| Nucleic Acids Research | Pages |
Demethylation and expression of methylated plasmid DNA stably transfected into HeLa cells
Introduction
Materials And Methods
Plasmids
In vitro methylation and transfection of plasmid DNAs
Methylation analysis
Results
Effects of in vitro CpG methylation on gene expression driven by various promoters
Variable extents of spontaneous demethylation in the body of the neo reporter gene in G418R transfectants containing an SV40 Pr/E-driven neo gene
Frequent spontaneous demethylation in the SV40 promoter region upstream of the neo gene in G418R transfectants
Discussion
Acknowledgements
References
Demethylation and expression of methylated plasmid DNA stably transfected into HeLa cells
ABSTRACT
INTRODUCTION
Although the vast majority of genetically programmed DNA methylation in vertebrates is at CG dinucleotides (CpGs), there are tissue-specific and stage-specific differences in exactly which CpGs in the genome are methylated (1). In vivo studies indicate that some of this DNA methylation negatively regulates gene expression and thereby plays critical roles in development, imprinting and carcinogenesis (2-5). Similarly, in vitro methylation of expression plasmids at all CpG sites generally greatly decreases reporter gene expression (>90%) upon transient transfection into cultured mammalian cells (6-8). The repressive effects of DNA methylation on transcription can be associated with a high density of m5C residues and the binding of sequence-nonspecific proteins that have a preference for highly methylated DNA regions (9-11). DNA methylation and linked gene repression may also involve specific sites that are recognized by positive- or negative-acting transcription factors whose DNA binding is altered by methylation (12-15).
Because of the role of chromatin structure and methylation-responsive DNA-binding proteins in mediating the effects of DNA methylation (1,9,16), it is important to examine the expression of methylated plasmids upon stable transfection. Stably transfected DNA should have more chromatin-like interactions with histone and non-histone DNA-binding proteins than transiently transfected DNA (17). In transient transfection experiments, the abnormally open structure of the plasmid DNA may not reveal the long-term, chromatin structure-dependent consequences of DNA methylation (18).
We determined, in stable transfection assays, the effect of in vitro DNA methylation on reporter gene expression using plasmids containing various promoters and enhancers. Included in these was the SV40 early promoter/enhancer (SV40 Pr/E), which has six CpG-containing sites recognized by the ubiquitous transcription factor Sp1. Multiple Sp1 sites confer resistance to de novo methylation and target localized demethylation during vertebrate embryogenesis or transfection of embryonal cells in culture (19-23) but they were reported not to do so in non-embryonal cells (24). Our study demonstrates that the SV40 promoter is frequently demethylated in a non-embryonal cell line, HeLa cells, upon stable transfection. The demethylation is probably related to our finding that this promoter is unusually resistant to silencing by in vitro methylation prior to stable transfection. More variable than demethylation in the SV40 promoter was demethylation in the body of its downstream reporter gene, which might be a consequence of demethylation extending from the promoter.
MATERIALS AND METHODS
Plasmids
The plasmid vector pMC1neo PolyA (abbreviated as pMC1neo; Stratagene), which contains the herpes simplex virus thymidine kinase promoter (HSV TK Pr) driving expression of the G418- and neomycin-resistance (neo) gene, was used to construct pPyEhyg by insertion of the enhancer B from polyomavirus DNA upstream of the HSV TK Pr. The episomal pChyg with a hygromycin-resistance (hyg) gene (25) contains a promoterless, unexpressed neo gene and a 9.1 kb fragment from pCEP4 (Invitrogen) with an HSV TK Pr-hyg cassette and oriP and EBNA-1 from Epstein-Barr virus (EBV) DNA. From pRC/RSV (Invitrogen), renamed pSVneo, we excised the 0.6 kb BglII/HindIII fragment containing the Rous sarcoma virus long terminal repeat (RSV LTR) or the 1.2 kb SmaI/HindIII fragment with the bovine growth hormone (BGH) transcription termination signal, f1 ori, and SV40 Pr/E to generate pSneo and pRneo. pRSneo was obtained by deleting the 1.5 kb HindIII/XhoI fragment containing the mouse mammary tumor virus LTR from pMAMneo (Clontech). pCSV was generated from a 2.7 kb BamHI fragment of pMAMneo containing the neo gene driven by the SV40 Pr/E that was inserted into the BamHI site of pCEP4's polylinker.
