The Absence of Enhancer Competition between Igf2 and H19 following Transfer into Differentiated Cells

Abstract

H19 and Igf2 are reciprocally imprinted genes that lie 90 kb apart on mouse chromosome 7. The two genes are coexpressed during development, with the H19 gene expressed exclusively from the maternal chromosome and Igf2 from the paternal chromosome. It has been proposed that their reciprocal imprinting is governed by a competition between the genes for a common set of enhancers. The competition on the paternal chromosome is influenced by extensive allele-specific methylation of the H19 gene and its 5′ flank, which acts to inhibit H19 transcription and thus indirectly leads to the activation of the Igf2 gene. In contrast, no allele-specific methylation has been detected on the maternal chromosome, and the basis for the preference for H19 transcription on that chromosome is unresolved. In this investigation, the mechanism controlling the silencing of the Igf2 gene on the maternal chromosome was explored by studying the transcriptional activity of a yeast artificial chromosome (YAC) containing Igf2 and H19 following transfer into differentiated tissue culture cells. Contrary to expectations, both H19 and Igf2 were expressed from a single integrated copy of the YAC. Furthermore, Igf2 expression appeared to be independent of the H19 locus, based on deletions of the H19 gene promoter and its enhancers. These results suggest that an active process is responsible for the transcriptional bias toward H19 on the maternal chromosome and that the hypomethylated state of this chromosome cannot be viewed as a “default” state. Moreover, the active process is not reproduced in a differentiated cell and may require passage through the female germ line.

The distal end of mouse chromosome 7 contains a cluster of genes subject to regulation by genomic imprinting. The linkage of six imprinted genes within a region of ∼1 Mb, namely, p57^KIP2, K_vlqt1, Mash2, Insulin-2 (Ins2), Insulin-like growth factor 2 (Igf2), and H19 (4, 10a, 11, 17, 21, 24), suggests the potential for long-range regional control of imprinting. A connection between the imprinting of Igf2 and H19 was first raised by the observation that, during development, the two genes are identically expressed in endoderm and mesoderm with H19 expressed exclusively from the maternal chromosome and Igf2 expressed solely from the paternal chromosome. The common pattern of expression was explained when a targeted deletion of two endoderm-specific enhancers 3′ of the H19 gene drastically reduced H19 endoderm expression when inherited on the maternal chromosome and reduced Igf2 expression upon paternal inheritance (33). Thus, the two genes require the same regulatory elements for expression, and their reciprocal imprinting may arise as the result of a competition between the genes for transcription (2).

The strong preference for Igf2 expression on the paternal chromosome is thought to arise from paternal chromosome-specific methylation of critical CpG residues in the 5′ flank of the H19 gene. This methylation has all the properties of a gametic mark, as it is established in sperm but not in oocytes, is resistant to the global demethylation that occurs during preimplantation development, and is maintained in all somatic cells of the organism (3, 16, 55, 56). The consequence of this methylation is the silencing of transcription of the H19 gene, and so indirectly Igf2 is expressed (34).

On the maternal chromosome, no gametic mark has been identified, suggesting that perhaps the transcriptional status of the Igf2 and H19 genes is not mediated by epigenetic regulation. In that case the exclusive expression of H19 could be explained by either its closer proximity to the enhancers or the inherently greater strength of its promoter. According to this model, H19’s relative advantage with respect to the enhancers leads directly to the silencing of the Igf2 gene. This idea was tested in experiments involving two different targeted germ line deletions of the H19 gene and its promoter that differed in the extent of the deletion of the 5′ flank of the gene (32, 43). In both instances, the removal of the H19 gene and its promoter led to some degree of activation of the normally silent Igf2 gene on the maternal chromosome, consistent with the premise that transcription of the H19 gene precludes Igf2 expression.

We set out to determine whether we could recapitulate the transcriptional state of the maternal chromosome using a yeast artificial chromosome (YAC) that contains the two genes as well as the tightly linked Ins2 gene. YACs have proven useful in the analysis of large gene complexes in mammals, as they can be readily modified in yeast and then transferred into either tissue culture cells or mice (26, 27, 37, 46, 52). As Saccharomyces cerevisiae lacks the CpG methylation found in mammals, the YACs were expected to mimic the hypomethylated state of the maternal chromosome. Modified YACs were then transferred to Hep3B cells to determine whether the exclusive maternal expression of H19 could be established in a fully differentiated cell.

MATERIALS AND METHODS

Strains and microbial techniques.

Strains used in this study were AB1380 (MATa ura3 lys2-1oc ade2-101c trp1 his5am can1-100oc) and YPH925 (MATα ura3-52 lys2-801 ade2-101 trp1-63 his3-200 leu2-1 cyh2^R kar1-15). The standard yeast media and genetic techniques that were employed are described by Rose et al. (44). The FEW.A12^neo YAC was transferred from AB1380 to YPH925 as described by Spencer et al. (50). Transformation of YAC-containing strains utilized lithium acetate transformation (44).

