases (Cre, Flp) for the introduction of transgenes into predetermined sites or for the

stepwise decomposition of complete gene domains at a given genomic location. Initial reports on

this concept have been published (Bode et al., 1996, 1998; Iber et al., 1999).

While these experiments are in progress, we have completed the investigation of S/MARs

as parts of retroviral constructs (Schübeler et al., 1996, 1998 and Fig. 3). Besides integrating a

single intact copy into the genome of the host cell, retroviruses have an intrinsic apparatus by

which sites of an increased transcriptional potential are sensed and utilized (Mielke et al., 1996).

We have therefore proposed retroviral vectors as an efficient means to introduce transgenes at

sites with a transcriptional potential anticipating that the integration machinery will overcome any

targeting potential of a cloned sequence. The available data clearly support the idea that at least

part of the augmenting action of S/MARs is due to a cis-effect.

6 Episomes and S/MARs

Viruses depend on host cell functions for coming alive. It is obvious that cellular functions

that require a structural organization will also be used by the virus for transcription and replication.

Replication of the small DNA tumor virus SV40 is an excellent example, because in addition to

virion proteins, it codes for only a few regulatory proteins, the most important one being the SV40

large tumor (`large TA) antigen. Deppert and Schirmbeck (1995) have summarized the evidence

that all major viral processes during the life cycle from viral DNA replication to virion formation

occur within the structural systems of the nucleus, in particular the chromatin and the nuclear

matrix. Large T antigen itself becomes a member of the nuclear matrix where it binds to the origin

of replication (ORI) and starts the assembly of an initiation complex in concert with cellular factors.

It might also mediate the known matrix association of SV40 minichromosomes which grants their

replication and maintenance as episomes. Interestingly, the SV40 genome contains a S/MAR

which is part of the large T coding region (Pommier et al., 1990). Upon deletion of the S/MAR the

episomal status is no longer maintained leading to the integration of the crippled construct.

We have recently demonstrated that for SV40 derivatives episomal functions can be

restored by replacing the large-T coding region in an SV40 vector by S/MAR element `IA

(Piechacek et al., 1999). This S/MAR mediates the association of the episome with the nuclear

matrix and the metaphase scaffold, helps to utilize the centromer function of host cell

chromosomes as well as replication function of the host. Interestingly, bovine papilloma virus

(BPV) and Epstein Barr virus (EBV) use related strategies, i.e. the E1/E2 or the EBNA-1 proteins,

resp., for chromosome attachment and episome maintenance (review: Calos, 1998).

11

Although the participation of a S/MAR has not been directly demonstrated for BPV, its role

as a maintenance element during the construction of episomal vectors has become evident: the

potential of BPV-derived episomal vectors was increased dramatically when a hybrid plasmid

(BPV-BV1) was constructed which could be shuttled between E. coli and mouse cells. For this

function it had to contain a 69% subfragment of BPV and a minimum of 2.7 kb eukaryotic

`stabilizing sequenceA which had been found, by trial and error, in the large ß-globin intron.

Lateron we have demonstrated that this sequence coincides with a S/MAR (Klehr and Bode, 1988

and references therein).

In Burkitt lymphoma cells, the extrachromosomal EBV virus genome is stably associated

with the nuclear scaffold and this association is effected by a S/MAR which includes the origin of

latent viral replication (oriP). Cellular DNA polymerases replicate the latent genomes during early

S-phase and in synchrony with host cell DNA. Replication and partitioning between the daughter

cells are mediated by interactions of the S/MAR-ORI region with a latent viral protein, EBNA-1, and

the nuclear matrix. Other important signals for EBV replication and latent gene expression are also

found in close proximity to the S/MAR (Jankelevich et al., 1992). Treatment with phorbolester,

butyrate and TGF-ß induces the EBV lytic cycle with concomitant use of the lytic origin of

replication (oriLyt). During this induction, ori P association was shown to decrease while ori Lyt

becomes specifically associated with the nuclear scaffold. These findings suggest that the dynamic

matrix association of the two EBV ORIs might regulate viral gene expression (Mattia et al., 1999).

Besides their established function in replication there is one convincing demonstration that

S/MARs contained in replicating episomes also effect transcriptional augmentation. To define the

elements of the Ig-κ gene involved in deregulation of the c-myc gene after translocation, Hörtnagel

et al. (1995) have assembled different parts of the Ig-κ locus in an EBV-derived episomal vector.

