Skip Navigation

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (1204K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (24)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Rollini, P.
Right arrow Articles by Fournier, R. E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Rollini, P.
Right arrow Articles by Fournier, R. E.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Nucleic Acids Research 27:3779-3791 (1999)
© 1999 Oxford University Press

Article

Identification and characterization of nuclear matrix-attachment regions in the human serpin gene cluster at 14q32.1

Pierre Rollini, Stephanie J. Namciu, Mark D. Marsden and R. E. K. Fourniera

Division of Basic Sciences, A2-025, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, PO Box 19024, Seattle, WA 98109-1024, USA


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Matrix-attachment regions (MARs) are DNA elements that are defined by their abilities to bind to isolated nuclear matrices in vitro. The DNA sequences of different matrix-binding elements vary widely. The locations of some MARs at the ends of chromatin loops suggest that they may represent boundaries of individual chromatin domains. As such, MARs may play important roles in regulating transcription and chromatin structure. As a first step towards assessing the roles of MARs in these processes, we assayed DNA sequences from the human serine protease inhibitor (serpin) gene cluster at 14q32.1 for matrix-binding activity in vitro. This ~150 kb region contains the cell-specific genes encoding {alpha}1-anti­trypsin ({alpha}1AT) and corticosteroid-binding globulin (CBG), as well as an antitrypsin-related sequence termed ATR. A DNase I-hypersensitive site (DHS) map of the locus has recently been described. We report here that the {alpha}1AT–ATR–CBG region contains five distinct MARs. There is a strong matrix-binding element ~16 kb upstream of {alpha}1AT; three MARs are between ATR and CBG and one MAR is within the CBG gene itself. These MARs were matrix-associated in all cell types examined. DNA sequencing indicated that the serpin MARs contained predominantly repetitive DNA, although the types of DNA repeats differed among the MARs.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Eukaryotic chromatin is thought to be organized into a series of individual loops or domains (1). The bases of these loops seem to be defined by specific DNA elements that are involved in attachment of the chromatin fiber to a scaffold of non-histone chromosomal proteins that is called the nuclear matrix. These specific DNA elements have been termed scaffold-associated regions, matrix-associated regions or matrix-attachment regions (MARs) (2–4).

MARs are defined and identified by in vitro biochemical tests. Specifically, MARs are genomic DNA fragments that have the ability to bind to isolated nuclear matrices in vitro. Nuclear matrices are particulate protein structures that are prepared by histone depletion of isolated nuclei (2,4–6). MARs generally bind to nuclear matrices of diverse species (5,7). In a few instances MARs have been shown to co-localize with the limits of DNase I-sensitive domains of tissue-specific genes (3,8), suggesting that MARs may function as structural boundaries of individual chromatin domains (9). Consistent with this view, human MARs can function as insulator elements in Drosophila, shielding integrated transgenes from chromosomal position effects in vivo (10). Other MARs have been found in promoter regions or within introns, where they may function in the regulation of transcription, and some MARs seem to stimulate expression of heterologous reporters in stably but not transiently transfected cells (reviewed in 11–13). However, the functions of MARs as assessed by transfection experiments remain equivocal. Finally, MARs may be cis elements of chromosome dynamics, regulating chromosome shape and maintenance (14,15).

Although MARs tend to be AT-rich (12), they lack readily defined consensus motifs that mediate matrix binding. This raises the possibility that the matrix-binding activity of these elements may be encoded by complex sequence motifs that may vary between and within genomes. These kinds of motifs may affect higher order DNA structure. On the other hand, a number of simple sequence motifs have been reported to be enriched in DNAs with matrix-binding activities (16). These include various binding sites for topoisomerase II (17,18), DNA unwinding motifs (19) and simple sequence motifs called the A-, T- (6) and H-boxes (20). However, a MAR consensus sequence has been described in Arabidopsis (21). Clearly, additional sequence information from a diverse collection of MARs will be required to define the determinants of matrix-binding activity.

We have used the serine protease inhibitor (serpin) (for a review see 22) gene cluster on human chromosome 14q32.1 as a model system to study the regulation of gene activity and chromatin structure within individual chromatin domains. The serpins are a large family of proteins that are related by descent, but they have evolved to perform many different functions in vivo. Thus, many of the serpin genes are specifically expressed in different cell types. Furthermore, although serpin genes have been mapped on many different human chromosomes, some serpin genes are organized into discrete gene clusters. For example, we recently described the genomic organization of an ~370 kb region of human chromosome 14q32.1 that includes six serpin genes (23,24). This interval is organized into two discrete subclusters of three genes each that share a common genomic organization. The distal subcluster includes the genes encoding kallistatin (gene symbol PI4), protein C inhibitor (PCI) and {alpha}1-antichymotrypsin (AACT), and all three genes are transcribed in a proximal->distal orientation. The other serpin subcluster is located ~170 kb more proximal on the chromosome; it includes {alpha}1-antitrypsin ({alpha}1AT, gene symbol PI), corticosteroid-binding globulin (CBG) and an {alpha}1AT-related sequence termed ATR (gene symbol PIL). This subcluster is transcribed distal->proximal. All of the 14q32.1 serpin genes except ATR, which is thought to be a pseudogene (25–27), are highly transcribed in the liver and in cultured hepatoma cell lines, but they are not expressed in most other tissues. However, some cell types express a subset of the 14q32.1 serpin genes.

The organization of the 14q32.1 serpin gene cluster into two subclusters of three genes each separated by ~170 kb of genomic DNA suggested that the subclusters might be independently regulated, possibly by being sequestered into distinct chromatin domains (9). Conversely, the tight linkage of genes within each subcluster raised the possibility that they might share regulatory elements. To explore these issues, we concentrated our initial efforts on characterizing the proximal serpin subcluster, which includes {alpha}1AT, ATR and CBG. These three genes are arranged in a head-to-tail orientation, with ~12 kb between {alpha}1AT and ATR and ~40 kb between ATR and CBG (23). A detailed analysis of the chromatin organization of the region was accomplished by mapping DNase I-hypersensitive sites (DHSs) (28). Twenty-two expression-associated DHSs were found in this ~130 kb region, as well as seven additional sites that were present in all cell types. To gain insight into the domain organization of the locus and to assess the potential roles of MARs in controlling gene activity and/or chromatin structure, we assayed DNA fragments from an ~150 kb region around the {alpha}1AT, ATR and CBG genes for nuclear matrix-binding activity in vitro.

Three different matrix-binding assays were used to map MARs in the {alpha}1AT–ATR–CBG region and five MARs, which ranged in size from ~1.3 to ~8.4 kb, were identified. Four of the MARs were in intergenic regions, with one upstream of {alpha}1AT and three between ATR and CBG, and one MAR was located in CBG intron 1. We also mapped MARs in a serpin allele that had been activated by chromosome transfer and in cosmid transfectants that contained the {alpha}1AT–ATR–CBG locus integrated at novel chromosomal sites. The human MARs were matrix-associated in every case. Finally, we sequenced the MARs and compared their sequences with those of surrounding DNA. Strikingly, all five MARs contained highly repetitive DNA, although the kinds of repeats differed among the MARs. These studies provide a framework for the functional analysis of MARs within the {alpha}1AT–ATR–CBG domain.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Sequence names and accession numbers
The sequences of the DNA clones reported here are available under the following GenBank accession numbers: the {alpha}1AT MAR, accession no. AF156542; the CBG promoter (CBGp) MAR, accession no. AF156543; the CBG intron (CBGi) MAR, accession no. AF156544; 30 461 bp of contiguous sequence (accession no. AF156545) between ATR and CBG, which includes the ATR MAR, CBG 5" MAR and an ~15 kb inter-MAR segment.