Isolation and characterization of FEW.A12.

The FEW.A12 Igf2/H19 YAC was isolated from the Massachusetts Institute of Technology mouse YAC library by PCR-based screening (19, 30). High-molecular-weight DNA from YAC strains and Hep3B cell lines was prepared in agarose blocks (9). Following three 30-min washes at 25°C in 10 mM Tris-HCl–1 mM EDTA (pH 8.0) and three 15-min washes at the appropriate digestion temperature in 1× digestion buffer, plugs were digested with restriction enzymes and separated by pulsed-field gel electrophoresis (PFGE) in a 1% agarose gel with a contour-clamped homogeneous electric field (CHEF) gel apparatus at 15°C (10).

DNA analysis.

The structures of the Igf2/H19 locus in FEW.A12 and the three modified versions before and after transfection into Hep3B cell lines were analyzed by restriction enzyme digestion followed by either electrophoresis in 0.8% agarose gels or PFGE. The DNAs were transferred to nitrocellulose (49) and hybridized to DNA fragments that had been labeled with [³²P]dCTP by nick translation (42).

Generation of modified YACs. (i) FEW.A12^neo.

To construct an Igf2/H19 YAC containing the neomycin resistance gene (neo^R) integrated at Ins2, a yeast integrating plasmid was constructed; it contained a 785-bp fragment from Ins2 (probe 1 [see Fig. 1]) that was generated by PCR with the forward primer 5′-TGAAGTGGAGGACCCACAAG-3′ and the reverse primer 5′-GGATGCAGAGGGAACCAAAG-3′. The targeting vector also contained neo^R and the yeast genetic marker LYS2. The vector was linearized at a unique SmaI site within the Ins2 sequence prior to transformation of the AB1380 strain harboring FEW.A12. LYS⁺ transformants were assayed by digestion of genomic DNA with HincII and analysis by Southern blotting with probe 1, the yeast LYS2 gene, or a 630-bp PstI-XbaI fragment from neo^R. High-molecular-weight DNA from transformants was subjected to PFGE, Southern blotting, and hybridization with the EcoRI-SalI fragment encompassing the H19 gene (probe 8 [see Fig. 1]) to verify that the modified YAC was the same size as FEW.A12.

FIG. 1.

Structure of the Igf2/H19 YAC. The 630-kb FEW.A12 YAC is shown at the top. The positions of the Ins2, Igf2, and H19 genes (shaded boxes) and the H19 enhancers (shaded ovals) are indicated in the expanded diagram at the bottom. The orientation of the Igf2/H19 locus relative to the ends of the YAC is arbitrary and could not be determined. The telomeres of the YAC are represented by arrows. Restriction enzyme sites: B, BssHII; E, EagI; S, SacII. The hatched boxes indicate the positions of the hybridization probes used in the study, as follows: 1, a 785-bp probe within the Ins2 gene; 2, 545-bp rat Igf2 cDNA; 3, 5-kb intergenic fragment; 4, 4-kb intergenic fragment; 5, 1.6-kb SfiI-EcoRI intergenic fragment; 6, 4.4-kb HindIII intergenic fragment; 7, 4-kb EcoRI fragment containing the gametic imprint; 8, 3-kb EcoRI-SalI H19 gene fragment; 9, 2.4-kb XbaI-BglII enhancer fragment; 10, 7-kb EcoRI 3′ fragment; 11, 200-bp 3′ PCR-derived fragment.

(ii) FEW.A12^Δp.

To construct an Igf2/H19 YAC with a deletion of the H19 promoter (FEW.A12^Δp), a targeting vector was constructed in pRS405 (47); the vector contained a 1.0-kb BamHI-XbaI fragment from the 5′ flank of H19, a 680-bp DraIII-BamHI fragment from the region immediately downstream of the H19 transcription start site, and the yeast genetic marker LEU2. The vector was linearized at a unique BamHI site between the two H19 flanking sequences prior to transformation of the YPH925 strain harboring the FEW.A12^neo YAC. LEU⁺ transformants were assayed by digesting genomic DNA with EcoRI followed by Southern blotting. Blots were hybridized separately with the 4-kb EcoRI fragment from the 5′ flank of H19 (probe 7 [see Fig. 1]) or the 8-kb SalI-EcoRI fragment 3′ of the H19 gene. Additionally, HindIII-digested DNA was hybridized to a 750-bp XbaI-EcoRI fragment from within the deleted promoter region to verify that the region had been removed.

(iii) FEW.A12^Δe.