These experiments clearly showed that the S/MAR is required for the maximum c-myc activation

observed in Burkitt lymphoma cells. In order to differentiate between S/MAR and enhancer

functions, both elements were also tested in transient transfection experiments where the

enhancers provided a 30 fold activation while the S/MAR alone reduced transcription to the

background level. This study suggests that episomally replicating constructs allow the study the

role of S/MARs in transcription and should therefore be useful for their detailed analysis. A

complication may arise from the fact that these derivatives of vector pHEBo will, in addition,

depend on the ORI-associated S/MAR.

In summary, for an augmenting action S/MARs will either need the buildup of superhelical

tension which is granted after integration or in the framework of a covalently closed circular

structure and/or they require the ultimate, ordered chromatin structure which can only be

established by replication.

V Ubiquitous Components of the Nuclear Scaffold

12

The backbone of eukaryotic nuclei can be isolated and characterized according to protocols

which have been optimized for the removal of soluble components (Berezney and Coffey, 1974;

Mirkovich et al, 1984; Fey et al., 1986). The resulting skeleton resembles the nucleus in size and

shape and three of its substructures can be discerned by electron microscopy: i - the lamina-pore

complex at the nuclear periphery, ii - an intranuclear network of filaments with associated granular

material and iii - residual nucleolar structures.

As expected for an operationally defined entity, various procedures lead to matrix/scaffold

preparations that differ to some extent in ultrastructure (Belgrader et al., 1991) while the vast

majority of protein constituents appears to be identical (Ivanchenko and Avramova, 1992). Even

more important, the DNA elements (S/MARs) which are characterized by a specific re-association

with these structures in the presence of a vast excess of bacterial competitor DNA are mostly

independent of the origin of the nuclear skeleton.

Our laboratory has adapted the LIS-(lithium 3,5-diiodosalicylate-) based procedure for

nuclei isolated from cultured mammalian cells and plant tissues (Kay and Bode, 1995). The

resulting scaffolds are used for capturing scaffold attached regions under stringent conditions.

Relative binding affinities are derived by an `equal fractions approach@ and referenced to an

internal standard. The scaffolds resisting LIS extraction contain components from the three

mentioned nuclear compartments, i.e. the lamins A-C (lamina), nucleolin (nucleolus) and IF-type

proteins from the fibrogranular internal network. These IF-like proteins may even comprise another

lamin fraction since intranuclear lamins are uncovered by extracting the chromatin (Hozak, 1995).

If overexpressed, another IF-type S/MAR binder, NuMA, fills the nucleus with an ordered lattice

which is stable to detergent extraction suggesting a major structural role for this protein in the

architecture of the interphase nucleus (Gueth-Hallonet et al., 1998).

A scaffold of this composition binds supercoiled plasmids and ssDNA independent of

sequence. In addition, it binds linear dsDNA provided it contains a S/MAR element. Since all

S/MAR assays are performed with linearized fragments, and characterized by competition with

ssDNA, only components mediating this binding mode will be considered in the following. In doing

so, we learned that the two predominant protein constituents of the scaffold, SAF-A and the lamins

(including other IF-type proteins), fully account for the binding properties of a complete scaffold.

Other S/MAR binding constituents like histone H1, nucleolin and possibly actin (Ivanchenko and

Avramova, 1992) may contribute to the binding but they cause no major imprint on the specificity

toward dsDNA. ARBP, another abundant protein, is a special case due to its particular solubility

features. It is part of a nuclear matrix isolated by 2M NaCl but absent from LIS-scaffolds (Ludérus

et al., 1992 and references therein).

A SAF-Proteins: The SAF-Box

13

Mattern et al. (1997 and references therein) have identified the 21 most abundant proteins

that are exclusively present in the internal nuclear network. In line with earlier reports (Nakayasu

and Berezney 1991) many of these belong to the group of heterogenous nuclear

ribonucleoproteins (hnRNPs) supporting a model in which the major protein constituents are

involved in RNA metabolism, -packaging and -transport. A major member of this group, hnRNP-U

has first been characterized as the most prominent attachment factor in LIS-extracted scaffolds

(called SAF-A by Fackelmayer and Richter, 1994 and SP120 by Tsutsui et al., 1993). SAF-A

associates with multiple S/MARs and UV-crosslinking experiments show that this established RNA

binder is likewise associated with DNA in vivo. S/MAR binding is optimal, if AAT patches@ (short

stretches of consecutive A´s and T´s), are distributed according to certain rules (Tsutsui, 1998).