Cell lines, culture conditions and cosmid transfections
The human hepatocellular carcinoma cell line HepG2, the human cervical carcinoma line HeLa S3 and the human monocytic cell line THP-1 were obtained from the American Type Culture collection. F(14n)2 is a rat hepatoma microcell hybrid that contains a single copy of human chromosome 14; it was prepared as described (29). The cells were cultured in 1:1 Ham’s F12/Dulbecco’s modified Eagle’s medium (F/DV) supplemented with 10% fetal bovine serum (FBS) (Gibco). F(14n)2 cells were grown in medium containing 250 µg/ml G418. For cosmid transfections, exponentially growing Fao-1 rat hepatoma cells (~1.2 x 107 cells/transfection) were harvested and suspended in 1 ml of ice-cold FDV, and 12 µg of MluI-linearized Ycos72 or Ycos65 (24) were added. These cosmid vectors contain a neo expression cassette. The cells were electroporated at 960 µF and 300 V using a Bio-Rad Gene Pulser. Cells were incubated for ~30 h in non-selective medium (F/DV + 10% FBS) before adding 500 µg/ml G418. After ~3 weeks, clones were picked and expanded.

Cosmid and plasmid subclones
Cosmids {alpha}ATc1, {alpha}ATc4, Ycos65 and Ycos72 have been described (23–25,30). Appropriate subclones were prepared in pGEM-2 (Promega) or pBluescript II KS+ (Stratagene) by standard procedures (31). The restriction maps of relevant subclones are shown in Figures 2–5.



View larger version (65K):
[in this window]
[in a new window]
  		Figure 2. A MAR upstream of {alpha}1AT. (A) DNA map of the {alpha}1AT gene and upstream region. The map is drawn to scale, with position 0 defined as the EcoRI site in the macrophage-specific exon IA of {alpha}1AT (23,24). The {alpha}1AT gene is indicated as a black box, with the arrow showing its transcriptional orientation. The matrix-associated region ~16 kb upstream of {alpha}1AT is shown as a stippled box. Relevant cosmids and cosmid subclones are shown below the map. An enlarged view of an ~8.6 kb HindIII subclone from Ycos117 is shown below, and relevant restriction sites, matrix-binding fragments and probes are indicated. (B) In vitro MAR assay II using nuclear matrices of THP-1 cells extracted with 2 M NaCl. After digestion of the ‘halo’ DNA with a combination of restriction enzymes, the isolated matrices were incubated with end-labeled probe mixtures. Matrix-bound (P, pellet) and released (S, supernatant) DNA fragments were separated by centrifugation and analyzed by agarose gel electrophoresis and autoradiography. For end-labeling, the relevant plasmid subclones were digested with the following restriction enzymes: p20 kb Spe (an ~20 kb SpeI subclone from Ycos117) with EcoRI and BamHI; p12.5N/S (an ~12.5 kb NotI–SpeI subclone from Ycos117) with EcoRI, HindIII and NotI; p17.6S/N (an ~17.6 kb SpeI–NotI subclone from Ycos126) with BssHII, EcoRI, HindIII and SacII. Arrows indicate matrix-binding fragments (a and b) of the 8.6 kb HindIII subclone shown in (A). For the 20 kb SpeI subclone, an aliquot of the digested, end-labeled probe mixture (Probe) is shown. (C) In vitro MAR assay II as in (B) using various restriction digests of the 8.6 kb HindIII subclone shown in (A). (D) In vivo MAR assay. HepG2 or HeLa nuclei were extracted with LIS, and the resulting nuclear ‘halos’ were digested with BamHI, BglII and EcoRI. Matrix-associated (P) and released (S) DNA fragments were separated by centrifugation, purified and 5 µg of each fraction was analyzed by Southern hybridization using the probes indicated in (A). Total (T) genomic DNA purified from each cell line was used as a control. Lanes marked M contain labeled 1 kb DNA ladders (Gibco).

 


View larger version (49K):
[in this window]
[in a new window]
  		Figure 5. MARs in and around CBG. (A) HepG2 nuclei were extracted with 2 M NaCl and in vitro MAR assay II was performed using restriction-digested subclones 1–4 (B) as probes. The asterisks indicate DNA fragments with matrix-binding activity; the diamonds indicate vector DNA fragments. (B) DNA map of the CBG region. Exons are indicated as black boxes, and the positions of three MARs are shown as stippled boxes. Subclones 1–4 are indicated below the map. The different subclones and the restriction enzymes used to prepare probe mixtures were as follows: 1, an ~10.4 kb BamHI subclone from cosmid {alpha}ATc4 (Fig. 6) digested with BamHI, EcoRI, HindIII, NheI and XbaI; 2, an ~6.5 kb XbaI subclone from a HepG2 library (23) digested with HindIII and XbaI; 3, an ~7.8 kb HindIII subclone from a CBG-containing phage (32) digested with EcoRI, HindIII, NcoI and XbaI; 4, an ~10.3 kb EcoRI subclone from a CBG phage (32) digested with BglII, EcoRI, HindIII and TaqI. Fragments x, y and z were the 8.4 kb XbaI–BglII, 1.1 kb HindIII–XbaI and 5.3 kb PvuII–HindIII DNA fragments, respectively, whose sequences were determined.

 
In vitro MAR assay I: DNA- and histone-depleted nuclear matrices
These binding assays were performed as described (2). Briefly, purified nuclei were depleted of histones and DNA by exhaustive DNase I digestion followed by extraction with 2 M NaCl. Exogenous labeled DNA fragments were then incubated with the resulting nuclear matrices in the presence of increasing amounts of competitor Escherichia coli DNA. The matrices were pelleted by centrifugation, and bound DNA fragments were visualized by agarose gel electrophoresis and autoradio­graphy, as shown schematically in Figure 1.



View larger version (28K):
[in this window]
[in a new window]
  		Figure 1. MAR mapping assays. (Left) In vitro MAR assay I (2). Purified nuclei are extracted with high salt and digested exhaustively with DNase I. DNA- and histone-depleted nuclear matrices are incubated with end-labeled probe fragments in the presence of increasing concentrations of unlabeled competitor DNA. Matrix-associated and unbound probe fragments are separated by centrifugation and analyzed by gel electrophoresis and autoradiography. (Center) In vitro MAR assay II (5,6). Nuclei are isolated and histones and other non-histone proteins are extracted using either 2 M NaCl or lithium diiodosalicylate (LIS). The resulting ‘nuclear halos’ are digested with restriction endonucleases and incubated in the presence of end-labeled DNA probes from the region of interest. These probes compete with endogenous DNA sequences for matrix binding. Bound (P, pellet) and unbound (S, supernatant) probe fragments are separated by centrifugation and analyzed by gel electrophoresis and autoradiography. (Right) In vivo MAR assay (4). This assay measures the partitioning of endogenous DNA sequences between matrix-associated and non-matrix fractions. Purified nuclei are extracted with LIS, and the ‘halo’ DNAs are digested with combinations of restriction enzymes. The matrices are pelleted by centrifugation, and the partitioning of specific DNA sequences into matrix-associated (pellet) and non-associated (supernatant) fractions is assessed by Southern hybridization. Adapted from Cockerill and Garrard (2).

 
In vitro MAR assay II: histone-depleted nuclear matrices
These assays were performed essentially as described (5,6), with minor modifications. Cells were grown, harvested and washed three times in isolation buffer [10 mM Tris–HCl, pH 7.4, 40 mM KCl, 0.25 mM spermidine, 0.1 mM spermine, 1 mM EDTA/KOH, pH 7.4, 20 µg/ml Aprotinin (Sigma), 1% thiodiglycol]. The cells were resuspended in ice-cold homo­genization buffer [isolation buffer containing 0.1% digitonin (Fluka)], homogenized with a B-type pestle in a Dounce homogenizer, and the nuclei were washed three times in ice-cold homogenization buffer. Nuclei (at 10–20 A260 units/ml) were stored for up to 1 month at –20°C in homogenization buffer containing 50% glycerol.