In vitro methylation and transfection of plasmid DNAs
Plasmid DNAs were methylated in standard reactions with bacterial DNA methyltransferases M.HpaII, M.HhaI or M.SssI (New England Biolabs) using 2 U of enzyme per µg of DNA for 2 h in a volume of 10 µl. For mock methylation, either AdoMet or M.SssI was omitted. After the reaction, the DNA was extracted with phenol-chloroform and precipitated with ethanol. Each batch of methylated DNA was shown to be undigested upon incubation with HpaII and/or HhaI. Calcium chloride-mediated transfection of HeLa cells (ATCC CCL 2) with 10 µg of plasmid was performed as previously (25) using 400 µg/ml of hygromycin B (Hyg, Calbiochem) or 900 µg/ml of G418 (Gibco-BRL) for selection beginning 2 days after the donor DNA was added and changing the medium every 3-5 days. These antibiotics caused massive cell death by 8 or 9 days after transfection and cultures of mock-transfected control cells never recovered from this die-off. Fourteen days after plasmid transfection, either colonies were stained with Giemsa and those containing at least 50 cells were counted or transfectants were further propagated as clones or mixed populations pooled from at least 100 colonies. Unless otherwise stated, transfectants were propagated in the presence of Hyg or G418 for the minimal amount of time necessary to accumulate enough cells for DNA isolation.
Methylation analysis
For Southern blotting, DNA from the transfectants (7 µg) was incubated overnight with HpaII or MspI (20 U/µg of DNA), with internal controls to assure complete digestion. The 32P-labeled probes were as follows: a 2.0 kb BamHI fragment from pSVneo or a 2.7 kb BamHI fragment from pMAMneo containing the neo gene or a 0.9 kb SacII-BstBI fragment of pCEP4 containing the hyg gene. As copy-number standards, diluted BamHI-digested pSVneo plasmid was added to 7 µg of salmon sperm DNA. Bisulfite-based genomic sequencing on 2 µg of transfectant DNA was as previously described (26) except that the annealing temperature was 55°C and the following primers (20 pmol) were used for the first round of PCR (24 cycles) on 200 ng of bisulfite-treated DNA: 5[prime]-AGTGGATTTTTGTTTTAAATTGGAATAATA-3[prime] and 5[prime]-AATCATTTCCA(G)AATAACCTCTCCA-3[prime]. In the second round, the former primer plus 5[prime]-AATTTTGTGGAATGTGTGTTAGTTAGG-3[prime] were used and the second-round PCR products were sequenced.
RESULTS
Effects of in vitro CpG methylation on gene expression driven by various promoters
To study the effects of DNA methylation on gene expression, we stably transfected HeLa cells with various methylated or mock-methylated expression plasmids containing neo or hyg reporter genes (Fig.
