ABSTRACT
Using a novel procedure a representative human chromosome 19-specific library was constructed of short sequences, which bind
preferentially to the nuclear matrix (matrix attachment regions, or MARs). Judging by 20 clones sequenced so far, the library contains
>
50% of human inserts, about 90% of which are matrix-binding by the
in vitro
test. Computer analysis of sequences of eight human MARs did not reveal any significant homologies with the EMBL Nucleotide Data Base entries as well as
between MARs themselves. Eight MARs were assigned to individual positions on
the chromosome 19 physical map. The library constructed can serve as a good source
of MAR sequences for comparative analysis and classification and for further chromosome mapping of
MARs as well.
Libraries of short sequences with certain specific functions is an important
instrument for functional mapping of the human genome (
1
) considered to be the next step in the Human Genome Project once sufficiently
high resolution physical maps of human chromosomes are compiled (
2
-
6
). The most advanced area of functional mapping is apparently the mapping of
transcribed sequences, namely mRNA (
7
-
11
) and heterogeneous nuclear RNA (hnRNA) (
12
-
14
). This allows one to position protein coding and/or transcribed sequences on
the chromosomes and is finally aimed at the chromosome localization of all
estimated 50-100 thousand human genes (
15
). However, apart from the protein coding and transcribed sequences there exist
many kinds of other unique sequences also having specific functions, such as
promoters, enhancers, protein-recognizing sequences etc. Positioning of these sequences on the human
chromosomes could significantly expand the scope of the human genome functional
map.
The sequences preferably binding to the nuclear matrix or scaffold (matrix or
scaffold attachment regions, MARs or SARs) are an example of such functional
sequences. Being polymorphic and relatively evenly scattered over the human
genome these sequences can serve as chromosome markers (
16
). Such MARs (SARs) were previously characterized as relatively short (100-1000 bp) DNA sequences involved in anchoring chromatin loops (domains) to
the protein network termed nuclear matrix in interphase or chromosomal scaffold
in mitosis (for review see
16
-
20
). The general attributes assigned to MARs have been summarized (
16
). Among other elements, MARs often include potential origins of replication,
relatively long A[up arrow] + [up arrow]T-rich stretches harbouring topoisomerase II binding sites and
palindromic sequences. Some classes of MARs contain CT-rich stretches or, in some cases, are enriched in TG-motif. In addition, these sequences are rich in transcription
factors binding sites and contain potentially curved or kinked DNA (
16
). Functionally, MARs are bringing together the presumed active components of
the nuclear matrix (topoisomerase II and other enzymes of DNA and RNA
metabolism, transcription factors, etc.) and their target chromatin regions
which include, but are not limited to, MARs (
17
-
20
).
Here we report the construction of a human chromosome 19-specific library of DNA sequences preferably binding to the nuclear matrix
in vitro
as well as the early results of mapping some of these sequences on chromosome
19.
Growth and transformation of
Escherichia coli
cells, preparation of plasmids and [lambda] DNA, gel electrophoresis, blot-hybridization, and other standard manipulations were performed as
described (
21
,
22
).
Jurkat cells were grown in suspension using RPMI-1640 medium supplemented with 10% fetal calf serum. DMEM medium with 10%
fetal calf serum was used for other cell lines.
Nuclear matrices were prepared from Jurkat cells by a high salt extraction
method (
23
) or, alternatively, using a lithium 3,5-diiodosalicilate (LIS) extraction protocol (
24
,
25
). Binding of radioactively labelled MARs to the nuclear matrices
in vitro
was carried out as described (
25
).
A human chromosome 19 genomic library in a [lambda]-vector Charon 40 (provided by A.V. Carrano, Lawrence Livermore
National Laboratory, USA) with average insert of 16 kb was used as a source of
chromosome 19 DNA.