Purified nuclei (~1 A260 unit/assay) were washed two or three times in ~1 ml of wash buffer [5 mM Tris–HCl, pH 7.4, 20 mM KCl, 0.25 mM spermidine, 0.1 mM spermine, 1 mM EDTA/KOH, pH 7.4, 20 µg/ml Aprotinin, 1% thiodiglycol, 0.1% digitonin (Fluka)], resuspended in 50 µl of wash buffer and stabilized for 20 min at 42°C. Nuclei were then extracted (1 ml per assay, 10 min at room temperature) using either 2 M NaCl or 25 mM lithium 3,5-diiodosalicylate (LIS) (Sigma) in a buffer containing 5 mM HEPES, pH 7.4, 0.25 mM spermidine, 2 mM EDTA/KOH (pH 7.4), 2 mM KCl and 0.1% digitonin. The histone-depleted nuclear ‘halos’ were pelleted at 6000 g for 15 min at room temperature. The resulting soft pellets were carefully washed (two or three times if nuclei were extracted with 2 M NaCl, five or six times if extracted with LIS) in digestion buffer (20 mM Tris–HCl, pH 7.4, 20 mM KCl, 70 mM NaCl, 10 mM MgCl2, 0.125 mM spermidine, 0.05 mM spermine, 20 µg/ml Aprotinin, 0.1% digitonin). The ‘halo’ DNA was released by digestion with a combination of restriction enzymes (50–100 U/A260 unit of nuclei) for 5–6 h at 37°C. Digestion was stopped by adding EDTA to 20 mM, and a probe mixture (5–10 ng of a 32P-end-labeled test plasmid that had been digested with multiple restriction enzymes) was added and incubated overnight at 37°C. The restriction enzymes used to digest the ‘halo’ DNA were generally the same as those used to prepare the probe fragments. Occasionally, one or more additional restriction enzymes were added. Matrix-associated and released DNAs were separated by centrifugation (10 min at 10 000 g, room temperature) and the DNA purified (31) and analyzed by agarose gel electrophoresis and autoradiography. Equal amounts of DNA were loaded in each lane.

In vivo MAR assay
This assay was as described by Mirkovitch et al. (4). Five to seven A260 units of nuclei purified as described above were washed, resuspended in 250 µl of wash buffer, stabilized (20 min at 42°C) and extracted with LIS (9 ml of extraction buffer containing 25 mM LIS) as described above (Fig. 1). Following centrifugation (~5000 g for 20 min at room temperature), the soft pellets were carefully washed five or six times in digestion buffer. ‘Halo’ DNA was released by digestion with a combination of restriction enzymes (usually between three and six enzymes at 200–250 U each) overnight at 37°C. Matrix-bound and released DNAs were separated by centrifugation, purified (31) and quantified. DNA (5 µg) from each fraction was analyzed by Southern hybridization as described previously (28).

DNA sequence analyses
The MARs identified in this study were subcloned into pGEM-2 (Promega), and both strands were sequenced by dye terminator cycle sequencing on an ABI Prism 377 DNA Sequencer. In addition, a series of overlapping plasmid subclones that contained an ~30 kb genomic interval between ATR and CBG were sequenced. This region includes the ATR MAR (~3.7 kb) and the CBG 5" MAR (~8.4 kb) plus an inter-MAR segment of ~15.7 kb. The MAR-containing restriction fragments that were analyzed and the corresponding GenBank accession numbers were as follows: the {alpha}1AT MAR, the 1274 bp EcoRI fragment labeled a in Figure 2A (accession no. AF156542); the CBG promoter (CBGp) MAR, the 1115 bp HindIII–XbaI fragment labeled y in Figure 5B (accession no. AF156543); the CBG intron (CBGi) MAR, the 5303 bp PvuII–HindIII fragment labeled z in Figure 5B (accession no. AF156544); a contiguous sequence of 30 461 bp (accession no. AF156545) between ATR and CBG. This sequence overlapped the previously published ATR pseudogene sequence (previously assigned accession no. M19684) by 49 bp, and it included 2613 bp from ATR to the ATR MAR, the ATR MAR (the 3721 bp XhoI–NheI fragment labeled d in Fig. 3C), an inter-MAR interval of 15 709 bp and the CBG 5" MAR (the 8418 bp XbaI–BglII fragment labeled x in Fig. 5B).



View larger version (46K):
[in this window]
[in a new window]
  		Figure 3. A MAR downstream of ATR. (A) Histone- and DNA-depleted nuclear matrices from HepG2 cells were prepared and used in in vitro MAR assay I. Nuclear matrices were incubated in a typical binding assay with 32P-labeled fragments from a 12 kb BamHI–SalI subclone (C) that had been digested with BglII, NcoI, PvuI, XbaI and XhoI. DNA fragments that displayed matrix-binding activity in the presence of increasing concentrations of unlabeled competitor DNA (100, 300 and 700 µg/ml E.coli DNA) are indicated by the arrows; their locations on the DNA map of the region are shown in (C). (B) HeLa S3 cell nuclei were extracted with either 2 M NaCl or LIS, and the resulting nuclear matrices were used in in vitro MAR assay II using the end-labeled probe described in (A). After centrifugation, matrix-bound (P, pellet) and released (S, supernatant) DNA fragments were purified and analyzed by agarose gel electrophoresis and autoradiography. (C) Map of ATR and its 3" flanking region. ATR exons are indicated and location of the ATR MAR is shown. Positions of relevant restriction sites and plasmid subclones are shown. The ~12 kb BamHI–SalI and ~4.1 kb XhoI–SalI subclones were derived from cosmid {alpha}ATc1, whereas the ~8.8 kb XbaI subclone was derived from cosmid {alpha}ATc4 (Fig. 6). The locations of DNA fragments a, b and c of (A) and (B) are indicated; d is the 3.6 kb XhoI–NheI fragment whose sequence was determined.

 
Computational analysis
Interspersed repeats in MAR and inter-MAR DNA sequences were identified using RepeatMasker2 (http://ftp.genome. washington.edu/cgi-bin/RM2_resp.pl ). This program compares test sequences with a database of repetitive DNA elements, returning a masked query sequence for database searching and a table of the specific types and positions of repetitive elements in each test sequence. To examine the A/T composition of test sequences, the Windows program of the GCG sequence analysis software package v.9.1 (Genetics Computer Group, Madison, WI) was employed with a window setting of 100 bp and increments of 3 bp. Putative MAR motifs were mapped using the GCG Findpatterns program. Potential exons were identified using GRAIL 1.0 (http://avalon. epm.ornl.gov/Grail-bin/GrailForm_post ) and GENSCAN (http://genomic.stanford.edu/GENSCANW.html ). Putative exons were compared with sequences in the GenBank non-redundant, EST and STS databases using BLAST (http://www.ncbi.nml.nih.gov/BLAST/ ).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Experimental strategy
MARs are defined operationally as DNA fragments that can specifically bind to isolated nuclear matrices, residual structures produced by histone-depletion of nuclei. Three different methods were used to test DNA fragments from the {alpha}1AT–ATR–CBG locus for matrix-binding activity (Fig. 1). The first two methods have been termed in vitro MAR assays because the binding of exogenous, labeled DNA fragments to nuclear matrices is assessed. The third method has been termed an in vivo MAR assay; in this case, the partitioning of endogenous DNA sequences into matrix-associated versus non-associated fractions is assessed. The first method, which we term in vitro MAR assay I, uses DNA- and histone-depleted nuclei; it was originally described by Cockerill and Garrard (2). In this assay, nuclei from the cells of interest are purified and depleted of DNA by exhaustive DNase I digestion, and histones and other non-histone proteins are extracted with 2 M NaCl. This results in the generation of nuclear matrices that have been depleted of DNA, histones and other chromosomal proteins. These nuclear matrices are then incubated with exogenous, end-labeled DNA fragments from the region of interest in a binding reaction that includes increasing concentrations of unlabeled E.coli competitor DNA. Bound DNA fragments are collected by centrifugation, purified and visualized by agarose gel electrophoresis and autoradiography (Fig. 1).

