ABSTRACT
We describe the ability of novel episomally maintained vectors to efficiently
promote gene expression in embryonic stem (ES) cells as well as in established
mouse cell lines. Extrachromosomal maintenance of our vectors is based on the
presence of polyoma virus DNA sequences, including the origin of replication
harboring a mutant enhancer (PyF101), and a modified version of the polyoma
early region (LT20) encoding the large T antigen only. Reporter gene expression
from such extrachromosomally replicating vectors was approximately 10-fold higher than expression from replication-incompetent control plasmids. After transfection of different ES cell lines, the polyoma virus-derived plasmid variant pMGD20neo (7.2 kb) was maintained
episomally in 16% of the G418-resistant clones. No chromosomal integration of pMGD20neo vector DNA was
detected in ES cells that contained episomal vector DNA even after long term
passage. The vector's replication ability was not altered after insertion of up
to 10 kb
hprt
gene fragments. Besides undifferentiated ES cells, the polyoma-based vectors were also maintained extrachromosomally in differentiating
ES cells and embryoid bodies as well as in established mouse cell lines.
Establishment of the mouse embryonic stem (ES) cell system has opened new ways
to study gene function in a living organism. ES cells are isolated from the
inner cell mass of preimplantation embryos and retain their pluripotency when
cultured under suitable conditions (
1
). Microinjection of ES cells into a recipient blastocyst and reimplantation
into pseudopregnant mothers results in the formation of a chimeric embryo which
can transmit the ES cell genotype to the next generation (
2
). Thus, specific genetic changes of the ES cell genotype generated by homologous recombination
in vitro
can be transferred into living mice (reviewed in ref.
3
). On the other hand, ES cells can be induced to differentiate into embryo-like structures known as embryoid bodies (EBs) which resemble the 6-8 day egg cylinder stage observed during normal mouse embryogenesis (
4
). Since embryoid bodies recapitulate several aspects of early mouse development, they have been
proposed as an
in vitro
model of embryogenesis (
4
-
8
).
To explore the feasibility of maintaining exogenous DNA sequences on episomal
plasmids in ES cells, we have constructed the polyoma-based vector pMGD20neo that can replicate extrachromosomally in these
cells (
9
). The polyoma virus, one of the smallest oncogenic viruses known, is a DNA
virus existing in certain laboratory mouse colonies and in some wild mice as a silent infection (reviewed in refs
10
,
11
). The viral genome consists of a double-stranded, circular DNA of ~5.3 kb. The polyoma virus early region, which is transcribed early during
the viral life cycle, is one of the rare examples in eukaryotic cells in which
all three potential reading frames are used to encode three different proteins.
Alternative splicing of the common precursor transcript results in three mRNAs
encoding for: (i) large tumor (T) antigen, a 100 kDa nuclear protein essential
for initiating viral DNA replication; (ii) middle T, a 48 kDa protein involved
in cellular transformation (
12
); and (iii) small T, a 22 kDa protein of ill-defined function(s). The vector pMGD20neo (Fig.
3
A) contains a modified segment from the polyoma early region that can only express large T (LT20), thereby avoiding expression of the oncogenic middle T. This vector also contains the mutated PyF101 enhancer-ori segment (
13
) that permits efficient expression of large T and DNA replication in the early embryo as well as in ES cells. We
recently showed that pMGD20neo is maintained episomally at about 10-30 copies per cell for at least 74 ES cell generations in the presence of
G418 (
9
). In the present work, we analyzed the ability of polyoma-based vectors to enhance gene expression, to support homologous
recombination in ES cells, and to replicate in established mouse cell lines and
differentiating ES cells.
pUC/LT20 and derivatives.
The wild type polyoma enhancer was replaced by the 802 bp
Bgl
I-
Bam
HI fragment from plasmid pPyF101 (
13
) harboring the mutant enhancer. To restrict the expression of the early region,
the
Ava
I site at bp 660 (numbering according to ref.