Figure 1. Expression plasmids used for stable transfection after methylation in vitro. The structure of the plasmids around the promoters and neo or hyg selection marker genes under study is diagrammed (not to scale). TK Pr, HSV thymidine kinase promoter; PyE, the polyomavirus B enhancer; SV40 Pr/E, the SV40 virus' early promoter and upstream enhancer; CMV Pr/E, the cytomegalovirus immediate early promoter and upstream enhancer; f1 ori, a phage origin of DNA replication; pA, polyadenylation and transcription termination signals from SV40 or, in the case of the pSVneo sequence immediately downstream of the RSV LTR, from the BGH gene. In contrast to the small decrease in the number of antibiotic-resistant colonies upon in vitro methylation of SV40 Pr/E-neo plasmids, CpG methylation reduced by ~100-fold the number of antibiotic-resistant colonies obtained from plasmids in which neo or hyg gene expression was driven by the HSV TK Pr (Table 1). Similar results were obtained from the non-episomal pMC1neo and the episomally maintained pChyg. When the polyomavirus enhancer B was present upstream of the HSV TK Pr in the transfecting DNA, CpG methylation still repressed the formation of most of the antibiotic-resistant colonies although not to as large an extent as in the absence of this enhancer (Table 1, pPyEhyg). CpG methylation of a transfecting plasmid containing the RSV LTR with its enhancer and promoter sequences driving neo gene expression caused a >100-fold decrease in the number of G418R transfectants (Table 1, pRneo). The presence of the RSV LTR's enhancer did not significantly alter the frequency of G418R transfectants when it was present upstream of an SV40 Pr/E-neo cassette. Table 1. There was also a large difference in the extent of silencing by in vitro methylation of an HSV TK Pr-hyg cassette and an SV40 Pr/E-neo cassette when both were present on a single plasmid, namely, the episomal pCSV (Fig. Also, we tested the effect of methylating pCSV only at its 49 HhaI sites (5[prime]-GCGC-3[prime]), 64 HpaII sites (5[prime]-CCGG-3[prime]), or both types of sites. In the HSV TK Pr there are 25 CpGs, including one HhaI site overlapping a low-affinity Sp1 site (27) and one HpaII site further upstream. The HSV TK Pr contains one additional, moderate-affinity Sp1 site (27). Methylation at only the HhaI or HpaII sites or at both types of sites led to decreases of 79, 19 and 100%, respectively, in the yield of HygR colonies. Therefore, HSV TK Pr-hyg expression was much more sensitive to repression by methylation at a small subset of CpGs than was SV40 Pr/E-neo expression to in vitro methylation at all CpG sites. 
b5 × 105 HeLa cells were stably transfected with 10 µg of M.SssI-methylated or mock-methylated plasmid and the resulting G418R or HygR colonies were scored, depending on whether the neo or hyg gene was present in the plasmid. For pCSV, which contained both genes, selection was for G418R in the fourth row of the table and HygR in the fifth row. The average results from duplicate 100 mm plates are shown.
Variable extents of spontaneous demethylation in the body of the neo reporter gene in G418R transfectants containing an SV40 Pr/E-driven neo gene
Southern blot analysis on digests obtained with HpaII, which is inhibited by CpG methylation at its cognate sites, was used to assess DNA methylation in the region of the neo gene in transfectants. Although there are no CpG methylation-sensitive restriction sites in the SV40 Pr/E region, retention of methylation in the gene as well as in the transcription regulatory upstream region should be examined because even DNA methylation confined to the body of a reporter gene can greatly decrease reporter gene expression upon transfection (28,29). Methylation was assessed on DNA from a mixed population of stable G418R transfectants obtained with methylated pCSV. Because this plasmid also contains the hyg gene, methylation in this region too was analyzed. If the examined gene regions were completely demethylated, the hybridizing DNA fragments from HpaII-digested DNA should be of <1 kb (236, 233, 207, 187 and 117 bp with the hyg probe and 863, 434, 275, 181, 171, 131, 102 and 90 bp with the neo probe). In digested DNA from pooled transfectants that were G418R and Hyg-sensitive, no hybridizing bands of <4 kb were observed with the hyg probe, indicating a very high degree of retention of CpG methylation at the tested HpaII sites (Fig. Figure 2. Extensive methylation in the body of the neo and hyg reporter genes of the transfecting pCSV in mixed populations of HeLa transfectants. DNA (7 µg) was analyzed from a mixed population of G418R transfectants obtained with M.SssI-methylated episomal pCSV, which contains the HSV TK Pr-hyg and SV40 Pr/E-neo gene cassettes. The DNA was digested with HpaII (meth/H) or MspI (meth/M) and blot-hybridized to hyg (A) or neo (B) gene probes. The probes spanned from just downstream