The two other assay methods were developed by Laemmli and co-workers (4–6) (Fig. 1). In these assays, nuclei are isolated in a low salt/polyamine buffer in the presence of digitonin. In one assay, which we term in vitro MAR assay II, purified nuclei are extracted using either 2 M NaCl or lithium diiodosalicylate (LIS). This extracts most of the histones as well as some non-histone proteins. The resulting nuclear ‘halos’ are digested with restriction endonucleases and incubated in the presence of 32P-end-labeled DNA probes from the region of interest. These probes compete with endogenous DNA sequences for matrix binding. Bound and unbound probe fragments are separated by centrifugation and analyzed by gel electrophoresis and autoradiography. Finally, the in vivo MAR assay measures the partitioning of endogenous DNA sequences between matrix-containing and non-matrix fractions. In this assay, purified nuclei are extracted with LIS, and the released DNA is digested with a combination (usually three to five) of restriction enzymes. The matrices are pelleted by centrifugation, and the partitioning of specific DNA sequences into matrix-associated (pellet) and non-associated (super­natant) fractions is assessed by Southern hybridization using specific probes from the region of interest (Fig. 1).

We used all three MAR assays to assess the matrix-binding activities of DNA fragments from an ~150 kb region around {alpha}1AT–ATR–CBG. As positive controls for these experiments, the Drosophila histone MAR (4) and the human apolipoprotein B 3" MAR (8) were employed (data not shown). In a few instances, it was not possible to perform the in vivo MAR assay because particular DNA fragments in the region of interest could not be used as specific probes in Southern hybridizations because they contained highly repetitive DNA. Otherwise, all three methods were uniformly concordant in demonstrating matrix-binding or non-binding of DNA fragments throughout the locus, as summarized below.

A MAR upstream of {alpha}1AT
Matrix-associated regions upstream of the {alpha}1AT gene would be of interest as potential 5" boundaries of a putative {alpha}1AT chromatin domain. We used various subclones from the cosmid contig upstream of {alpha}1AT (24) to assay the region extending ~50 kb upstream of {alpha}1AT for matrix-binding activity. Using the three different matrix-binding assays, a MAR was identified ~15–17 kb upstream of {alpha}1AT. For example, when an ~20 kb SpeI subclone (Fig. 2A) extending from within the {alpha}1AT gene to ~17 kb upstream was digested with EcoRI and BamHI, a probe mixture containing >10 distinct DNA fragments was obtained (Fig. 2B, p20 kb Spe). When this probe mixture was incubated with restriction-digested, histone-depleted nuclear matrices (in vitro MAR assay II), one specific probe fragment, an EcoRI fragment of ~1.3 kb (Fig. 2B, fragment a), partitioned almost exclusively in the matrix-associated fraction. The other fragments in the probe mixture remained in the supernatant. Similar assays were performed using other DNA fragments from the region. For example, an ~12.5 kb NotI–SpeI subclone (Fig. 2A, p12.5N/S) just upstream of the SpeI fragment described above contained an ~1.0 kb subfragment that was weakly enriched in the matrix-associated fraction (Fig. 2B, fragment b), whereas the other fragments in this probe mixture remained in the supernatant. The DNA map of the region indicated that fragments a and b were adjacent in the genome (Fig. 2A). Other DNA probes from the region failed to display matrix-binding activity (Fig. 2B, p17.6S/N and data not shown). These data indicate that a single MAR is present in the region extending ~50 kb upstream of {alpha}1AT.

The matrix-binding activities of DNA sequences in the region 15–17 kb upstream of the {alpha}1AT gene, which we term the {alpha}1AT MAR, were confirmed in other in vitro MAR assays. In addition, the strong matrix-binding activity of fragment a and the weak binding of DNA fragments just upstream of a (fragment b in Fig. 2B and fragment c in Fig. 2C; these overlapping fragments differ in their 3"-ends; Fig. 2A) were clearly apparent in in vitro MAR assays using probe mixtures generated by digestion with other restriction endonucleases (Fig. 2C). For example, digestion of an 8.6 kb HindIII subclone (Fig. 2A) with EcoRI alone, EcoRI + XbaI or EcoRI + NcoI yielded a strongly binding fragment a plus a weakly binding fragment c (c was cut into two weakly binding subfragments of ~0.8 and ~0.7 kb by NcoI; Fig. 2A and C). Interestingly, when the probe was cut with BglII + HindIII, sequences from the a and c subfragments were present on restriction fragments of ~2.6 and ~1.3 kb, respectively. The larger DNA fragment (d in Fig. 2A) displayed less matrix-binding activity than was apparent when the same sequences were present on smaller DNA fragments (a in Fig. 2C). These and other data (not shown) emphasize the need to use a variety of restriction endonucleases to generate different probe mixtures that are required to rigorously assess the matrix-binding activities of specific DNA fragments. In general, matrix-binding activity was most readily apparent in these assays when the relevant DNA fragments were ~1–2 kb. Because of this size constraint in in vitro MAR assays, the precise boundaries of matrix-binding DNA fragments are generally difficult to define. We consider the {alpha}1AT 5" MAR to extend from ~17.5 to 15.2 kb upstream of {alpha}1AT, but the core fragment with strong matrix-binding activity is only ~1.3 kb (~16.5–15.2 kb).

The {alpha}1AT MAR was also analyzed using the in vivo MAR assay. Nuclei from human HepG2 and HeLa cells were extracted with LIS and digested with restriction endonucleases as described in Materials and Methods. Released and matrix-associated DNAs were separated by centrifugation and analyzed by gel electrophoresis and Southern hybridization (Fig. 2D). Probe 1 (Fig. 2A) was the ~1.3 kb EcoRI DNA fragment that had strong matrix binding activity in the in vitro MAR assay (Fig. 2B and C, fragment a). Genomic DNA that hybridized with this probe was found almost exclusively in the matrix-associated fraction in both cell lines. In contrast, DNA sequences homologous to probe 2, which was ~5 kb downstream of the {alpha}1AT MAR (Fig. 2A), were strongly enriched in the supernatant (Fig. 2D). These data confirmed results from the in vitro MAR assays and, further, they indicated that the {alpha}1AT MAR was matrix-associated in both expressing (HepG2) and non-expressing (HeLa) cell types. In vivo MAR assays using the ~1.5 kb EcoRI fragment c (Fig. 2A) as a hybridization probe revealed weak matrix binding (data not shown), consistent with data from in vitro MAR assays.

A MAR downstream of ATR
Plasmid subclones from cosmid {alpha}ATc1 (23,30) and other cosmids in the region (24) were used to assess matrix-binding of DNA fragments from the region extending from ~7 kb upstream to ~30 kb downstream of {alpha}1AT. This region includes both {alpha}1AT and ATR. One example showing results obtained using the two different in vitro MAR assays is shown in Figure 3. An ~12 kb BamHI–SalI subclone of {alpha}ATc1, containing ATR exons and downstream sequences, was digested with BglII, NcoI, PvuI, XbaI and XhoI, producing a probe mixture that contained 11 different restriction fragments. This probe mixture was incubated with either DNA- and histone-depleted nuclear matrices (in vitro MAR assay I; Fig. 3A) or histone-depleted nuclear matrices (in vitro MAR assay II; Fig. 3B). In both assays, three specific probe fragments were found to be matrix-associated. Furthermore, these fragments were matrix-associated in in vitro MAR assay II using either LIS- or high salt-extracted matrices (Fig. 3B). The three matrix-binding fragments identified in these assays (a–c) were contiguous in the genome (Fig. 3C), defining an ~3.7 kb MAR that is located ~3.0–6.7 kb downstream of ATR exon V. We call this matrix-binding element the ATR MAR. No other DNA fragments from the region ~7 kb upstream to ~30 kb downstream of {alpha}1AT had matrix-binding activity in these assays (data not shown).