Total [lambda]-library DNA, 10 [mu]g, was divided into halves and digested to completion with 20
U of either
Sau
3A (New England Biolabs) or
Csp
6I (Fermentas, Lithuania) in 100 [mu]l of the corresponding manufacturer supplied buffer for 2 h at 37oC. The fragments obtained were extracted with phenol-chloroform, ethanol precipitated and dissolved in 30 [mu]l TE buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA).
To the resulting sticky ends, the following PCR adapters were ligated:
5'-GATCTGTTCATGG-3'
adapter
Sau
3A
3'-ACAAGTACCTAT
5'-TATGTTCATGG-3'
adapter
Csp
6I
The library primer was identical for both enzymes and contained an
Xho
I site (underlined).
For ligation, 10 [mu]l of each digested DNA was mixed with 1 nmol of the library primer and 1 nmol of the corresponding adapter in a final
volume of 20 [mu]l 50 mM Tris-HCl, pH 7.5, 5 mM MgCl
2
, 5 mM dithiotreitol, 0.5 mM ATP and 50 [mu]g/ml bovine serum albumin (BSA). The reaction was done with 8 U (Weiss)
ligase for 24 h at 13oC for
Sau
3A digest and 4oC for
Csp
6I digest. After the ligation, the mixtures were diluted 100-fold with TE buffer, and 1 [mu]l of each was PCR-amplified in 100 [mu]l 10 mM Tris-HCl, pH 8.3, 50 mM KCl, 3 mM MgCl
2
, 0.25 mM of each dNTP, 25 [mu]M of the library primer with 2 U
Taq
DNA polymerase. The amplification profile was 94oC for 0.8 min; 50oC for 0.6 min; 72oC for 1 min for 30 cycles. The resulting DNA products were
extracted with phenol-chloroform, precipitated with ethanol, washed, dried, dissolved in 50 [mu]l TE buffer and pooled together.
Binding of the DNA to nuclear matrices was done according to ref.
23
with modifications. A suspension of nuclear matrices from 5 * 10
7
cells was washed three times with 1 ml of buffer 1 (10 mM Tris-HCl pH 7.5, 50 mM NaCl, 2 mM EDTA, 0.25 M sucrose, 0.25 mg/ml BSA). Two
times concentrated buffer 1 (50 [mu]l), 20 [mu]l (40 [mu]g) of sonicated
E
.
coli
DNA and 20 [mu]l of water was next added to the matrix pellet and the mixture was
preincubated on a rocker platform for 3 h at room temperature. The fragmented
chromosome 19 library DNA (20 ng) with ligated primers was then added to the
preincubation mixture and the binding was performed for 30 min at room
temperature. Then the matrices were spun down for 1 min in a microcentrifuge
and washed once with buffer 2 (10 mM Tris-HCl pH 7.5, 300 mM NaCl, 1 mM MgCl
2
, 0.25 M sucrose, 0.25 mg/ml BSA), then with the same buffer but with 2 M NaCl,
and finally with the same buffer containing 50 mM NaCl. The matrices were then
resuspended in 50 [mu]l TE buffer, 0.5% SDS, proteinase K was added up to 50 [mu]g/ml and the mixture was incubated for 1 h at 37oC to digest proteins. The mixture with 10 [mu]l 10 M ammonium acetate added was extracted twice with phenol-chloroform and precipitated with 2 vol of ethanol
overnight at -20oC. Matrix-bound DNA was then collected in a microcentrifuge for 10 min
at 4oC, washed with 75% ethanol, dried and dissolved in 10 [mu]l TE buffer.
This DNA, 5 [mu]l, was again PCR-amplified and purified as described above and then used for the next
round of binding to the nuclear matrices under the same conditions. This
selection procedure was repeated five times.
The PCR-amplified mixture of the putative MARs after the fifth round of binding
with nuclear matrices was digested with
Xho
I, taking advantage of the introduced site, and cloned into
Xho
I site of a pGEM7zf(+) phagemid vector (Promega). After transformation the XL1
blue
E
.
coli
cells were plated on X-gal/IPTG agar plates and 384 white colonies were arrayed on 96-well clusters.