14
) was cleaved, blunted and ligated to the
Hae
III (bp 782) site. This deletion removed the 5' splice donor site at bp 748 for both middle T and small T reading
frames. Subsequently, the
Kas
I-
Hin
cII fragment (2757 bp) of the polyoma early region including this deletion was
subcloned together with a
Bam
HI-
Kas
I segment composed of a 130 bp DNA sequence containing the SV40 early region
polyA site in addition to the polyoma origin of replication and PyF101 enhancer
into pUC19 resulting in pUC/LT20.
pMGD20neo
. A
Bgl
II linker was introduced into the
Dra
II site of pMC1neopolyA (
15
) and followed by the substitution of the
Eco
RI-
Bss
HII (820 bp) fragment of the resulting pMC1neopoly- A
Bgl
II+ with the 1047 bp segment (
Eco
RI-
Bss
HII) from PGKneobpA (
16
) generating PGKneopolyA. Subsequently, this vector was cleaved with
Bgl
II and
Ssp
I and the
Bam
HI-
Hin
cII fragment from pUC/LT20 was inserted, giving rise to pMGD20neo (see Fig.
3
A). Note that the
neo
gene used in our preparations does not contain the point mutation in nucleotide 2096 which has been shown to reduce the resistance of transformants to G418
selection (
17
). Following the same cloning steps, we generated a variant of pMGD20neo in
which the LT20 fragment was replaced by the cDNA (pPyLT1) version of large T (
18
).
pMC1neo-hGH and derivatives
. The SVtk hybrid promoter (
Pvu
II-
Bgl
II fragment) used to drive
neo
expression in pSVtkneo[beta] (
19
) was linked to the hGH reporter gene fragment
Bam
HI-
Ssp
I (
20
), and the resulting segment was subcloned into the
Nde
I-
Bam
HI site of pMC1neopolyA
Bgl
II+ (see above) thereby replacing the
neo
cassette. Insertion of the MC1neopolyA cassette into the
Bgl
II site resulted in pMC1neo-hGH while coinsertion with a
Bam
HI-
Hin
cII fragment containing either the PyF101/LT20 polyoma sequence (from pUC/LT20)
or the corresponding wild type (PyF101/wt) or cDNA (PyF101/LT1) viral sequences
resulted in plasmid pMC1neo-hGH-PyF101/LT20 and its corresponding derivatives (wt and LT1).
pRVi6.8-LT20
.
The RVi6.8
hprt
fragment-which contains the
neo
expression cassette in the
The ES cell line CCE (
23
) was grown on gelatine-coated dishes without feeder cells in freshly prepared Dulbecco's modified Eagle's
medium (DMEM; Gibco) supplemented with 20% heat-inactivated (56oC, 30 min) foetal calf serum (FCS; Boehringer-Mannheim), 1000 U/ml leukemia inhibitory factor (LIF; Gibco),
150 [mu]M monothioglycerol (Sigma), 1* minimal essential medium (MEM) non-essential amino acids (Gibco), 100 U/ml penicillin (Gibco) and
0.1 mg/ml streptomycin (Gibco) in a humidified environment containing 5% CO
2
at 37oC and passaged every 2-3 days as described (
7
,
8
). Routinely, 10
7
ES cells in 800 [mu]l PBS were electroporated with 10-20 [mu]g DNA using a BioRad Gene Pulser at 240 V/960 [mu]F. Selection for
neo
was started 24-40 h later using G418 (500 [mu]g/ml active substance). Selection against a functional
hprt
gene was started exactly 6 days after transfection by adding 6-thioguanine to the medium to a final concentration of 1 [mu]g/ml. Differentiation of ES cells to EBs was performed in a semisolid
medium exactly as described (
7
,
8
).
Mouse embryonic carcinoma cells (F9), mouse renal adenocarcinoma cells (RAG) and mouse L-fibroblasts (L-929) were cultured in freshly prepared DMEM supplemented with 10% FCS, 150 [mu]M monothioglycerol, 1* MEM non-essential amino acids, 1 mM sodiumpyruvate (Gibco), 2 mM L-glutamin (Gibco) and antibiotics. Mouse erythroleukemia (MEL) cells were grown
in the same medium but containing 15% FCS.