of the HSV TK Pr past the transcription termination sequence of the hyg gene or from the SV40 Pr/E almost to the end of the neo gene. Purified diluted, unmethylated pCSV plasmid DNA digested with HpaII was present in the first lane of each panel. The sizes of molecular-weight markers are indicated. By Southern blotting, we also confirmed that plasmids isolated from the mixed G418R population obtained with the M.SssI-methylated episomal pCSV were mostly maintained as episomes, as seen by their linearization by HindIII (expected single cleavage of pCSV) and resistance to cleavage by AflII (expected no cleavage of pCSV). By Southern blotting using diluted untransfected plasmid as a copy-number standard, we also determined that transfectants obtained with methylated DNA had an average plasmid copy number of about 10 while those obtained with mock-methylated plasmid had a copy number of about one. However, examination of individual transfectant clones generated with the non-episomal pSVneo indicated that transfectants with low-copy-number plasmid DNA could retain most of their in vitro-introduced neo methylation and, nonetheless, express the plasmid's selection marker (pSVmeth clones 3 and 6; Table 2 and Fig. Table 2. Figure 3. Methylation of the body of the neo reporter gene in the transfecting pSVneo DNA in different HeLa clones. DNA from mixed populations (meth mix 1 or 2) or clones (meth 1, 2 or 6) of G418R transfectants obtained with the non-episomal M.SssI-methylated pSVneo plasmid or transfectant clones of the analogous mock-methylated plasmid (mock 4 or 5) was digested with HpaII (H) or MspI (M). Southern blotting was performed as in Figure 2 with a neo gene probe and 3 µg of transfectant cell DNA for lanes 3, 4, 7, 8, 11 and 12; 1.5 µg for lanes 9, 10 and 15-20; and 0.3 µg from SVmeth clone 1 (meth 1 dil.) for lanes 5 and 6. In lanes 1 and 13, there was MspI-digested DNA from 106 molecules and lanes 2 and 14 had 10 times as much of the same DNA. The neo probe for this analysis contained additional sequences from the F1 ori upstream of the SV40 Pr/E (Fig. 1) so that an extra, weak 1.5-kb fragment hybridizing with ~130 bp of the probe is seen from vector sequences upstream of the SV40Pr/E, which was not seen with the 0.7-kb shorter neo probe used in Figure 2B. Molecular weight markers are shown in lane m. The extent of retention of methylation at HpaII sites within the neo gene varied greatly among SVmeth transfectant clones (Table 2 and Fig. 
bRearrangements in the transfectants in the ~2 kb region, including the 0.7 kb neo gene, were assessed by the appearance of unexpected sizes of hybridizing bands in Southern blot analysis of BamHI digests (data not shown) or MspI digests, as seen in lanes 11 and 12 of Figure 3.
cThe relative extents of methylation at HpaII sites in the neo gene wereapproximated from the fraction of low-molecular-weight hybridizing DNA fragments after HpaII digestion as illustrated in Figure 3. For mix 1, no bands of unmethylated hybridizing DNA fragments were visible (Fig. 3, lanes 3 and 5). ND, not determined.
Frequent spontaneous demethylation in the SV40 promoter region upstream of the neo gene in G418R transfectants
Because CpG-rich regions (CpG islands) overlapping the 5[prime] ends of expressed genes are often hypomethylated in contrast to the sequences in the body of the genes (1,30), demethylation of the promoter and the 5[prime] end of the gene in the SV40 Pr/E-neo cassette may be more important for its expression than demethylation of the central portion of the gene. In order to assay for demethylation of the M.SssI-methylated SV40 promoter region upstream of the neo gene in G418R transfectants, we did genomic sequencing using bisulfite-modified DNA from HeLa transfectant clones obtained with methylated pSVneo (pSVmeth clones). After bisulfite modification, subsequent deamination, PCR amplification and molecular cloning, the methylated C residues of individual strands of DNA are evidenced by persisting C-specific (or complementary G-specific) bands in DNA sequencing gels while unmethylated C residues are converted to T residues. We used PCR primers to amplify ~240 bases that encompassed 13 CpGs from the transcribed strand of the SV40 promoter and the immediate cap-downstream region, but not the adjacent CpG-deficient SV40 enhancer. From the amplified DNA of five transfectant clones, a total of 17 PCR-derived molecular clones were sequenced. Only ~1% of the C residues at non-CpG sites were unconverted after the bisulfite modification and PCR and these were in random locations. Therefore, the efficiency of modification and deamination at unmethylated Cs was ~99% and no significant level of atypical, non-CpG methylation was observed. As a control, five molecular clones from the M.SssI-methylated pSVneo used for transfection were subject to genomic sequencing. They were fully methylated at the examined CpGs.