Single-stranded DNA of the pGEM7zf(+) phagemids containing MAR inserts was
prepared using R408 helper bacteriophage according to a Promega protocol (
26
). The sequencing reaction was done using the M13 direct sequencing primer, [[alpha]-
33
P]dATP (Obninsk, Russia) and Sequenase version 2.0 DNA sequencing kit (US
Biochemicals) according to the manufacturer recommendations.
PCR labelling of MAR probes for high density filter hybridization was performed
as described earlier (
14
). Probes for binding with nuclear matrices were prepared as follows. MAR-containing fragments were PCR-amplified in 50 [mu]l reaction volume using the corresponding plasmid template and
library primer. Negative control fragments were amplified from the
corresponding phage DNA using the following primers (positions in the
bacteriophage sequences are in parentheses).
[lambda]-phage (162 bp fragment)
Direct primer 5'-TCCGTGAGGTGAATGTGGTG-3' (916-935)
Reverse primer 5-'TAGTCGGCTCAACGTGGGTT-3' (1077-1058)
T7 phage (139 bp fragment)
Direct primer 5'-ACCGTGAAGGAACGTGACC-3' (34257-34275)
Reverse primer 5'-CGCCAGCAGCATTCATTAAG-3' (34395-34376)
A pUCMAR10 positive control was provided by I. Shatsky and PCR-amplified using the M13 direct and reverse 17 nt sequencing primers. The
fragments were purified by low melting point agarose gel electrophoresis, their
bands cut out from the gel, diluted 3-fold with TE buffer and used as templates for subsequent PCR labelling.
The PCR labelling was conducted in a final volume of 50 [mu]l containing ~10 ng of the template fragment, 10 [mu]M each primer (for controls) or 20 [mu]M library primer (for MAR fragments), 200 [mu]M each dCTP, dGTP and dTTP, 20 [mu]M dATP, 50 [mu]Ci [[alpha]-
32
P]dATP (Obninsk, Russia), 3 mM MgCl
2
, 10 mM Tris-HCl pH 8.3 and 50 mM KCl. The reaction profile was 94oC for 0.8 min; 50oC for 0.6 min; 72oC for 1 min for 20 cycles. Reaction products were
extracted twice with phenol-chloroform, precipitated with ethanol and dissolved in TE buffer.
DNA isolated from the nuclear matrices as described above was labelled using the
Ready-To-Go labelling mixture (Pharmacia) according to the manufacturer's
protocol.
The general scheme of the construction of a chromosome-specific MARs library is shown in Figure
1
.
To verify the absence of [lambda]-derived clones in the library the colonies were transferred onto
nitrocellulose filters and hybridized with the [lambda]-probe. Only seven of 384 colonies were [lambda]-positive, indicating that the extent of library
enrichment in MARs was >50 times, given an ~1:1 initial ratio of the vector and insert DNA in the mixture (data not
shown).
The presence and the size of inserts in all library clones were checked by PCR
with the library primer. Twenty clones with 250-500 bp inserts were selected and sequenced. The nucleotide sequences
obtained were compared to those of the EMBL Nucleotide Data Base using BLAST
algorithm (
27
). No significant (>80%) homologies with the data bank entries were found. The
sequences were analysed on mutual homology using PC/Gene package. Only two
clones (M0A3 and M0A12) were >95% homologous. As 19 out of 20 clones were
individual, it can be concluded that the library obtained is highly
representative. Nevertheless, it could not be excluded that very long or short MARs are
underrepresented in the library due to the PCR reaction bias.
The source [lambda]-library of human chromosome 19-specific DNA was constructed by sorting chromosomes isolated
from human-hamster hybrid UV5HL9-5B cells containing chromosome 19 as a sole human component. Due to
limitations of the sorting, a considerable number of the library clones were of
hamster origin. It was necessary, therefore, to check the human chromosome 19
origin of the clones before mapping. To this end internal pairs of primers
specific for each clone were designed using Primer software (version 0.5,
Whitehead Institute for Biomedical Research) and synthesized. Sequences of the
primers are presented in Table
1
. These primer pairs were subsequently used in PCR with DNA isolated from human,
hamster and hybrid UV5HL9-5B cells (
14
). The detection of a PCR product of the expected length for a clone with only
human and hybrid templates, but not with hamster ones was considered to be a
proof that the clone did originate from human chromosome 19 (Fig.