HeLa-cells contain about 2 million SAF-A proteins per nucleus, half of which associate with

the nuclear matrix in a salt-resistant manner. The other half is either bound to hnRNP particles or

resides in a DNAseI extractable fraction. In vitro, SAF-A shows a pronounced propensity to self-

polymerize and this state is required to recognize S/MAR DNA. The primary structure of SAF-A

reflects its dual function as there are two independent nucleic acid binding domains, i - a C-

terminal RNA/ssDNA binding domain (RGG box), and ii - a S/MAR specific 45 amino acids N-

terminal domain, called ASAF-box@, which is split and inactivated during apoptosis. The SAF-box

resembles a homeobox lacking the DNA recognition helix and represents the first characterized

protein domain specifically recognizing S/MARs (Fackelmayer 1999, submitted). It recurs in

several eukaryotic proteins, from yeast to man, among these SAF-B (see IVB3).

For a cloned SAF-box, a characteristic association with S/MARs can only be demonstrated

in pull-down experiments and in case a critical protein density is reached on the surface of

Sepharose-beads. These data indicate a cooperative binding mode which is also typical for

scaffold-S/MAR interactions (Kay and Bode, 1994): each individual domain interacts only weakly

with a DNA element, which may be an UE according to Fig. 2. Only the simultaneous binding of

multiple SAF-boxes will then confer a strong and at the same time specific interaction (Zuckerkandl

and Villet, 1988). This model would explain the well known phenomenon that there are hardly any

naturally occuring S/MARs below a critical length of 300 bp. Another feature typical for S/MAR-

scaffold interactions could also be demonstrated, i.e. DNA binding is sensitive to distamycin (which

binds to the minor groove of oligo(dA) @oligo(dT) tracts), but not to chromomycin (specific for the

minor grove in GC-rich regions). The failure of ssDNA to compete for the interaction of S/MARs

with SAF-boxes shows that many but not all criteria of scaffold-S/MAR interactions can be

explained by SAF-box proteins.

B Lamins and other Intermediate Filament Proteins

Intermediate filament (IF) proteins are built according to a common tripartite structure: a

central α-helical rod which is flanked, on both sides, by nonhelical domains with ssDNA binding

14 potential. IF proteins self-assemble into long polymers by coiled-coil formation between the rods.

The spatial distribution of S/MAR binding centers in the nuclear scaffold has been studied

by Ludérus et al. (1994 and references therein). S/MAR sites appear to be distributed equally over

the peripheral nuclear lamina and the internal fibrogranular network. In case of the lamina, lamins

B (and later A) have been determined as the major binding partners. The specificity of a S/MAR-

lamin interaction could be confirmed with paracrystal-like lamin polymers which were prepared by

dialyzing a lamin solution into a buffer of low ionic strength. Two different types of interaction were

discovered and these appear to be related to different features of S/MARs. One type involves the

minor groove of the DNA double strand and the second single strand regions. Both modes of

association are interdependent as S/MAR binding is almost completely inhibited by the presence

of single-stranded competitors. These characteristics resemble other intermediate filament

proteins which, in addition to their intrinsic ssDNA binding property, gain S/MAR binding potential

in their filamentous state (Traub and Shoeman, 1994). It is hypothesized, that in this state two

ssDNA binding domains act in concert to recognize individual strands in the BURs (Fig. 4).

It has been claimed by the respective authors, that either lamin B or SAF-A mirror the

S/MAR-binding behavior of a complete scaffold. However, while both proteins display a

comparable recognition profile for native and artificial S/MAR elements and respond similarly to

competition by distamycin versus chromomycin, they clearly differ regarding the competition of

S/MARs by ssDNA. The 50% competition limit found for the complete scaffold (Kay and Bode,

1994) may therefore reflect the contribution of these two major contributors.

VI How other S/MAR-Binding Proteins Deal with BUR-Associated Structures

The prediction of S/MARs from primary sequence data has met with unexpected difficulties

since the major scaffold proteins recognize a structural consensus rather than a primary

sequence. This has been exemplified above by the prototype S/MAR binders, SAF-A and lamin

A/B. A prominent S/MAR-specific feature is a regular assembly of UEs (Fig. 2). The resulting base-

unpairing regions have been shown to attract many S/MAR binding proteins which can be isolated

by ̀ BUR-affinity chromatographyA (review: Kohwi-Shigematsu et al., 1997). So far this procedure

led to the recovery of a ̀ special-AT-rich sequence binding proteinA (SATB1), nucleolin, a cancer-

associated protein p114 with relation to PARP and to a complex between PARP and DNA-PK

(Galande and Kohwi-Shigematsu, 1999 and references therein). Interestingly, Ku autoantigen, the

DNA-binding subunit of DNA-PK, is also a prominent ORI-binding factor. Related recognition

properties also assigned to BRIGHT (Herrscher et al., 1995) and mutant forms p53 (p53mut; Will et

al., 1998 and references therein). Among these examples, SATB1, nucleolin, PARP/DNA-PK and

p53mut appear to be prototype binders recognizing BURs according to radically different

mechanisms.

