© 2000 Macmillan Magazines Ltd
Mitotic stability of an episomal vector containing
a human scaffold/matrix-attached region is
provided by association with nuclear matrix
NATURE CELL BIOLOGY | VOL 2 | MARCH 2000 | cellbio.nature.com
A. Baiker*, C. Maercker*, C. Piechaczek*, S. B. A. Schmidt*, J. Bode†, C. Benham‡ and H. J. Lipps*§
*Institut für Zellbiologie, Universität Witten/Herdecke, Stockumer Strasse 10, D-58448 Witten, Germany
†Gesellschaft für Biotechnologische Forschung mbH, Mascheroder Weg 1, D-38124 Braunschweig, Germany
‡Department of Biomathematical Sciences, Mount Sinai School of Medicine, New York, New York 10029, USA
NA replication occurs in tight association with the nuclear
matrix, where binding of the replication origin to the nuclear
matrix must precede the onset of S phase1–4. We have shown
previously that the origin of replication of the simian virus 40
(SV40) genome linked to a human scaffold/matrix-attached region
(S/MAR) allows sustained episomal replication (where an episome
is autonomous, self-replicating DNA) that is independent of the
expression of the virally encoded large T-antigen5. A vector with
this combination of SV40 origin and potential for matrix associa-
tion is maintained in cultured cells for at least 100 cell generations,
in the absence of selection5. Here we show, by in situ hybridization
and nuclear-fractionation procedures, that there is a specific inter-
action of this vector with the nuclear matrix and the chromosome
scaffold, presumably through proteins that both structures have in
common. This interaction correlates with replication of the vector
as an episome. These observations allow a mechanistic explanation
for the episomal replication and mitotic stability of this new type of
Chinese hamster ovary (CHO) cells, transfected either with the
S/MAR–origin-containing vector (pEPI-1, Fig. 1a) or the corre-
sponding truncated vector lacking the S/MAR (pGFP-C1, Fig. 1d)
were subjected to fluorescence in situ hybridization (FISH) on met-
aphase spreads. Confirming previous results5, FISH analyses
showed that the truncated vector was prone to integrate into the
host-cell DNA. A single intense signal was seen at the same chromo-
somal locus of different cells derived from the same clone (Fig. 1e,
f); however, the site of integration varied between different clones.
In contrast to this predictable behaviour of integrated transgenes6,
clones established using the complete vector contained cells with a
greater and variable number of fluorescent spots that were associ-
Figure 1 In situ hybridization of S/MAR–origin-containing vector (pEPI-1) and its S/
MAR-depleted version (pGFP-C1) to CHO metaphase chromosomes. a, Restriction
map of pEPI-1 (ref. 5). Only the functional elements of the vector are indicated.
Arrows indicate the positions of primers used for PCR analysis. b, c, In situ
hybridization to a CHO clone transfected with pEPI-1. b, Spreading of metaphase
chromosomes was done as gently as possible. c, During spreading, increased shear
forces were applied as described in Methods. d, Map of pGFP-C1 (ref. 5). e, f, In situ
hybridization to a CHO clone transfected with pGFP-C1. Spreading of metaphase
chromosomes was done in e and f as in b and c respectively. In situ hybridization
was done according to ref. 17 using biotin-labelled pGFP-C1 as a probe. Scale bar
represents 5 µm. kb, kilobase.
© 2000 Macmillan Magazines Ltd
NATURE CELL BIOLOGY | VOL 2 | MARCH 2000 | cellbio.nature.com
ated, but not strictly coincident, with the chromosomes (Fig. 1b, c).
By analysing metaphase spreads of 10 different clones, we estimated
the number of vector molecules per cell to be between 4 and 13
(average copy number between 5 and 9; standard deviations
between 0.83 and 3.1) (see Supplementary information). These fig-
ures did not change significantly between 20 and 50 generations
after transfection in the absence of selection. The position of these
signals varied from cell to cell and the non-covalent association of
the vector and host chromosomes was evident, as hybridization was
seen always on only one of the sister chromatids in metaphase
spreads. The non-covalent nature of this association was further
shown by applying increased shear forces during spreading. Under
gentle lysis conditions more than 95% of the signals were associated
with the chromosome (Fig. 1b); however, when cells were dropped
onto the slides from a distance exceeding 10 cm, a significant
number of vector molecules appeared to be separated from the
chromosome structures (Fig. 1c).