Fine mapping of the ATR MAR was accomplished using both in vitro and in vivo assays (Fig. 4). For example, when an ~4.1 kb XhoI–SalI subclone (Fig. 4A) from cosmid {alpha}ATc1 (30) was digested with various restriction enzymes and used in in vitro MAR assay II, virtually all of the DNA fragments of the plasmid insert were matrix-binding (Fig. 4B). The various restriction fragments with matrix-binding activity (labeled a–f in Fig. 4B) are identified on the DNA map shown in Figure 4A. However, small DNA fragments (~0.4–0.6 kb) tended to bind less efficiently than larger fragments, with optimal fragment sizes being ~1–2 kb, as observed in previous experiments. Two examples of in vitro MAR assays using subclones of genomic DNA that did not contain matrix-binding sequences (Fig. 4B, p6.1 kb Xho and p4.7 kb Bam), both from within the {alpha}1AT gene (see Fig. 2A), are also shown.



View larger version (53K):
[in this window]
[in a new window]
  		Figure 4. Fine mapping the ATR MAR using in vitro and in vivo assays. (A) DNA map of the ATR MAR showing relevant restriction sites and probes used in MAR assays. Subclone p4.1 kb XhoI–SalI was derived from cosmid {alpha}ATc1 (Fig. 6). There were no BamHI, HindIII or XbaI sites in the region. (B) In vitro MAR assay II. Plasmid subclones were digested with the indicated restriction enzymes, end-labeled and used in binding assays with HeLa S3 nuclear matrices prepared using 2 M NaCl. S, supernatant (released) fraction; P, pellet (matrix-bound) fraction. p4.7 kb BamHI and p6.1 kb XhoI were plasmid subclones from cosmid {alpha}ATc1 (see Fig. 2A). (C) In vivo MAR assay. HeLa S3 (left and right) or HepG2 (center) nuclei were extracted with LIS, and the resulting nuclear ‘halos’ were washed extensively and digested with the indicated combinations of restriction enzymes. Matrix-bound (P) and released (S) DNA fragments were separated by centrifugation, purified and each fraction (5 µg) was analyzed by Southern hybridization. Total (T) genomic DNA from each cell line was used as a control. The probes used in the left and center panels are indicated in (A). Probe 1 contained some repetitive sequences that hybridized weakly with other genomic DNA fragments under the conditions used. Probe 2 cross-hybridized weakly to an ~0.8 kb EcoRI fragment (not matrix-bound) located at position ~+21 kb between {alpha}1AT and ATR (Fig. 6). The filter shown in the left panel was stripped and rehybridized to an ~1.8 kb BamHI–EcoRI fragment just upstream of {alpha}1AT (position –1.8 to 0 kb; Fig. 6) to generate the data shown in (C), right.

 
The ATR MAR was also analyzed using the in vivo MAR assay. Genomic DNA sequences homologous to probe 1, from within the ATR MAR (Fig. 4A), were enriched in the pellet fractions of HeLa cell nuclear matrices that had been digested with a variety of restriction enzymes (Fig. 4C, left). As observed previously in this and other reports (see 5 for example), the matrix-binding activities of specific DNA sequences were most readily apparent when they were present on ­relatively small restriction fragments. For example, digestion of HeLa cell nuclear matrices with BamHI, BglII and EcoRI generated an ~4.1 kb restriction fragment homologous to probe 1, and this fragment was moderately enriched in the matrix-associated fraction (Fig. 4C, left). However, digestion with BamHI, EcoRI, HindIII, NcoI and XbaI produced a fragment homologous to probe 1 of only ~1.1 kb, and this fragment ­partitioned almost exclusively in the matrix-associated fraction (Fig. 4C, left). Using another probe from the region (probe 2, Fig. 4A), the same sequences were found to be matrix-­associated in HepG2 nuclear matrices (Fig. 4C, center). Other sequences from the region, such as a probe just upstream (0 to –1.8 kb) of {alpha}1AT, were not matrix-associated (Fig. 4C, right). These data define an ATR MAR ~3.0–6.7 kb downstream of ATR exon V, and they confirm that it is the only matrix-­binding element in the region from ~7 kb upstream to ~30 kb downstream of the {alpha}1AT gene.

MARs in and around CBG
Three different MARs in the interval from the ATR MAR to ~20 kb downstream of CBG were identified using various in vitro and in vivo MAR assays. For example, results of an in vitro MAR assay II using 2 M NaCl-extracted nuclei from human HepG2 cells are shown in Figure 5. Probe 1 (Fig. 5B) was prepared by digesting an ~10.4 kb BamHI subclone from cosmid {alpha}ATc4 (Fig. 6) with BamHI, EcoRI, HindIII, NheI and XbaI. This produced a probe mixture containing 10 different human DNA fragments and one vector fragment (indicated by diamonds in Fig. 5A). Four of the human DNA fragments had matrix-binding activity (asterisks in Fig. 5A). These matrix-binding fragments were contiguous in the genome, defining an ~8.4 kb matrix-binding element (CBG 5" MAR) that is located ~9–17 kb 5" of the CBG gene. Probe 2 was an ~6.5 kb XbaI subclone from a HepG2 genomic library (23) that had been digested with HindIII and XbaI. One of the human DNA fragments of this probe mixture had matrix-binding activity (Fig. 5A, lane 2). Probe 3 was an ~7.8 kb HindIII subclone from a CBG-containing phage clone (32) that had been digested with EcoRI, HindIII, NcoI and XbaI. Three of the human DNA fragments in this probe mixture had matrix-binding activity (Fig. 5A, lane 3). Probe 4 was an ~10.3 kb EcoRI subclone from a CBG-containing phage clone (32) that had been digested with BglII, EcoRI, HindIII and TaqI. Four of the human DNA fragments in this probe mixture had matrix-binding activity (Fig. 5A, lane 4). Taken together, the in vitro matrix-binding data from probes 2, 3 and 4 define two distinct MARs in the CBG gene, an ~1.1 kb element in the 5" flanking region [2.3–1.2 kb upstream of exon I, the CBG promoter (CBGp) MAR] and an ~5.3 kb element in CBG intron 1 [the CBG intron (CBGi) MAR]. The CBG 5" MAR and CBGi MAR also displayed matrix-binding activities in the in vivo MAR assay, but the CBGp MAR could not be tested in this assay because it contained highly repetitive DNA (data not shown). No other DNA sequences in the region from the ATR MAR to ~20 kb downstream of CBG had matrix-binding activity (data not shown).



View larger version (12K):
[in this window]
[in a new window]
  		Figure 6. Chromatin organization of the {alpha}1AT–ATR–CBG region. The map is drawn to scale, with position 0 being the EcoRI site in exon IA of {alpha}1AT (23). Exons are indicated as black boxes, with arrows showing the transcriptional orientations of the genes. The five MARs that were found in the region from ~–47 to +108 kb are shown as stippled boxes. The positions of relevant cosmids used in subcloning and/or transfection experiments are indicated below the map. Twenty-nine DHSs that are present in the chromatin of human HepG2 cells are indicated. Long arrows indicate strong DHSs and short arrows weak sites, as described (28).

 
The locations of the five matrix-binding elements within the ~150 kb region that includes the human {alpha}1AT, ATR and CBG genes are summarized in Figure 6. The figure also shows the positions of 29 DNase I-hypersensitive sites (DHSs) that are present in the chromatin of {alpha}1AT- and CBG-expressing human HepG2 cells, as recently described (28). The map of this region suggests that {alpha}1AT and ATR may reside in an ~50 kb chromatin domain that is bounded by MARs, with CBG sequestered into an adjacent chromatin domain. Chromatin mapping experiments using cells that express {alpha}1AT but not CBG (33,34) and functional tests of mutant serpin alleles (35,36) should allow us to test the validity of this model.