For SATB1, chemical interference assays demonstrated binding along the minor groove of

15 double stranded S/MARs with very little contact to the bases. Association was ascribed to a

S/MAR-domain and a homeodomain which together direct the protein to a specific CUE within the

IgH-associated BUR. SATB1 does not bind to nor is it competed off by ssDNA suggesting that the

CUE is recognized indirectly through an altered sugar phosphate backbone structure. Binding of

SATB1 may therefore prevent unwinding and suppress promoter/enhancer functions (Kohwi-

Shigematsu et al., 1997). As in the other cases, the CUEs consist of a special AT-rich sequence

context in which one strand is well- mixed A's. T's and C's, excluding G's. Within S/MARs these

AATC sequences@ typically occur in clusters and they show a rather close correlation with base

unpairing properties (Krawetz and Bode, unpublished). SATB1 associates with quite a number of

different established S/MARs and its binding has been used as an alternative means for the

detection of S/MAR elements .

Nucleolin binds to RNA and single-stranded DNA. Within S/MARs it recognizes the region

of highest base-unpairing potential. In contrast to the highly selective binding of SATB1 to

double-stranded S/MARs, nucleolin also accepts single strands of S/MAR-DNA and among these

the T-rich strand is preferred. Its affinity for S/MARs is severalfold higher than to RNA and human

telomere DNA which are other established binding partners.

Both PARP and DNA-PK are usually activated by DNA strand breaks, they can be

recovered as a complex and are implicated in DNA-repair, -recombination, -replication, and -

transcription. While it has been described that PARP and the DNA-binding subunit of DNA-PK

associate with free DNA ends, the highly specific association with a BURs in closed circular

supercoiled DNA substrates could recently be demonstrated; this interaction is abolished after and

may therefore be regulated by ADP-ribosylation (Galande and Kohwi-Shigematsu, 1999).

p53mut specifically interacts with an AATATATTT core-unwinding region and it catalyzes

DNA strand separation when this motif is located within a structurally labile sequence environment.

Since there is no recognition of individual single strands, it is concluded that the active process of

strand separation provides a basis for its high-affinity.

HMG proteins are other prominent factors with a binding preference for S/MARs. Proteins

with multiple HMG boxes (prototypes: HMG-1/2) recognize irregular DNA structures in a

sequence-nonspecific manner. A preferred substrate is cruciform DNA which is bound with an

affinity exceeding B-DNA. Cruciforms are formed by intrastrand pairing at inverted repeats which

are rather prominent S/MAR features (review: Bode et al., 1998). Since base unpairing is a

prerequisite for cruciform formation, it directly depends on SIDD properties.

An 11-residue motif named AAT-hook@, first detected in the HMG-I(Y) protein, represents

a DNA-binding unit different from the known α-helix, ß-sheet and Zn++-finger motifs. AT hooks

recognize the narrow minor groove of AT-rich double strands in a way resembling distamycin and

netropsin. High-affinity binding sites contain several A/T-tracts separated by 6-8 base pairs. Once

16 again, motifs of this architecture are common in BURs.

In summary, the majority of proteins associating with S/MARs are attracted by specific

features of base unpairing regions such as particular dsDNA structures [SATB1, HMG I(Y), SAF-

A], the recognition of single strands individually (lamins, nucleolin) or conjointly (p53mut) or

secondary structures depending on prior unpairing (HMG-1/2). Sometimes special structures and

single strands are recognized by different domains on the same protein (SAF-A). Together with

sequence-specific factors these proteins form a regulatory network in which the transcriptional

state of a chromatin domain can be precisely adjusted.

Acknowledgements

We are most grateful to our colleagues Frank Fackelmayer (Konstanz), Steven Krawetz (Detroit)

and Stefan Stamm (Martinsried) for intense discussions on the occasion of the Gene Therapy

Molecular Biology meeting in Crete (August 1998) and for obtaining relevant data prior to

publication. The excellent cooperation with Hans-Joachim Lipps and Armin Baiker (Universität

Witten-Herdecke) on the principles of episomal replication is gratefully acknowledged.

Development of the present concepts would not have been possible without the continuous input

and encouragement over many years by Terumi Kohwi-Shigematsu (Lawrence-Berkeley

Laboratory). Work from the authors labs was supported by an EU grant to JBO (BIO4-CT98-0203),

DFG grants (Bo 419/5; Bo 419/6), Danish Cancer Society Grant 97-100-32 to CMI and a grant

from the NIH to CJB.