To confirm the association of the S/MAR-containing vector
with the nuclear matrix, we used nuclear-fractionation procedures7.
We separated the various protein fractions by SDS–polyacrylamide
gel electrophoresis (SDS–PAGE) and visualized the DNA-binding
constituents by southwestern analysis. In the presence of a 1,000-
fold excess of nonspecific DNA, both vectors showed comparable
binding to a protein in fraction 2 (0.25 M ammonium sulphate),
which contains the soluble nuclear components including histone
H1. Subsequent digestion with DNaseI removed DNA and the
remaining histones (fraction 3), leaving the complete matrix in the
pellet fraction. After extraction with 2 M NaCl, matrix proteins not
part of the core filament network will be in the supernatant (frac-
tion 4). The core filament network was then solubilized with SDS-
sample buffer to yield fraction 5. In the matrix-containing fractions,
a major protein band bound strongly to the S/MAR-containing vec-
tor but not to its truncated form (Fig. 2a, b). However, although
migration of the S/MAR-binding protein and the protein in frac-
tion 2 indicates that they might belong to the histone H1 group, this
correspondence could not be confirmed by using several available
antibodies against common H1 subspecies that, in a control,
showed crossreaction with endogenous CHO-cell histone H1.
To prevent chromosomal rearrangements from occurring dur-
ing fractionation, we added a crosslinking step to the protocol. Cis-
DDP is a reagent that links matrix proteins to endogenous S/MAR
with high specifity8. Cis-DDP can be applied to the living cell and
activated by reducing the concentration of chloride ions. In these
experiments, we replaced digestion with DNaseI during the frac-
tionation procedure by digestion with six restriction enzymes that
do not cut pEPI-1, together with either one restriction enzyme that
linearizes the vector (EcoRI or BglII, Fig. 1a) or two enzymes that
excise the S/MAR element (EcoRI and BglII, Fig. 1a). After nuclear
fractionation, the crosslinked complexes were treated with protein-
ase K and the purified DNA was amplified by the polymerase chain
reaction (PCR) to trace the complete vector and its S/MAR-
depleted variant. Vectors without the S/MAR were found in frac-
tion 3 and, to a lesser extent, in fraction 4 (Fig. 2c, PCRII) whereas
S/MAR-containing vectors were found in the nuclear-matrix frac-
tion 5 and to a lesser extent in fraction 4 (Fig. 2c, PCRI).
Further control experiments support our view that the combi-
nation of an SV40 replication origin and a scaffold/matrix-attached
region enables replication in the episomal state and that this func-
tional state correlates with a specific interaction with components
of the nuclear matrix. Plasmids in which either the S/MAR (Fig. 1d–
f) or the replication origin (data not shown) was deleted lost this
property and became integrated. In addition, other control
sequences cloned in place of the S/MAR element did not support
the episomal state. The AT-rich NTS-1 and NTS-2 sequences are
associated with an endogenous origin of replication in mouse cells9.
We cloned these sequences in place of the S/MAR element either as
monomers or as multimers (up to a tetramer). In no case was epi-
somal replication of these vectors observed. While the 69% AT-rich
S/MAR element10 used in this study has a normalized binding affin-
ity of 94 ± 5% to the nuclear matrix11 and an extended base-unpair-
ing region over its entire length12, the amplification-promoting
sequences9 NTS-1 (370 base pairs, 56% A+T, binding affinity
<20%) and NTS-2 (424 base pairs, 65% A+T, binding affinity 50 ±
10%) contain only one (NTS-1) or two (NTS-2) restricted unpair-
ing elements, which are not sufficient to confer S/MAR character12.