All but two of the 17 sequenced molecular clones from G418R HeLa transfectants obtained with M.SssI-methylated pSVneo were demethylated in ~30-90% of the 13 CpGs in the SV40 Pr and 5[prime] region of the neo gene (Table 3). Those two molecular clones were from one transfectant colony, pSVmeth 6, and were unmethylated at one CpG in this region while the third molecular clone from this transfectant was unmethylated at 11 of the 13 examined CpGs. In all the molecular clones except the two above-mentioned hypermethylated ones from transfectant pSVmeth 6, the CpGs in four Sp1 sites were unmethylated, namely, those at positions -106, -94, -85 and -73, relative to the major transcription start site of the gene (31). At most of the other examined CpGs, there was considerable variation in the exact patterns of methylation from DNA molecule to DNA molecule, which is seen sometimes (26), but not always (32), at m5CpG-rich regions in human tissue DNAs. Even the SV40 promoter's CpGs at positions -63 and -52, which are also located in Sp1 sites, albeit moderate- or low-affinity ones (27), showed heterogeneity in their methylation state among the molecular clones.
Table 3. 
bThe percentage of analyzed molecular clones that were demethylated at a given site is given.
DISCUSSION
Upon stably transfecting HeLa cells with methylated reporter plasmids, we demonstrated, for the first time, that a non-embryonal cell type could frequently demethylate Sp1 sites in a promoter, namely the SV40 early promoter, and that a large extent of resistance to methylation-induced gene silencing was associated with this demethylation. In vitro methylation of transfected episomal or non-episomal plasmids containing the SV40 promoter and enhancer (SV40 Pr/E) driving neo expression gave no noticeable decrease in transfectant colony size and a decrease of only ~30-45% in the yield of G418R transfectants compared to an ~100-fold reduction in transfectant yield when neo expression from the plasmid was driven by the HSV TK promoter or the RSV LTR. Therefore, the high frequencies of SV40 Pr/E-neo expression and of SV40 promoter demethylation are probably a reflection of their frequency in vivo and not the result of selection for G418R cells.
The effect of in vitro DNA methylation on gene expression has generally been analyzed in transient rather than stable transfection experiments. In some transient transfection studies on HeLa, L929 or NIH 3T3 cells, gene expression driven by the SV40 Pr/E was highly susceptible to CpG methylation-induced silencing (29,33,34), while in others, it showed only a low sensitivity (6,35,36). In vitro methylation of stably transfected plasmids containing a variety of promoter-reporter gene cassettes greatly inhibits gene expression (28,37-39). However, preferential demethylation of several restriction sites in the 5[prime] region of a transfected dihydrofolate reductase minigene was seen in stable transfectants of Chinese hamster ovary cells even without selection for minigene expression (39). Also, in a transfected M.SssI-methylated episomal plasmid, EBV oriP sequences undergo very frequent and targeted demethylation at sites recognized by a viral DNA-binding protein, EBNA-1, starting several days after transfection into a human embryonic kidney carcinoma cell line (40).
In our study, four of the six tightly clustered Sp1 sites in the SV40 Pr were hotspots for demethylation upon stable transfection of HeLa cells. There was much more variability in the extent of demethylation in DNA sequences surrounding these Sp1 sites and in the central portion of the reporter gene. The variation in methylation patterns could be partly due to differences in plasmid copy number. Demethylation can be inhibited by a high copy number as well as a high m5CpG density (41). The transfectant clone with the highest copy number (pSVmeth clone 1) had the most methylation of the neo gene and a lack of detectable unmethylated 5[prime]-CCGG-3[prime] sites (Fig.