REFERENCES Auten J., Agarwal M., Chen JY, Sutton R, Plavec I (1999) and references therein: Effect of scaffold attachment

region on transgene expression in retrovirus vector-transduced primary T cells and macrophages. Hum Gene Ther 10:

1389-1399.

Belgrader P, Siegel A J, Berezney R (1991): A comprehensive study on the isolation and characterization of

the HeLa S3 nuclear matrix. J Cell Sci 98: 281-291.

Benham C, Kohwi-Shigematsu T, Bode J (1997): Stress- induced duplex DNA destabilization in scaffold/matrix

attachment regions. J Mol Biol 274: 181-196.

Berezney R, Mortillaro MJ, Ma H, Wei X. Samarabandu J. (1995) The nuclear matrix: A structural milieu for

genomic function. Int Rev Cytol 162A: 1-65.

Berezney R, Coffey DS (1974): Identification of a nuclear protein matrix. Biochem Biophys Res Comm 60:

1410-1417.

Bhattacharya SK, Ramchandani S, Cervoni N, Szyf M (1999): A mammalian protein with specific demethylase

activity for mCpG DNA. Nature (London) 397: 579-583.

Bode J, KohwiY, Dickinson L, Joh RT, Klehr D, Mielke C, Kohwi-Shigematsu T (1992): Biological significance

of unwinding capability of nuclear matrix- associating DNAs. Science 255: 195-197.

Bode J, Schlake T, Ríos-Ramírez M, Mielke C, Stengert M, Kay V, Klehr-Wirth D (1995):Scaffold/Matrix-

Attached Regions (S/MARs): Structural Properties Creating Transcriptionally Active Loci. Int Rev Cytol 162A: 389-453.

Bode J, Stengert-Iber M, Schlake T, Kay V, Dietz- Pfeilstetter A (1996): Scaffold/matrix-attached regions:

Topological switches with multiple regulatory functions. Crit Rev Eukaryot Gene Expr 6: 115-138.

Bode J, Bartsch J, Boulikas T, Iber M, Mielke C, Schübeler D, Seibler J, Benham C (1998): Transcription-

17 promoting genomic sites in mammalia: their elucidation and architectural principles. Gene Therapy and Mol Biol 1: 551-

580.

Boulikas T (1995): Chromatin domains and prediction of MAR sequences. Int Rev Cytol 162A: 279-388.

Calos MP (1998): Stability without a centromere. Proc Natl Acad Sci USA 95: 4084-4085.

Campoy FJ, Meehan RR, McKay S, Nixon J, Bird A (1995): Binding of histone H1 to DNA is indifferent to

methylation at CpG sequences. J Biol Chem 270: 26473- 26481.

Deppert W, Schirmbeck, R (1995): The nuclear matrix and virus function. Int Rev Cytol 162A: 485-537.

Fackelmayer FO, Richter A (1994): Purification of two isoforms of hnRNP-U and characterization of their

nucleic acid binding activity. Biochemistry 33: 10416-10422.

Fey EG, Krochkalnic G, Penman S (1986): The nonchromatin substructures of the nucleus: the

ribonucleoprotein (RNP)-containing and RNP-depleted matrices analyzed by sequential fractionation and resinless

section electron microscopy. J Cell Biol 102: 1654-1665.

Galande S, Kohwi-Shigematsu T (1999): Poly(ADP-ribose) polymerase and Ku autoantigen form a complex

and synergistically bind to matrix attachment sequences. J Biol Chem 274: 20521-20528.

Gasser S M, Laemmli UK (1987): A glimpse at chromosomal order. Trends in Genetics 3: 16-22.

Gueth-Hallonet C, Wang J, Harborth J, Weber K, Osborn M (1998). Induction of a regular nuclear lattice by

overexpression of NuMA. Exp Cell Res 243: 434-452.

Herrscher RF, Kaplan MH, Lelsz DL, Das C, Scheuermann R, Tucker PW. (1995): The immunoglobulin

heavy-chain matrix-associating regions are bound by Bright: A B cell-specific trans-activator that describes a new

DNA- binding protein family. Genes Dev 9 3067-3082.

Hörtnagel K, Mautner J, Strobl LJ, Wolf DA, Christoph B, Geltinger, C Polack A (1995): The role of

immunoglobulin kappa elements in c-myc activation. Oncogene 10:1393-1401.