Human MARs are matrix-associated in microcell hybrids and cosmid transfectants
To determine whether the human serpin MARs could mediate matrix attachment in heterologous cells, organizing the locus into an appropriate chromatin domain, microcell hybrids and cosmid transfectants were employed. We previously reported that transfer of a neo-marked human chromosome 14 from non-expressing fibroblasts to {alpha}1AT- and CBG-expressing rat hepatoma cells resulted in activation of human {alpha}1AT and CBG gene expression. This gene activation event was accompanied by chromatin reorganization of the entire ~150 kb locus (28), as both generalized sensitivity of the locus to DNase I and expression-associated DHSs were induced upon gene activation. All of these events required the cell-specific transactivators HNF-1{alpha} and HNF-4 (37). Therefore, we used the in vivo MAR assay to determine whether the human serpin MARs were matrix-associated in a rat hepatoma hybrid [F(14n)2] that contained human chromosome 14. We also analyzed matrix binding in cosmid transfectants in which parts of the human 14q32.1 serpin locus were integrated randomly into the rat genome. For these studies, cosmids Ycos72 and Ycos65 (24) were stably transfected into Fao-1 rat hepatoma cells by electroporation, and single copy transfectants (72A and 65M) were chosen for analysis. Both 72A and 65M expressed high levels of human {alpha}1AT mRNA (data not shown).

F(14n)2, 72A and 65M nuclei were extracted with LIS and digested with various restriction endonucleases, and the partitioning of specific DNA sequences into matrix-bound (pellet) and unbound (supernatant) fractions was assessed by Southern hybridization using specific DNA probes. Representative results of these in vivo MAR assays are shown in Figure 7. For example, Figure 7A shows the partitioning of four different DNA fragments in 72A and F(14n)2 nuclear matrices. Two of these DNA fragments were clearly matrix-associated in both the cosmid transfectant and the microcell hybrid (Fig. 7A, left panels); these DNA fragments were derived from the {alpha}1AT MAR and the CBGi MAR, respectively [probe sizes and positions on the DNA map (Fig. 6) are indicated in Fig. 7]. In contrast, DNA sequences upstream of {alpha}1AT (–10 kb) and within the {alpha}1AT gene were not matrix-associated (Fig. 7A, right panels). Similarly, DNA sequences from the ATR MAR (Fig. 7B, left) and the CBG 5" MAR (Fig. 7B, center) were matrix-associated in 65M and F(14n)2 cells, but sequences from {alpha}1AT exon II were not (Fig. 7B, right). These and other data (not shown) indicate that the two human MARs ({alpha}1AT and ATR MARs) that were present in the cosmid transfectants were matrix-associated in vivo. Furthermore, all four MARs that could be assayed using the in vivo test ({alpha}1AT, ATR, CBG 5" and CBGi MARs) were matrix-associated in F(14n)2 cells. Taken together with nuclease accessibility tests reported previously (28,37), these findings demonstrate that the human 14q32.1 serpin locus contains regulatory information that is sufficient for its organization into a characteristic chromatin state, as defined by DHSs, generalized nuclease sensitivity and specific matrix attachments, in heterologous cells.



View larger version (65K):
[in this window]
[in a new window]
  		Figure 7. Matrix associations in cell hybrids and transfectants. (A) In vivo MAR assay using cosmid transfectant 72A and microcell hybrid F(14n)2. The same gel was hybridized sequentially with the probes indicated below each panel. The restriction fragments used as probes and the positions of the (left) end of each probe fragment on the DNA map of the locus (Fig. 6) are shown. F(14n)21 and F(14n)22 are two different preparations of F(14n)2 nuclear matrices. The DNAs of LIS-extracted nuclei were digested with the following restriction enzymes: 72A and F(14n)21 with BglII, EcoRI, HindIII and XhoI; F(14n)22 with the same enzymes plus BamHI and PstI. The probes used in the two upper panels are probes 1 and 2 of Figure 2A. The 0.6 kb PvuII–BglII probe used in the lower left panel was from the CBG intronic (CBGi) MAR, whereas the 0.5 kb BamHI–SphI probe (lower right) is a fragment from within the {alpha}1AT gene. (B) The ATR MAR was analyzed (left) using an ~0.7 kb TaqI fragment (c in Fig. 4A) as probe. The right three lanes of the gel were subsequently rehybridized with a probe from the CBG 5" MAR (middle) and with a probe from {alpha}1AT exon II (right). Endogenous DNAs of the nuclear matrices were digested with the following enzymes: 65M and F(14n)23 with BamHI, BglII, EcoRI and PstI; F(14n)24 with BamHI, EcoRI, HindIII and XbaI.

 
Sequence analysis of the serpin MARs
The DNA sequences of the five serpin MARs and the ~15.6 kb genomic interval between the ATR MAR and the CBG 5" MAR were determined and compared. This analysis demonstrated that the serpin MARs are composed predominantly of interspersed repetitive DNA elements that are degenerate copies of retrotransposons (38,39). Figure 8 shows the distributions of LINEs (long interspersed nuclear elements, shown in green), SINEs (short interspersed nuclear elements, shown in red) and LTR-type retrotransposons (shown in blue) in each of the serpin MARs. The ATR, CBG 5", CBGp and CBGi MARs were composed of 83–93% repetitive DNA. LINEs (green) were the predominant class of repetitive DNA in these MARs, although a few SINEs (red) and LTR-type elements (blue) were also present in each MAR. In marked contrast, the {alpha}1AT MAR contained 85% repetitive DNA, all of which was of the LTR retrotransposon type (Fig. 8). Thus, the serpin MARs were highly enriched for repetitive DNA (the average density of interspersed repeats in the human genome is ~35%; 39), but the matrix-binding activities of these elements were not correlated with the presence of any particular type of interspersed repeat.



View larger version (24K):
[in this window]
[in a new window]
  		Figure 8. Sequence analysis of MARs. The upper bars show the distributions of interspersed repeat elements (SINEs, red; LINEs, green; LTR transposons, blue) within each serpin MAR, identified using RepeatMasker2 (http://ftp.genome.washington.edu/cgi-bin/RM2_resp.pl ). The percent AT in each sequence was determined and plotted using the Windows program of the GCG sequence analysis software package v.9.1 (Genetics Computer Group, Madison, WI), with a window setting of 100 bp and increments of 3 bp. The boxes at the bottom of the figure show the locations of putative MAR motifs, identified using the GCG Findpatterns program. These include DNA unwinding motifs (AATATATT and AATATT), H-boxes (uninterrupted A/T/C of at least 25 bp), A-boxes (AATAAAYAAA) and T-boxes (TTWTWTTWTT). Drosophila topoisomerase II (Topo II) binding sites (GTNWAYATTNATNNR) which had identity to the 6 bp core and 12 out of 15 bp matches overall are also indicated.

 
The arrangement of interspersed repeat elements within each MAR was complex. For example, the ~8.4 kb CBG 5" MAR was composed of parts of three different LINE elements, the L1MC3, L1MB8 and L1MA7 repeats of ~2, 1.0 and 1.8 kb, respectively. An Alu element (SINE) and an LTR retro­trans­poson were found within the L1MC3 repeat. Furthermore, the L1MB8 repeat was interrupted by the insertion of a Tigger2 transposon, whereas the L1MA7 repeat contained part of another LINE element plus a simple sequence repeat. ­More­over, all the Alu (SINE) elements within the MARs were insertions within individual LINE repeats. This mosaic pattern of interspersed repetitive DNA was typical of the ATR, CBG 5", CBGp and CBGi MARs (Fig. 8).

The DNA sequences of most matrix-binding elements are AT-rich, and the five serpin MARs shared this property. The ATR, CBG 5", CBGp and CBGi MARs all consisted of DNA that was ~65% AT, which is typical of MARs in general (11,12). The {alpha}1AT MAR was 75% AT-rich. The AT composition of each MAR varied along its length, with dramatic reductions in AT content within SINE elements, which were predominantly GC-rich Alu repeats (Fig. 8).