Quantitation of hGH protein in the supernatant of transfected ES cells was performed using the immunoradiometric assay TANDEM-R HGH (Hybritech, San Diego, CA) following the manufacturer's instructions exactly. Low-molecular-weight DNA was extracted according to a modified Hirt protocol as
described (
9
). Total DNA extraction and Southern blotting was performed by standard
methodology. For Western blot analysis, cells were lysed with RIPA buffer [150
mM NaCl, 50 mM Tris-HCl (pH 7.2), 0.5% Nonidet P-40, 1% Triton X-100, 0.1% SDS, and 1% sodium deoxycholic acid] and spun 20
min at 50 000 r.p.m. in a Beckman TL100 ultracentrifuge. The extracts were
loaded onto a 7.5% polyacrylamide gel and subsequently electroblotted onto a Nytran filter (Schleicher & Schuell). Polyclonal rat antibodies against polyoma T antigens were obtained
from W. Eckhart (San Diego, CA). Chemiluminescent detection was performed using an anti-rat IgG antibody conjugated to horseradish peroxidase.
Since the transforming activity of polyoma middle T might alter the pluripotency
of the ES cells, we sought to express large T only (required for viral DNA
replication) from constructs harboring either the large T cDNA derived from pPyLT1 (
18
) or our polyoma LT20 version (
9
) which contains a mutated intervening sequence lacking the 5' splice donor site for middle T and small T reading frames (see Materials
and Methods). Both large T-encoding DNAs were fused to a mutant version of the polyoma enhancer-ori segment termed PyF101 (
13
). Subsequently, these polyoma early region variants were subcloned into a
modified version of pMC1neopolyA (
15
) containing the human growth hormone (hGH) reporter gene (pMC1neo-hGH). After electroporation and G418 selection for 17 days, the cellular
extracts of pooled ES cell colonies were tested for expression of large T by
Western blot analysis using a polyclonal antibody recognizing all three T
antigens (Fig.
1
). As positive control we used the T antigens-producing cell line MOP6 (
24
), and for negative control, untransfected NIH3T3 and ES cells as well as ES
cells transfected with plasmids which do not contain sequences encoding polyoma
large T (-/- and PyF101/-). ES cells transfected with vector pMC1neo-hGH-PyF101/LT1 containing the large T cDNA produced an unspecific banding pattern which
was indistinguishable from the negative controls (Fig.
1
). In contrast, transfection with an analogous plasmid harboring the LT20 version (pMC1neo-hGH-PyF101/LT20) resulted in efficient production of the viral
replication protein, indicating that an intervening sequence in the polyoma early region is necessary for efficient expression of large T.
We were interested in determining the ability of the polyoma sequence present in
our constructs to increase expression of other genes located on the same
plasmid. Therefore, we measured the expression levels of the reporter gene hGH
after transfection of ES cells. Since hGH protein is secreted by the cells (
20
), kinetic analysis of hGH expression was performed from periodically collected
samples of the culture medium. Forty hours after transfection with circular
plasmid DNA, the media was replaced and G418 selection was started. Every 1-2 days, the hGH concentrations were determined and the G418-containing medium was replaced. Figure
2
illustrates a typical transfection experiment: ES cells electroporated with the basic vector pMC1neo-hGH (-/-) only, or with the same vector containing the polyoma large
T cDNA sequence (PyF101/LT1) showed about the same hGH concentrations. This observation was consistent with the immunoblot presented in Figure
1
where no polyoma large T was detected when expressed from a cDNA construct. In
contrast, ES cells transfected with pMC1neo-hGH harboring either the polyoma wild type early region segment
(PyF101/wt) or our modified large T version (PyF101/LT20) increased their hGH
expression by a factor of 3-5 and 6-10, respectively.
We recently reported that 15% of the 87 G418-resistant CCE ES cell clones have been found to maintain pMGD20neo
episomally (
9
). Further transfection experiments (Table
1
) using this CCE ES cell line (derived from the mouse strain 129/Sv) as well as
the 129/OLA-derived ES cell line E14 and its
hprt
-deficient subclone E14TG2a (
27
) confirmed this observation: 32 out of 204 (16%) of the G418-resistant ES cell clones contained the plasmid as an extrachromosomal element. Previously, analysis of individual clones (e.g. clone 1.19) derived from experiment 2 (Table
1
) revealed that the transfected DNA persisted as an episome without detectable
chromosomal integration of plasmid DNA for 28 cell generations (
9
). To investigate whether plasmid copies integrate into the chromosome as the
cells are further passaged in culture, we analyzed chromosomal DNA from ES
cells grown for 78 cell generations. Total DNA (containing chromosomal and
episomal DNA) as well as low-molecular-weight DNA was extracted from the vector-containing clone 1.19 grown in the presence of G418. The
isolated DNA was digested with
Asp
718 which linearizes pMGD20neo and thus gives rise to a 7.2 kb fragment
indicative of episomal DNA as well as of any integrated concatemeric DNA.