Targeting of demethylation to multiple Sp1 sites and the spreading of demethylation to the 5[prime] end of the neo gene could have led to the relative resistance of the SV40 Pr/E to in vitro methylation-induced inactivation. This targeted demethylation could be relevant to the establishment and maintenance of hypomethylation of CpG islands spanning the 5[prime] ends of genes, which often have multiple potential Sp1 sites (42). Clustered Sp1 sites confer local protection against de novo methylation in murine embryonal stem cells, embryonal carcinoma cell lines and early embryos (20,21,23). Also, in murine embryonal cells transfected with in vitro-methylated plasmids containing Sp1 sites in the adenine phosphoribosyl transferase (APRT) promoter or an inactive synthetic promoter, the wild-type Sp1 sites direct local demethylation of CpGs (20,22). However, it was reported that Sp1 site-associated demethylation, detected by Southern blot analysis, occurred in embryonal cells but not in non-embryonal L-cell fibroblasts and L8 myoblasts transfected with a methylated plasmid containing the hamster APRT promoter (24). In our study, it appears that demethylation in HeLa cells, monitored precisely by genomic sequencing, was specifically targeted to the four distal Sp1 sites of the SV40 Pr, three of which are high-affinity sites and one a moderate-affinity site, as opposed to the proximal sites, which have only moderate or low affinity for Sp1 (43,44).
DNA demethylation, which is most prevalent very early in embryogenesis, can be mediated by demethylating activities associated especially (45), but not exclusively (46-48), with embryonic cells. Demethylation can also be mediated passively by inhibition of maintenance methylation or by a combination of active and passive processes (19,40). That we observed a high degree of demethylation at Sp1 sites in the SV40 promoter of transfectants generated with a CpG-methylated plasmid introduced into HeLa cells suggests that targeted demethylation may occur in vivo in normal postnatal cells more frequently than realized and may restore function in genes subjected to aberrant methylation-induced silencing. Such demethylation might serve as an epigenetic repair system to compensate for some physiologically inappropriate methylation of certain CpG islands subject to background levels of adventitious de novo methylation. However, this epigenetic repair may fail or be overwhelmed by selection for fast-growing cells during carcinogenesis so that tumor suppressor genes in cancers are often inactive due to aberrant CpG island methylation.
ACKNOWLEDGEMENTS
This research was supported in part by Public Health Service Grant GM-39999 and development funds from the Tulane Cancer Center.
REFERENCES
This article has been cited by other articles:
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification: 14 May 1999
Copyright©Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
N. Fujita, S. Watanabe, T. Ichimura, Y. Ohkuma, T. Chiba, H. Saya, and M. Nakao
MCAF Mediates MBD1-Dependent Transcriptional Repression
Mol. Cell. Biol.,
April 15, 2003;
23(8):
2834 - 2843.
[Abstract]
[Full Text]
[PDF]
C. Modin, F. S. Pedersen, and M. Duch
Lack of Shielding of Primer Binding Site Silencer-Mediated Repression of an Internal Promoter in a Retrovirus Vector by the Putative Insulators scs, BEAD-1, and HS4
J. Virol.,
December 15, 2000;
74(24):
11697 - 11707.
[Abstract]
[Full Text]
S. Suzuki-Ishigaki, K. Numayama-Tsuruta, A. Kuramasu, E. Sakurai, Y. Makabe, S. Shimura, K. Shirato, K. Igarashi, T. Watanabe, and H. Ohtsu
The mouse L-histidine decarboxylase gene: structure and transcriptional regulation by CpG methylation in the promoter region
Nucleic Acids Res.,
July 15, 2000;
28(14):
2627 - 2633.
[Abstract]
[Full Text]
[PDF]
This Article 
Abstract
Print PDF (203K)
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (9)
Request Permissions
Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Articles by Qu, G. Z.
Articles by Ehrlich, M.
Search for Related Content
PubMed
PubMed Citation
Articles by Qu, G. Z.
Articles by Ehrlich, M.
Social Bookmarking
What's this?