Several simple sequence motifs have been associated with MAR sequences (6,16–20). These include the A-box (AATAAAYAAA), the T-box (TTWTWTTWTT), DNA unwinding motifs (AATATATT, AATATT), SATB1 binding sites (H-box, A/T/C25) and consensus topoisomerase II (topo II) sites of vertebrates (RNYNNCNNGYNGKTNYNY) and Drosophila (GTNWAYATTNATNNR). Many of these motifs are AT-rich. Their distributions within the serpin MAR sequences are shown in Figure 8; data for both DNA strands are shown. The H-box was the most prevalent motif among the five serpin MARs. Each MAR also contained five or more degenerate Drosophila topo II sites (GTNWAYATTNATNNR, with identity in the 6 bp core and 12 of 15 matches overall). The A-box, T-box and DNA unwinding motifs were not highly represented among the MAR sequences. A single vertebrate topo II site was found in the CBGi MAR (not shown).

To determine whether any characteristic features might distinguish MAR sequences from their surrounding DNA, the DNA sequence of the inter-MAR segment between the ATR MAR and the CBG 5" MAR was determined. An analysis of the ~34 kb region from within ATR to the CBG 5" MAR is shown in Figure 9. The most striking result of this analysis was that virtually all of the LINE elements (green) in the region were within MARs. Other kinds of interspersed repetitive DNA elements were found throughout the inter-MAR region, whereas ATR itself was composed predominantly of unique sequence DNA. The ATR and CBG 5" MARs were more AT-rich (67 and 65%, respectively) than either inter-MAR (53%) or upstream (48%) DNA. The various simple sequence motifs were found throughout the region, with an apparent enrichment of topo II and H-box motifs within the ATR and CBG 5" MARs (Fig. 9).



View larger version (31K):
[in this window]
[in a new window]
  		Figure 9. Sequence analysis of the ATR–CBG 5" MAR interval. The figure shows an analysis of 34 171 bp of contiguous sequence extending from the ATR pseudogene (GenBank accession no. M19684) to the CBG 5" MAR. The map at the top of the figure shows the positions of ATR exons, the ATR and CBG 5" MARs and exon remnants identitied using GRAIL 1.0 (http://avalon.epm.ornl.gov/Grail-bin/GrailForm_post ) and GENSCAN (http://genomic.stanford.edu/GENSCANW.html ). Interspersed repeats, AT base composition and putative MAR motifs were identified as described in the legend to Figure 8.

 
Finally, we searched for potential protein coding sequences within MAR and inter-MAR sequences using RepeatMasker 2 and GENSCAN or GRAIL 1.0. GENSCAN found two potential protein coding exons in the inter-MAR region. The conceptual translation products of these sequences were 71 and 31 amino acid peptides with homology to two regions of opossum {alpha}1AT encoded within exon III. A putative 86 bp exon identified by GRAIL contained a 44 bp segment with 72% identity to opossum {alpha}1AT cDNA exon II (40). The positions of these putative exons are indicated in Figure 9. No ESTs or STSs were identified in the inter-MAR region or within the MARs themselves.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Matrix attachment regions are AT-rich DNA elements that are found throughout eukaryotic genomes. These elements are thought to play important roles in organizing higher order chromatin structure in both interphase nuclei and metaphase chromosomes (11,12). MARs have also been implicated in the regulation of gene expression. For example, some MARs are associated with enhancers or other cis-regulatory elements (2,6,41) and they can stimulate expression of stably integrated reporter genes in transfected cells (12,13,42). Other transfection studies suggest that MARs may possess insulator activity, shielding integrated transgenes from chromosomal position effects, but this remains controversial (11). Recently, human MARs were shown to encode an insulator function in a well-developed position effect assay in Drosophila (10). Clearly, a useful approach to study MAR function would be one in which MARs and other potential regulatory elements are modified systematically within individual human chromosomes, and the structures and functions of mutant alleles are rigorously assessed. The development of a system for efficient modification of human chromosomal loci by homologous recombination (35,36) provides a means to achieve this goal.

The human serpin gene cluster at 14q32.1 is an attractive model system to study the regulation of gene activity and chromatin structure. The serpins are an ancient, highly conserved gene family that is the product of divergent evolution (22,43). As such, serpin proteins perform a wide variety of functions in vivo, and they are specifically expressed in different cell types. Moreover, some serpin genes are organized into gene clusters. For example, the six serpin genes at 14q32.1 are organized into two distinct subclusters of three genes each, with a common arrangement within each subcluster (23,24). The proximal subcluster, which includes {alpha}1AT, ATR and CBG, is highly expressed in liver cells, but is transcriptionally repressed in most other cell types. These different functional states are correlated with different chromatin configurations of the locus in the respective cell types: expressing cells display 29 specific sites in this ~150 kb region that are hypersensitive to digestion with DNase I, but only seven of those DHSs are present in non-expressing cells (28). To gain insight into the potential chromatin domain organization of the {alpha}1AT–ATR–CBG locus, MAR mapping experiments were performed.

Using a combination of in vitro and in vivo MAR assays (2,4–6), five MARs were found in the ~150 kb {alpha}1AT–ATR–CBG region. The matrix-binding activities of all five elements were readily apparent and highly reproducible in both in vitro MAR assays; moreover, no other DNA fragments in the region had matrix-binding activity in these assays. Four of the five MARs we defined could also be tested using the in vivo MAR assay, and all four MARs were clearly matrix-associated in vivo. Furthermore, these in vivo associations were apparent in all cell types tested, so matrix-binding in this interval was not cell-specific. The positions of MARs within the locus raise the possibility that {alpha}1AT and ATR may reside within an ~50 kb chromatin domain that is bounded by the {alpha}1AT MAR and the ATR MAR, with CBG in an independently regulated chromatin domain. This model is consistent with observations that there are many DHSs around the {alpha}1AT gene but few around the CBG gene in macrophages (P.Rollini and R.E.K.Fournier, unpublished observations), which express {alpha}1AT but not CBG (33). It is not yet clear whether the putative 5" boundary of the CBG domain is the ATR MAR or the CBG 5" MAR, and the 3" boundary of this domain has yet to be defined.

Two MARs were located in the CBG gene: an ~1.1 kb MAR in the promoter region and an ~5.3 kb MAR in intron 1. Intronic MARs have been found in several genes, including the mouse immunoglobulin {kappa} light chain gene (2), the mouse heavy chain µ gene (41), the human ß-globin gene (44), the hamster dihydrofolate reductase gene (45) and the human topo I gene (46). In many cases, these MARs were associated with important cis-regulatory elements. For example, the intronic MAR of the mouse {kappa} gene is near a tissue-specific enhancer and both elements are required for the proper regulation of Ig{kappa} during development (47,48). A tissue-specific DHS is also present just 5" of the CBGi MAR (28), so this might represent a similar arrangement. We are currently testing whether the CBG promoter and/or intronic MARs regulate aspects of CBG transcription.

The MARs in the human {alpha}1AT–ATR–CBG locus were also matrix-associated in heterologous cells. These matrix associations were readily apparent when the in vivo MAR assay was used to assess matrix-binding of the serpin MARs in microcell hybrids, which contained human chromosome 14 in a rat hepatoma cell background, or in cosmid transfectants, which had parts of the human serpin locus integrated randomly into the rat genome. These microcell hybrids and cosmid transfectants also displayed the characteristic arrays of DHSs (28; unpublished observations). Taken together, these results indicate that the human serpin locus contains cis-regulatory information that is sufficient for establishing and maintaining a proper chromatin configuration of the locus in heterologous cells. Moreover, they indicate that the heterologous rat cells used as recipients in these experiments contain functionally conserved, trans-acting regulatory factors that are sufficient for the elaboration of these cell-specific chromatin structures.