Hozak P, Sasseville AM -J, Raymond Y, Cook PR (1995): Lamin proteins form an internal nucleoskeleton as

well as a peripheral lamina in human cells. J Cell Sci 108: 635- 644.

Iber M, Schübeler D, Seibler J, Höxter M, Bode J (1999): Efficient FACS selection procedure for cells

undergoing Flp-mediated site-specific conversions. Trends Genet (tto01668: 1-4).

Ivanchenko M, Avramova Z (1992): Interaction of MAR- sequences with nuclear matrix proteins. J Cell

Biochem 50: 190-200.

Jankelevich S, Kolman JL, Bodnar J W, Miller G (1992): A nuclear matrix attachment region organizes the

Epstein-Barr viral plasmid in Raji cells into a single DNA domain. EMBO J 11: 1165-1176.

Jenuwein T, Forrester WC, Fernandez-Herrero LA, Laible G, Dull M, Grosschedl R (1997): Extension of

chromatin accessibility by nuclear matrix attachment regions. Nature (London) 385: 269-272.

Käs E, Poljak L, Adachi Y, Laemmli UK (1993): A model for chromatin opening: Stimulation of topoisomerase

II and restriction enzyme cleavage of chromatin by distamycin. EMBO J. 12: 115-126.

Kay V, Bode J (1994): Binding specificity of a nuclear scaffold: Supercoiled, single-stranded, and scaffold-

attached- region DNA. Biochemistry, 33: 367-374.

Kay V, Bode J (1995): Detection of scaffold-attached regions (SARs) by in vitro techniques; activities of these

elements in vivo. Meth Mol Cell Biol 5:186-194.

Kirillov A, Kistler B, Mostoslavsky R, Cedar H, Wirth T, Bergman Y (1996): A role for nuclear NF-KB in B- cell

specific demethylation of the Ig kappa locus. Nature Genet 13:, 435-441.

Klehr D, Bode J (1988): Comparative evaluation of bovine papilloma virus (BPV) vectors for the study of gene

expression in mammalian cells. Mol Gen (Life Sci Adv ) 7: 47-52

Kohwi-Shigematsu T, Maass K, Bode J (1997): A thymocyte factor SATB1 suppresses transcription of stably

integrated matrix-attachment region-linked reporter genes. Biochemistry 36: 12005-12010.

18

Kohwi-Shigematsu T, deBelle I, Dickinson LA, Galande S. KohwiY (1998): Identification of base-unpairing

region-binding proteins and characterization of their in vivo binding sequences. Meth Cell Biol 53: 323-354.

Krawetz SA, Kramer JA, McCarrey JR (1999): Reprogramming the male gamete genome: a window to

successful gene therapy. Gene 234: 1-9.

Ludérus MEE, Den Blaauwen, JL, De Smit OJB, Compton DA, Van Driel R (1994): Binding of matrix

attachment regions to lamin polymers involves single-stranded regions and the minor groove. Mol Cell Biol 14, 6297-

6305.

Mattern KA, van Driel R, de Jong L (1997): Composition and structure of the internal nuclear matrix.. Nuclear

Structure and Gene Expression 87-110.

Michalowski SM, Allen GC, Hall GE, Thompson WF, Spiker S (1999): Characterization of randomly-obtained

matrix attachment regions (MARs) from higher plants. Biochemistry in press.

Mielke C, Kohwi Y, Kohwi-Shigematsu T, Bode J (1990): Hierarchical binding of DNA fragments derived from

scaffold-attached regions: Correlations of properties in vitro and function in vivo. Biochemistry, 29, 7475-7485.

Mielke C, Maass K, Tümmler M, Bode, J (1996): Anatomy of highly expressing chromosomal sites targeted by

retroviral vectors. Biochemistry 35: 2239-2252.

Mirkovitch J, Mirault ME, Laemmli UK (1984): Organization of the higher-order chromatin loop: specific DNA

attachment sites on nuclear scaffold. Cell 39: 223-232.

Nakayasu H, Berezney R (1991): Nuclear matrins: Identification of the major nuclear matrix proteins. Proc Natl

Acad Sci USA 88: 10312-10316.

Nayler O, Strätling W, Bourquin JP, Stagljar I, Lindemann L, Jasper H, Hartmann AM, Fackelmayer FO, Ullrich

A, Stamm S (1998): SAF-B protein couples transcription and pre-mRNA splicing to SAR/MAR elements.. Nucl Acid

Res,26: 3542-3549.

Piechaczek C, Fetzer C, Baiker A, Bode J, Lipps HJ (1999): A vector based on the SV40 origin of replication

and chromosomal S/MARs replicates episomally in CHO cells. Nucl Acid Res 27: 426-428.