However, integrated DNA will also produce additional fragments of varying
sizes. As shown in Figure
4
, Southern blot analysis failed to detect any additional bands indicative of
integrated copies for up to 78 cell generations after transfection, implying
that pMGD20neo DNA is maintained solely as an extrachromosomally replicating
plasmid in these ES cells during long term passage. Moreover, the episomal DNA
was stable during this long-term passage as the plasmids retained a functional bacterial gene that
confered resistance to ampicillin in
Escherichia coli
(data not shown).
Spontaneous differentiation of ES cells occurs by culturing the cells in the absence of feeder cells and/or of the leukemia inhibitory factor (LIF). To analyze whether the episomal maintenance of pMGD20neo in ES
cells is affected as the cells differentiate, we cultured the vector-containing clone 1.19 (
9
) in the absence of LIF for up to 27 cell generations. As shown in Figure
6
, ES cell differentiation did not alter the episomal state of pMGD20neo as judged from Southern blot analysis of Hirt-extracted DNA digested with
Asp
718. ES cells can also be differentiated into EBs if cultured in a semisolid
medium lacking LIF (
7
). Under these conditions, ES cell clone 1.19, like its ES cell progenitor,
differentiated into typical embryoid bodies. DNA extracted by the Hirt
procedure at day 4 and 9 of differentiation contained episomal plasmid DNA that
was indistinguishable in size from that found in the parental clone 1.19 cells
(data not shown).
There have been a few reports in the past on the maintenance of polyoma-based plasmids in mouse cells. Polyoma-pBR322 recombinants containing the wild-type early region and origin of replication have been shown to
replicate efficiently in mouse fibroblasts and to be maintained as episomes for
at least 6 days at ~1000 copies per cell (
35
). Plasmids recovered from individual clones were structurally identical to the
parental plasmid. However, after 60 days, there was less than 1 copy of free
plasmid per 10 cells. Another recombinant vector, pSV5gpt (
36
), which includes the polyoma early region containing two copies of ori and the
coding sequence for all three T antigens, persisted episomally in mouse
hepatoma cells for at least 50 generations without significant rearrangement of
the vector DNA (
37
). On average, the cells contained 50-100 copies of plasmid DNA and only one integrated copy. The morphology of
the transfected cells was significantly altered and their albumin synthesis
decreased drastically, but the cause of these effects was not established. More
recently, polyoma-related plasmids that persisted as episomes have been found in mouse L (tk
-
) cells (
38
) and in the embryonal carcinoma cell line F9 (
39
). The polyoma-related DNA in the L (tk
-
) cells had numerous sequence changes compared with the input DNA, and when
transfected into F9 cells it was maintained as a plasmid in the cells although
most of it was rearranged (
40
). The morphology of the embryonal carcinoma cells containing the episomal DNA
was not altered nor was their ability to differentiate in the presence of
retinoic acid affected.
To express genes from an episomal vector rather than from integrated copies, we
constructed polyoma-based vectors that were maintained as episomal elements in ES cells (
9
). Apart from introducing a mutant polyoma enhancer fragment (PyF101) enabeling
viral replication in embryonic cells, our vectors contained two different
versions of the early region aiming to express large T only (LT1 and LT20). ES
cells transfected with circular plasmids containing the polyoma early region
mutant LT20 fragment (which requires splicing of the primary transcript)
produced sufficient amounts of large T to support replication of the introduced
plasmid. This in turn led to an increased production of hGH. In contrast, ES
cells transfected with plasmids harboring the polyoma large T cDNA segment
(LT1) failed to produce detectable amounts of viral protein and did not
replicate after transient transfection. These observations are consistent with
those reported by Nilsson and Magnusson (
41
). Based on the stable and long-term episomal maintenance of the vector which was not impaired after
insertion of an up to 6.8 kb DNA fragment (i.e. pRVi6.8-LT20) and on the efficient expression of several genes (i.e. large T,
neo
, hGH) from the replicating vector, we believe that this system will facilitate
many genetic manipulations of ES cells such as efficient overexpression of a gene of interest or genetic complementation by expressing a desired cDNA library from an extrachromosomally replicating vector.