3
). Of 20 sequenced clones, 10 were found to be human chromosome 19-specific, eight were of hamster origin and two were human, but belonging
to some other human chromosome. Only human chromosome 19-specific clones were used for subsequent analysis.
Finally, we mapped the characterized MAR sequences on human chromosome 19 (see
Table
1
). The strategy of mapping was described in detail elsewhere (
14
). The F and R cosmid libraries of chromosome 19, arrayed on high density
filters (
5
,
6
), were constructed at Lawrence Livermore National Laboratory (USA) and provided
by A.V. Carrano. They were used for hybridization with the PCR-labelled MAR probes. To verify the hybridization results the hybridization-positive cosmids were used as templates for PCR with clone-specific internal primer pairs. The detection of a product of
the expected length was considered to be a proof that the sequence of the
corresponding MAR is located in a given cosmid (data not shown).
MAR positive cosmids were positioned on the chromosome 19 physical map by virtue
of their membership in cosmid contigs generated by restriction fingerprinting,
and subsequent mapping of these contigs using fluorescence
in situ
hybridization (FISH). A description of the physical map is available (
5
). Subsequent versions of this map can be found using the following URL:
http://www-bio.llnl.gov/bbrp/genome/genome.html.
The nuclear matrix attachment regions (MARs) are structurally and functionally
important elements of eukaryote genome (
16
-
20
). Being polymorphic and evenly scattered over the human genome (
16
), these sequences can serve not only as markers for physical mapping, but also
provide an opportunity for functional characterization of the human genome,
such as location of sites of initiation of DNA replication and transcription.
For the functional mapping chromosome specific representative libraries of
sequences serving particular functions (e.g. binding to the nuclear matrix) are
superior to non-specific ones (discussed in detail in ref.
14
). Such a library in our case is a source of hundreds of chromosome specific
MARs also suitable for comparative analysis which might help to reveal the
nature of the nuclear matrix binding sites and their interaction with MARs.
Assuming that the length of chromosome 19 DNA is ~6 * 10
7
bp and the average size of chromatin loops is 60 kb (
18
), the total number of the chromosome DNA fragments attached to the nuclear
matrix in interphase should be ~1000. The number of different sequences, potentially capable of binding the
nuclear matrix, however, can be significantly larger as a large proportion of
the potential sites can be detached from the matrix during the cell cycle (
29
). Despite this ambiguity, it seems reasonable to assume that the number of MARs
in the chromosome is around several thousands. Consequently, to be a good
source of MAR sequences for mapping, a MARs library should contain at least
several hundred independent clones.
Basically there are two possible approaches for MAR library construction: either
to clone directly the residual matrix DNA or to clone the fragments capable of
binding nuclear matrices
in vitro
. The first one has an obvious advantage in that all clones obtained are by
definition a part of the nuclear matrix. This approach was employed recently by
Boulikas and Kong (
30
). However, 90 of 150 clones of the resulting library contained the same insert.
Another serious limitation of this kind of libraries if used for mapping is
that they cannot be constructed in a chromosome specific manner, and therefore
imply the laborious and time-consuming chromosome assignment of each clone.
We used the second approach which permitted us to obtain both a chromosome-specific and highly representative library. Of 20 sequenced inserts 19
were independent, 10 were human chromosome 19-specific and nine of them were bound specifically to the nuclear matrix
in vitro
. This indicates that our arrayed library potentially contains about 150 MARs
specific for human chromosome 19, which is about five times more than the total
number of eukaryotic MARs characterized so far (
20
,
31
). It should be noted that this number can be readily increased by arraying
additional clones. Moreover, blot-hybridization (Fig.