DNA sequencing studies revealed that all five serpin MARs were composed predominantly of interspersed repetitive DNA elements that are degenerate copies of retrotransposons. These interspersed repeats represented 83–92% of each MAR sequence. The ATR, CBG 5", CBGp and CBGi MARs consisted mostly of LINE elements of the L1 family, although each MAR also contained LTR-type repeat elements and/or SINEs. In marked contrast, 85% of the {alpha}1AT MAR was composed of interspersed repetitive DNA, all of which was of the LTR type. Interspersed repeats have been observed in other human MARs (49–52), but, in general, these MARs contained much less repetitive DNA (~10–40%) than the serpin MARs. Other human MARs seem devoid of interspersed repeats (46,53), although they may contain other kinds of repetitive DNA, such as simple sequence repeats (8). Some rodent MAR sequences are enriched for interspersed repeats (54), whereas others are not (55). These observations suggest that while repetitive DNA is a common feature of MAR primary structure, there is no specific type of repeat that encodes matrix-binding activity. Conversely, not all interspersed repeats have matrix-binding activity. For example, the human ß-globin locus contains two ~5 kb LINEs of the L1 family (56), but neither of these LINEs has matrix-binding activity (44). We presume that there exist a variety of complex DNA sequence motifs that encode matrix-binding activity, and these sequences may vary widely both within and between species. The common association of interspersed repetitive DNA elements with MARs suggests that these exogenous, transposon-derived sequences may have evolved within the eukaryotic host to function in matrix-binding. It is also possible that some of the ancestral transposon insertions may have targeted matrix-binding sequences. In any event, the association of interspersed repeats with matrix-binding elements provides another example in which transposable elements have played important roles in the genomic evolution of eukaryotes (38,39).

To determine whether any aspects of DNA primary sequence might distinguish the serpin MARs from surrounding DNA, we compared the DNA sequences of the ATR and CBG 5" MARs with the 15.7 kb genomic region between them. The most obvious conclusion from this study was that LINE sequences were found almost exclusively within the MARs. In contrast, SINEs and LTR-type repeats were present both in the MARs and the inter-MAR segment. Another interesting feature of the inter-MAR region was the observation that although this interval contains cell-specific DHSs (28), no functional genes were present in the region: only small exon remnants related to {alpha}1AT were found. This suggests that the region downstream of ATR may be an ancient duplication of {alpha}1AT and/or ATR. If so, the ATR and CBG MARs might be related by descent. This possibility is also supported by the observation that the corresponding region of the distal 14q32.1 serpin subcluster, which is likely an ancestral duplication of the proximal subcluster (or vice versa), seems to lack the inter-MAR segment (24). The conservation of chromatin domain organization, as defined by gene and MAR arrangement, has been documented in plants (57).

The mapping of DHSs (28) and matrix-attachment regions in the {alpha}1AT–ATR–CBG locus provides insight into the chromatin organization of the locus in different cell types. This information provides a useful conceptual framework for the design of homologous recombination experiments in which the structure of the locus is modified systematically and the functions of specific mutant serpin alleles are critically assessed.


   ACKNOWLEDGEMENTS
 
We thank J. Mirkovitch for reagents and advice and Steve Henikoff and Meng-Chao Yao for their comments on the manuscript. These studies were supported by grant GM26449 from the National Institute of General Medical Sciences. P.R. was supported by fellowships from the Swiss National Science Foundation, the Swiss League against Cancer and the Schweizerische Stiftung für Medizinisch-Biologische Stipendien. S.N. was supported by fellowships from the National Institutes of Health (DK09188) and the Arnold Guild, Fred Hutchinson Cancer Research Center.


   FOOTNOTES
 
a To whom correspondence should be addressed. Tel: +1 206 667 5217; Fax: +1 206 667 6522; Email: kfournie{at}fhcrc.org Back

DDBJ/EMBL/GenBank accession nos AF156542–AF156545.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Paulson,J.R. and Laemmli,U.K. (1977) Cell, 12, 817–828.

  2. Cockerill,P.N. and Garrard,W.T. (1986) Cell, 44, 273–282.

  3. Phi-Van,L. and Stratling,W.H. (1988) EMBO J., 7, 655–664.

  4. Mirkovitch,J., Mirault,M.E. and Laemmli,U.K. (1984) Cell, 39, 223–232.

  5. Izaurralde,E., Mirkovitch,J. and Laemmli,U.K. (1988) J. Mol. Biol., 200, 111–125.

  6. Gasser,S.M. and Laemmli,U.K. (1986) Cell, 46, 521–530.

  7. Cockerill,P.N. and Garrard,W.T. (1986) FEBS Lett., 204, 5–7.

  8. Levy-Wilson,B. and Fortier,C. (1989) J. Biol. Chem., 264, 21196–21204.

  9. Dillon,N. and Grosveld,F. (1994) Curr. Opin. Genet. Dev., 4, 260–264.

  10. Namciu,S.J., Blochlinger,K.B. and Fournier,R.E.K. (1998) Mol. Cell. Biol., 18, 2382–2391.

  11. Bode,J., Stengert-Iber,M., Kay,V., Schlake,T. and Dietz-Pfeilstetter,A. (1996) Crit. Rev. Eukaryot. Gene Expr., 6, 115–138.

  12. Laemmli,U.K., Kas,E., Poljak,L. and Adachi,Y. (1992) Curr. Opin. Genet. Dev., 2, 275–285.

  13. Holmes-Davis,R. and Comai,L. (1998) Trends Plant Sci., 3, 91–97.

  14. Hart,C.M. and Laemmli,U.K. (1998) Curr. Opin. Genet. Dev., 8, 519–525.

  15. Strick,R. and Laemmli,U.K. (1995) Cell, 83, 1137–1148.

  16. Boulikas,T. (1995) Int. Rev. Cytol., 162A, 279–388.

  17. Spitzner,J.R. and Muller,M.T. (1988) Nucleic Acids Res., 16, 5533–5556.

  18. Kas,E. and Laemmli,U.K. (1992) EMBO J., 11, 705–716.

  19. Bode,J., Kohwi,Y., Dickinson,L., Joh,T., Klehr,D., Mielke,C. and Kohwi-Shigematsu,T. (1992) Science, 255, 195–197.

  20. Dickinson,L.A., Joh,T., Kohwi,Y. and Kohwi-Shigematsu,T. (1992) Cell, 70, 631–645.

  21. van Drunen,C.M., Oosterling,R.W., Keultjes,G.M., Weisbeek,P.J., van Driel,R. and Smeekens,S.C. (1997) Nucleic Acids Res., 25, 3904–3911.

  22. Potempa,J., Korzus,E. and Travis,J. (1994) J. Biol. Chem., 269, 15957–15960.

  23. Rollini,P. and Fournier,R.E.K. (1997) Mammalian Genome, 8, 913–916.

  24. Rollini,P. and Fournier,R.E.K. (1997) Genomics, 46, 409–415.

  25. Kelsey,G.D., Parkar,M. and Povey,S. (1988) Ann. Hum. Genet., 52, 151–160.

  26. Bao,J.J., Reed-Fourquet,L., Sifers,R.N., Kidd,V.J. and Woo,S.L. (1988) Genomics, 2, 165–173.

  27. Hofker,M.H., Nelen,M., Klasen,E.C., Nukiwa,T., Curiel,D., Crystal,R.G. and Frants,R.R. (1988) Biochem. Biophys. Res. Commun., 155, 634–642.

  28. Rollini,P. and Fournier,R.E.K. (1999) Genomics, 56, 22–30.

  29. Shapero,M.H., Langston,A.A. and Fournier,R.E.K. (1994) Somat. Cell Mol. Genet., 20, 215–231.

  30. Kelsey,G.D., Povey,S., Bygrave,A.E. and Lovell-Badge,R.H. (1987) Genes Dev., 1, 161–171.

  31. Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning: A Laboratory Manual, 2nd Edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

  32. Underhill,D.A. and Hammond,G.L. (1989) Mol. Endocrinol., 3, 1448–1454.