Experiments comparing the targeting efficiency of molecules with no breaks
(supercoiled DNA), single-strand breaks (nicked circle DNA), and double-strand breaks (linear DNA) within the region of homology showed that
linear DNA targeted at a 10-fold higher frequency than nicked circular DNA and at a 34-fold higher frequency than supercoiled DNA (
42
). A reason for the low targeting efficiency of the circular vectors might be
the requirement for DNA strand break events to promote homologous recombination. It has been reported that double strand breaks within the region
of homology increased the number of targeted events by 5- to 10-fold (
21
). Since these experiments were all performed with a non-replicating vector, we tested whether the use of a replication-competent vector that is present episomally in the cells during
several cell cycles will enhance the frequency of homologous recombination. To
this end, we targeted the mouse
hprt
locus with linear or replicating (circular) vectors. Only two homologous
recombination events, however, were found when using circular replication-competent constructs. In contrast, a total of 47 G418
r
, 6-TG
r
clones was found when linear vectors were used to transfect ES cells. Thus, our data underline the requirement of DNA strand breaks to promote homologous recombination.
Transfection of replication-competent vectors into several different mouse cell lines revealed that extrachromosomal maintenance of these
plasmids was not restricted to ES cells. Polyoma virus is known to infect mice
mainly through the respiratory tract and is replicated predominantly in lung,
liver, kidney and colon. It can be propagated in mouse fibroblasts as well as
in primary cell cultures derived from mouse kidney or embryo (
43
). Consistent with the properties of the infectious virus, vector pMGD20neo was
found to replicate efficiently in mouse L-fibroblasts, in mouse renal adenocarcinoma (RAG) cells and in the teratocarcinoma cell line F9 which closely resembles ES cells. Furthermore, the mouse erythroleukemia cell line MEL was also capable to support replication of pMGD20neo. Thus, we presume that this polyoma-based vector carrying a gene of interest can be useful to study gene expression in a variety of cell lines.
Continued replication and persistence of a polyoma based expression vector as an
episome in differentiating ES cells could provide a useful tool to analyze
different states of embryonic development
in vitro
. Moreover, vector-containing ES cells yielded several viable chimeras (
9
) indicating that our plasmid expressing polyoma large T does not affect either
the ES cells' pluripotency or normal embryogenesis. The direct microinjection
of extrachromosomally maintained vectors containing a polyoma replication unit
into fertilized mouse eggs might allow studies in a variety of molecular events such as DNA methylation, replication, recombination and repair during the earliest stages of development, since these events might be analyzed directly from the isolated
extrachromosomal vector. Compared with linear, non-replicating constructs, polyoma-derived vectors might enable a more efficient expression of given genes in developing embryos.
Studies of such genetically altered mice should not be disturbed by the polyoma
large T, as the latency of tumorigenesis in transgenic mice expressing this
viral protein from the polyoma early region promoter has been reported to be
very long: development of pituitary tumors in these transgenic mice began at
about 9-13 months of age (
44
,
45
).
We are indebted to A. Müller (Freiburg i. Brg., Germany) for performing part of the transfections
mentioned in Table
1
, as well as to M. Dieckmann and W. Baier-Kustermann for technical assistance. We also acknowledge F. Fujimura, H.
Zieler, W. Eckhart, K.R. Thomas, P. Hasty , P. Ratcliffe and V. O'Donnall for gifts of material,
A.G. Smith, W. Schaffner, G. Barsh and C. Bauer for helpful discussions, J.
Silke and D. Legler for critical reading of the manuscript and C. Gasser for
the artwork. This project was supported by grants from the Swiss National
Science Foundation (31-36369.92) and the Sandoz-Stiftung (both to M.G.) and by a fellowship from the `Sondermassnahmen des Bundes zur Förderung des akademischen Nachwuchses' (to R.H.W). M.G. and G.D. wish
to thank Paul Berg (Stanford, CA) for his encouragement, support and advice
during the course of this project.
REFERENCES
Return