Poljak L, Seum C, Mattioni T, Laemmli UK (1994): SARs stimulate but do not confer position independent gene

expression. Nucl Acid Res 22: 4386-4394.

Pommier Y, Cockerill PN Kohn KW Garrard WT (1990). Identification within the simian virus 40 genome of a

chromosomal loop attachment site that contains topoisomerase II cleavage sites. J Virol 64: 419-423.

Schübeler D, Mielke C, Maass K, Bode J (1996). Scaffold/matrix-attached regions act upon transcription in a

context-dependent manner. Biochemistry 35: 11160-11169.

Schübeler D, Maass K, Bode J (1998): Retargeting of retroviral integration sites for the predictable expression

of transgenes and the analysis of cis-acting sequences. Biochemistry 37: 11907-11914.

Stein GS, VanWijnen AJ, Stein JL,Lian JB, Pockwinse SH, McNeil S (1999): Implications for interrelationships

between nuclear architecture and control of gene expression under microgravity conditions. Faseb J 13: S157-S166.

Strissel PL, Dann HA, Pomykala HM, Diaz MO, Rowley JD, Olopade OI (1998): Scaffold-associated regions in

the human type I interferon gene cluster on the short arm of chromosome 9th Genomics 47: 217-229.

Thompson EM, Christians E, Stinnakre M-G, Renard J-P (1994):Scaffold attachment regions stimulate

HSP70.1 expression in mouse preimplantation embryos but not in differentiated tissues. Mol Cell Biol 14: 4694-4703.

Traub P, Shoeman RL (1994): Intermediate filament proteins: ytoskeletal elements with gene-regulatory

function?. Int Rev Cytol 154: 1-104.

Tsutsui K (1998): Synthetic concatemers as artificial MAR: importance of a particular configuration of short AT-

tracts for protein recognition. Gene Ther Mol Biol 1: 581-590.

Tsutsui K, Tsutsui K, Okada S, Watarai S, Seki S, Yasuda T, Shohmori T (1993): Identification and

characterization of a nuclear scaffold protein that binds the matrix attachment region DNA. J Biol Chem, 268: 12886-

19 12894.

Weitzel JM, Buhrmester H, Strätling W H (1997): Chicken MAR-binding protein ARBP is homologous to rat

methyl- CpG-binding protein MeCP2nd. Mol Cell Biol 17: 5656-5666.

Wiersma EJ, Ronai D, Berru M, Tsui FWL Shulman MJ (1999): Role of the intronic elements in the

endogenous immunoglobulin heavy chain locus - Either the matrix attachment regions or the core enhancer is sufficient

to maintain expression. J Biol Chem 274: 4858-4862.

Will K, Warnecke G, Wiesmüller L, Deppert, W (1998): Specific interaction of mutant p53 with regions of MARs

with a high potential for base unpairing. Proc Natl Acad Sci USA 95: 13681-13686

Zong RT, Scheuermann RH (1995): Mutually exclusive interaction of a novel matrix attachment region binding

protein and the NF-µR enhancer repressor - Implications for regulation of immunoglobulin heavy chain expression. J

Biol Chem 270: 24010-24018.

Zuckerkandl E, Villet R (1988): Generation of high specificity of effect through low-specificity binding of proteins

to DNA. FEBS Lett. 231, 291-298.

20 Figure 1: DNA Recognition Profile of a Nuclear Scaffold

A nuclear scaffold has been prepared by LIS-(lithium 3,5-diiodosalicylate-) extraction of cultured

murine cells. The reassociation of a pUC-based vector with an 800 bp S/MAR insert (`IVA) is

studied in the presence of a vast (100 000x) excess of bacterial competitor DNA. Due to

nucleolytic activities of a scaffold an initially supercoiled S/MAR vector (see total input, T) becomes

first nicked, then linearized during the reassociation process. Traces P (pellet) and S (supernatant

mark fragments which associate with the scaffold or remain unbound. The supercoil strongly

associates with the scaffold (i.e. occurs in the pellet fraction as does a S/MAR-free control). Unlike

the S/MAR-free control the S/MAR vector exhibits evident but minor affinity for the scaffold, i.e. it is

distributed over the P- (15%) and S- (85%) fractions. Complete scaffold association after

linearization is exclusively found for the S/MAR containing vector.