5
) showed that all human MARs from this library had close homologs among residual
matrix DNA fragments. The only obvious disadvantage of this library is that it
contains a significant (~50%) proportion of clones of hamster origin due to a high content of
hamster DNA in the source [lambda]-library.
Although the number of MARs mapped on chromosome 19 is not yet sufficient to
draw general conclusions, the first mapping attempt indicated that MARs were
evenly distributed on both chromosome 19 arms (Table
1
). These findings are in accord with generally accepted loop models of
interphase chromatin or metaphase chromosome organization (
17
-
20
). All eight MARs were mapped to one or several cosmids belonging to the same
contig. Seven of these contigs were assigned to specific locations on the
chromosome using high resolution FISH to localize one or more selected cosmids
from each contig (
32
). The position of the contig containing M0B7 could be determined only with an
accuracy of up to a single chromosome band.
Within the limits of the accuracy, only one position on the chromosome could be
assigned to each individual MAR clone. However, we cannot exclude that the same
or closely related MARs can be found on other human chromosomes or within one
and the same cosmid. Moreover, MARs M0A3, M0A7, M0B1, M0B6 and M0B9 were also
mapped to additional cosmids, which are either not assigned at all (`orphan'
cosmids) or assigned to contigs with yet unknown locations. Even with the
limitations above we can conclude that all characterized MARs are individual
and do not belong to any family of repeated sequences. This finding is in
contrast to the results obtained for the hncDNA clones where more than one-third of the mapped sequences belonged to various repeated families (
14
).
Using the library constructed we also compared some properties of the chromosome
19 MARs (summarized in Table
2
). The only common characteristic of all (except one) MARs was that they were
relatively enriched in purines (or pyrimidines in complementary chains).
However, it should be kept in mind that the actual matrix-binding sequences are most likely only a part of the MAR inserts.
Therefore integral characteristics like A + G content can be deceptive.
To characterize the MAR elements in more detail, we analysed their sequences
using the PC/Gene package. Four out of eight MARs contained the A + T-rich regions (>75% A + T), which is characteristic of one of the major
classes of these sequences (
16
,
19
,
20
). Nevertheless, four other MARs with high affinity for the nuclear matrix did
not contain any A + T-rich sequences longer than 20 bp. Similarly, the number of inverted
repeats (potential hairpin or cruciform structures) did not correlate with the
matrix affinity: for example, clones M0B2 and M0B7 with high affinity contained
many less inverted repeats than M0B6, a clone with the lowest affinity (Table
2
). On the other hand, a clone M0B1 with the highest affinity behaves like a
typical A + T-rich MAR with many inverted repeats. Significant sequence differences
between MARs in combination with their ability to compete with each other for
binding with the matrix (
23
,
25
) make a serious challenge to researchers trying to find out the structural
basis for this kind of specific interactions. The sequence variability can be
caused, for example, by multiple overlapping DNA-protein interaction sites each with its own sequence requirements, which
can, in addition, be degenerated. The experiments aimed at location of the DNA-protein contact sites within the MAR sequences will hopefully clarify the
nature of this specificity.
The authors are grateful to V. K. Potapov, Y. B. Lebedev, S. V. Volik and Y. O. Shevchenko for participating in some experiments, to A. A.
Volodin for sequencing of some MARs, to B. O. Glotov for helpful discussion and critical reading of the manuscript and to I. N.
Shatsky for providing us with the pUCMAR10 plasmid. We are grateful also to A.
V. Carrano, E. Branscomb, J. Mazrimas and other colleagues from Lawrence
Livermore National Laboratory for indispensable help in mapping of sequences on
chromosome 19. Work in Russia was supported by grant of Human Genome State
Project of Russia and International Science Foundation grant #MGT000. Work at
LLNL was performed under the auspices of US Department of Energy by LLNL under
Contract # W-7405-ENG-48.
REFERENCES
Return