The association of the vector molecules with the host chromo-
somes, as shown by in situ hybridization, is reminiscent of the behav-
iour of episomally replicating viruses such as Epstein–Barr virus,
Kapos’s-sarcoma-associated herpesvirus or bovine papilloma virus,
in which maintenance of the viral genome is mediated by virally
encoded proteins (EBNA-1, LANA and E2, respectively) that ensure
efficient segregation of the replicated genome during mitosis13–15. In
the case of our vector it may be that association with elements of the
Figure 2 Nuclear fractionation and identification of the S/MAR-containing vector
by southwestern and PCR analysis. a, b, Southwestern analysis of nuclear
proteins fractionated according to ref. 7. The proteins were separated by SDS-gel
electrophoresis, blotted and probed either with DIG-labelled pGFP-C1 (a) or with
pEPI-1 (b). Fraction 1 was extracted with cytoskeletal buffer including 100 mM
NaCl; fraction 2 was extracted with extraction buffer containing 250 mM
ammonium sulphate; fraction 3 was extracted with digestion buffer after digestion
with DNaseI; fraction 4 is the supernatant obtained after adjusting to 2 M NaCl;
fraction 5 is the pellet fraction resulting after extraction with 2 M NaCl. Mr (K),
relative molecular mass (in thousands). c, Distribution of vector DNA in nuclear
fractions 1–5, after crosslinking with cis-DDP8, of cells transfected with pEPI-1.
For this experiment, cellular DNA was not digested by DNaseI but by restriction
digestion using six non-cutting restriction enzymes together with either a
restriction enzyme that linearizes the vector (the S/MAR is retained with the
portion of the vector amplified by PCR (PCRI)) or two restriction enzymes (the S/
MAR is released from the portion of the vector amplified by PCR (PCRII)). The same
volumes of each fraction were used for PCR analysis; primers were derived from
the neomycin-resistance genes indicated in Fig. 1a; the DNA concentrations in the
corresponding fractions were identical in PCRI and PCRII. K+ and K–, control PCRs
using either pEPI-1 as template (K+) or no template (K–); M, 1-kb ladder (Gibco).
PCR conditions were as described20.
© 2000 Macmillan Magazines Ltd Download full-text
NATURE CELL BIOLOGY | VOL 2 | MARCH 2000 | cellbio.nature.com
nuclear matrix is also essential for segregation during mitosis; fur-
thermore, such an association must be vital in allowing the plasmid
to replicate as an episome. In SV40 the expression of the large T-anti-
gen is essential for SV40 replication: it unwinds the SV40 origin and
allows the replication complex to bind16. In the S/MAR-containing
vector the S/MAR might complement some of the functions of the
large T-antigen and, as it is bound to the nuclear matrix, the SV40
origin could be co-replicated during cellular DNA replication using
the replication machinery that forms on the chromosome. To our
knowledge, this is the first report that the persistence of an episomal
vector that is capable of sustained, selection-independent replication
in the absence of virally encoded trans-acting factors correlates with
its association with the nuclear matrix.
Culturing of CHO cells, transfection of CHO cells and selection of transfected clones were done as
FISH analysis was done according to ref. 17 using biotin-labelled pGFP-C1 (Fig. 1d) as a probe.
Spreading of metaphase chromosomes was done as gently as possible (Fig. 1b), or increased shear forces
were applied by dropping the cells onto the slides from at least a 10-cm distance (Fig. 1c). The labelled
probe was detected by an avidin/tetramethyl rhodamine isothiocyanate (TRITC)-labelled anti-avidin
antibody/biotin (Sigma) sandwich procedure17. After immunostaining, metaphase chromosomes were
counterstained with 4,6-diamidinophenylindole (DAPI; 0.1 µg ml–1; Sigma). Fluorescence microscopy
was done using a Leitz DM RB microscope and photographed using Kodak Elite Chrome 100 ASA films.
Nuclear proteins were fractionated according to ref. 7, separated on a 12.5% SDS gel18 and blotted
onto a nylon membrane (Amersham); a southwestern analysis19 was performed using either digoxygenin
(DIG)-labelled pGFP-C1 or pEPI-1 as a probe. Before the incubation with the labelled probes, filters were
saturated with a 500–1,000-fold excess of sonicated Escherichia coli DNA.