Figure 2: Anatomy of a Prototype S/MAR: Stress-Induced Duplex Destabilization (SIDD)

Although S/MARs are not necessarily AT-rich, they contain many AT-patches which form the basis

of their local strand separation under stress. While stress in vivo can arise by various mechanisms,

a convenient way of its standardization is the application of superhelical tension in the context of a

bacterial plasmid (i.e. at superhelical densities of σ = -0.055). Under these conditions very

restricted regions of the plasmids open and can be trapped by single-strand specific agents (see

the peaks flanking the Ampicillin-resistance gene and the phage f1-ORI). Under the same

conditions S/MARs exhibit a multitude of more or less regularly spaced unwinding elements (UEs).

Among these the most prominent one (the one reacting preferentially with KMnO4; insert shows

KMnO4 reactivity in the living cell) is called the `core-unwinding elementA (CUE). Altogether the

UEs in a S/MAR form a `base-unpairing regionA (BUR).

Figure 3: Transcriptional Augmentation: Relevance of Distance

S/MAR-elements have been inserted at various positions along a bicistronic retroviral vector

construct. If this construct is transfected and its expression monitored during the transient

expression phase S/MARs are seen to exhibit a slightly negative effect throughout. If an authentic

single copy of the construct is integrated by retroviral infection and its expression is studied, the

initiation rate is significantly augmented relative to the control (Sp) if the S/MAR is at a certain

distance downstream from the promoter (construct c) or downstream and upstream from the

promoter (´double-copy´ vector 5´3´). At a position immediately downstream from the

transcriptional initiation site S/MARs inhibit the passage of RNA polymerase. A possible

mechanism is discussed in Fig. 4.

Figure 4: Augmentation explained by SIDD-properties of S/MARs

21 At least some S/MARs are particularly accessible to single-strand specific agents like

chloroacetaldehyde (CAA, Bode et al., 1997), OsO4 (Paul and Ferl, 1993) or single-strand specific

nucleases (Targa et al. 1994, Iarovaia et al., 1995). The association of ss-specific proteins

(triangles) would lead to strand separation within S/MARs and raise transient topological problems

(B). Their ultimate resolution by topoisomerases leads to the scaffold-associated ground state (C).

Transcription-dependent buildup of positive superhelicity can then be resolved by breaking some

(or all) ssDNA contacts enabling the immediate initiation by other polymerases.

Table 1: Ubiquitous S/MAR-Binding Proteins with a Relaxed Sequence Specificity

(Prototype S/MAR-Binders)

Prominent S/MAR binders have been listed and their binding specificity indicated. The binding of

supercoils by the nuclear scaffold has been described before (Tsutsui et al., 1988; Kay and Bode,

1994) but is not the subject of the present study which describes the recognition of linearized

S/MARs. It is evident that the specificity of the S/MAR-scaffold interaction is reflected by the major,

not tissue-specific, protein constituents.

22

Protein

Recognition Function

Name

Alternative Function

Abundance per Nucleus (Scaffold)

RNA

scDNA

ssDNA

ds S/MAR

Else

TopoII (Sc1)

Enzymatic plus structural functions; Sc1xSc2 (=UB2)

3E5 (60-80% in metaphase)

++

-

+

minor groove

HMG I(Y)

Nucleosome phasing

Chromatin bound

+ 3 AT-hooks

minor groove; bending protein

HMG 1,2

1E5 Chromatin bound

++

+

+ Cruciforms

bending protein

H1

25E6 (1-15%)

++

-

+ (S/TPXX)

ARBP

MeCP2

1E5

-

+ KD = 2E-10M

minor and major groove (GGTGT); methyl. CpGs; ds S/MAR>350 bp

Lamins A,B

also: internal network

100%

+

+ end domains (S/TPXX)

minor groove; 80-90% competition by ssDNA; noncooperative binding

NuMA

mitotic spindle maintenance

2E5 (most)

+

+

Nucleolin

rDNA transcript., rRNA packaging, ribosome assembly

(variable)

+

+

+ KD = 9E-9M

BUR, T-rich strand

PARP

DNA-repair; complex with DNA-PK (Ku-antigen)

1E6 (50%)

+

+

BUR under superhel. tension

SAF-A

hRNP-U; packaging of hnRNAs

2E6 (50%)

+

+ (RGG-box)

+ (SAF-Box) KD(min) = 3E-9M

AT-patches in dsS/MARs>300 bp, no competition by ss DNA; cooperative binding

SAF-B

Platform for transcriptosome/ splicing complex assembly

1E5 (5%) Chromatin bound

+ (SAF-Box)

Interaction with spliceosomes and Pol II

TOTAL SCAFFOLD

++

++

++

BUR; 50% competition by ssDNA; dsS/MAR > 300 bp cooperative binding

23

24

25

26

27