To trace the vector DNA in the different nuclear fractions by PCR analysis, we included a cis-DDP
crosslinking step8. The digestion with DNaseI during the fractionation process was replaced in these
experiments by digestion with restriction enzymes, using six restriction enzymes that do not cut pEPI-1
and, in addition, either a restriction enzyme (EcoRI or BglII) that linearizes the vector (Fig. 2c, PCRI) or
two restriction enzymes (EcoRI and BglII) that delete the S/MAR from the vector (Fig. 2c, PCRII)5. PCR
analysis was done according to ref. 20. The same volumes of each fraction were used for PCR analysis and
the DNA concentrations in the corresponding fractions were identical in PCRI and PCRII. Amplification
was done using primers derived from the neomycin-resistance gene (neo 1: 5′-
GGAGAGGCTATTCGGCTATGAC; neo 2: 5′-CGTCAAGAAGGCGATAGAAGGC). Primer positions
are indicated in Fig. 1a.
RECEIVED 4 NOVEMBER 1999; REVISED 10 JANUARY 2000; ACCEPTED 27 JANUARY 2000;
PUBLISHED 10 FEBRUARY 2000.
1. Dijkwell, P. A. & Hamlin, J. L. Int. Rev. Cyt. 162A, 455–484 (1995).
2. Cook, P. Science 284, 1790–1795 (1999).
3. Cossons, N. et al. J. Cell. Biochem. 67, 439–450 (1997).
4. Bode, J., Benham, C., Knopp, A. & Mielke, C. Crit. Rev. Eukar. Gene Expr. (in the press).
5. Piechaczek, C., Fetzer, C., Baiker, A., Bode, J. & Lipps, H. J. Nucleic Acids Res. 27, 426–428 (1999).
6. Robins, D. M., Ripley, S., Henderson, A. S. & Axel, R. Cell 23, 29–39 (1981).
7. Nickerson, J. A., Krockmalnic, G. & Penman, S. in Cell Biology, a Laboratory Handbook (ed. Celis, J.
E.) 184–192 (Academic, London, 1998).
8. Ferraro, A., Cervoni, L., Eufemi, M., Altieri, F. & Turano, C. J. Cell. Biochem. 62, 495–595 (1996).
9. Wegner, M. et al. Nucleic Acids Res. 17, 9909–9931 (1989).
10. Mielke, C., Kohwi, Y., Kohwi-Shigematsu, T. & Bode, J. Biochemistry 29, 7475–7485
11. Kay, V. & Bode, J. Methods Mol. Cell Biol. 5, 186–194 (1995).
12. Benham, C., Kohwi-Shigematsu, T. & Bode, J. J. Mol. Biol. 274, 181–196 (1997).
13. Calos, M. P. Proc. Natl Acad. Sci. USA 95, 4084–4085 (1998).
14. Lehman, C. W. & Botchan, M. R. Proc. Natl Acad. Sci. USA 95, 4338–4343 (1998).
15. Ballestas, M. E., Chatis, P. A. & Kaye, K. M. Science 284, 641–644 (1999).
16. Fanning, E. & Knippers, R. Annu. Rev. Biochem. 61, 55–85 (1992).
17. Liu, C. G., Maercker, C., Castanon, M. J., Hauptmann, R. & Wiche, G. Proc. Natl Acad. Sci. USA 93,
18. Laemmli, U. K. Nature 227, 680–685 (1970).
19. Miskimis, W. K., Roberts, M. P., McClelland, A. & Ruddle, F. H. Proc. Natl Acad. Sci. USA 82, 6791–
20. Saiki, R. K. et al. Science 239, 487–491 (1988).
This work was supported by the Alfried Krupp von Bohlen und Halbach foundation (H.J.L.), by the EC
(grant BIO4-CT98-0203) and DFG (grant Bo 419/5-3) (J.B.) and by the NSF (grant DBI99-04549) and
NIH (grant ROI-GM47012) (C.B.). We thank F. Grummt for providing the NTS-1 and NTS-2 sequences
and C. Fetzer for his help with the graphs.
Correspondence and requests for material should be addressed to H.J.L.
Supplementary information is available on Nature Cell Biology’s World-Wide Web site (http://
cellbio.nature.com) or as paper copy from the London editorial office of Nature Cell Biology.