To construct an Igf2/H19 YAC with a deletion of the H19 enhancers (FEW.A12^Δe), a targeting vector was constructed in pRS403 (47); the vector contained a 2.6-kb SacI-XbaI fragment from the region 5′ of the H19 enhancers, a 3.7-kb BamHI-EcoRI fragment from the region 3′ of the enhancers, and the yeast markers CYH2 and HIS3. The plasmid was linearized at a unique NsiI site within the SacI-XbaI fragment after destroying an NsiI site within the noncoding region of HIS3. Genomic DNA from HIS3⁺ transformants was digested with BglII or XmnI followed by Southern blotting with probe 8, the 2.4 kb XbaI-BglII enhancer fragment (probe 9), and a 200-bp PCR product amplified from a region 3′ of the DNA used in the targeting vector (probe 11) (33). Two-step gene replacement resulted in deletion of the H19 enhancer region.

Lipofection of Hep3B cells.

The Hep3B cell line and the transfected cell lines were maintained in Dulbecco modified Eagle medium with 10% fetal calf serum. Transfected cell lines were maintained in media with 250 μg of active G418 per ml. One day prior to lipofection, 2.5 × 10⁶ Hep3B cells were seeded onto 6-cm tissue culture plates to achieve 90 to 95% confluence on the day of lipofection. Approximately 2 h prior to lipofection, cells were washed once with OptiMEM (Gibco BRL) and incubated in 3 ml of OptiMEM. YAC lipofection into Hep3B cells was done essentially as described by Lee and Jaenisch (31). High-molecular-weight DNA from YAC strains was embedded in agarose blocks and subjected to PFGE in a 1% low-melting-point agarose gel with a single slot trough in a CHEF gel apparatus at 15°C. The appropriate gel slice was excised from the unstained portion of the gel, dialyzed in 20 mM Tris-HCl (pH 7.6)–1 mM EDTA–200 to 400 μM spermine at 25°C and used within a few days.

On the day of transfection, poly-l-lysine was added to the gel slice to a final concentration of 4 μg/ml. The tube was warmed to 65 to 68°C until the gel melted and was then equilibrated at 40°C for 5 min. Agarose was digested with 15 U of β-agarase at 40°C for 1 to 2 h. DOTAP lipid (20 μg) was added, and the DNA-lipid complex was incubated at room temperature for 30 min. The mixture was then equilibrated in Dulbecco modified Eagle medium, and 1.5 ml of OptiMEM was added. The monolayer of Hep3B cells was washed with OptiMEM, and the transfection complex was applied to the monolayer with a wide-bore pipette. After 4 h, the cells were incubated in fresh medium with serum for 12 to 18 h. Two days later, 250 μg of active G418 per ml was applied; resistant colonies were picked approximately 2 to 3 weeks later.

Identification of YAC-containing cell lines.

A small aliquot of cells was removed for quick DNA preparation with lysis buffer containing 50 mM KCl, 10 mM Tris-HCl (pH 8.3), 2 mM MgCl₂, 0.45% Nonidet P-40, and 0.45% Tween 20. The mouse H19 and Igf2 genes were identified by PCR with, for H19, the forward primer 5′-GTACCCACCTGTCGTCC-3′ and the reverse primer 5′-GTCCACGAGACCAATGACTG-3′ and, for Igf2, the forward primer 5′-CTTCACTGGTCATTCCATCAC-3′ and the reverse primer 5′-GCAGCAGCAGAACTAGATGATTGG-3′. Cell lines that were positive for both genes were lysed (25), and the DNA was prepared (18). To verify that the DNA surrounding the two genes was intact, DNA (15 μg) was digested separately with EcoRI and BglII and hybridized successively to probes 2 and 8. The copy number was determined by comparison to signals from the endogenous genes. Finally, high-molecular-weight DNA was digested with BssHII, EagI, or SacII followed by PFGE and hybridization to probes 2 and 8. Cell lines in which the YAC H19 and Igf2 genes mapped to the same high-molecular-weight DNA fragment were chosen for subsequent expression analysis.

DNA methylation analysis.

Genomic DNA (15 μg) was digested overnight with SacI (H19) or EcoRI plus HindIII (Igf2 differentially methylated region [DMR] 1) in combination with HpaII or MspI and hybridized to probe 7 (for H19) and a 717-bp EcoRI-XbaI probe from the 5′ upstream region of Igf2 (DMR 1) (6).

RNA isolation and analysis.

Total RNA was isolated by guanidinium hydrochloride extraction (1). Single-stranded RNase protection probes were synthesized by in vitro transcription in the presence of [³²P]CTP and purified by gel electrophoresis. RNase protection was performed with a non-allele-specific H19 probe (8) or a 400-bp XbaI-BamHI fragment derived from the 3′ untranslated region of Igf2 according to the directions in the Ribonuclease Protection Assay kit (Ambion). For Northern analysis, RNA (10 μg) was electrophoresed in 1.5% agarose gels in 0.02 M 3-[N-morpholino]propanesulfonic acid–5 mM sodium acetate–1 mM EDTA. RNA was transferred to nitrocellulose (54) and hybridized to a 1.4-kb PstI-EcoRI probe from the 3′ end of the human H19 cDNA or probe 2.

RESULTS

Isolation and characterization of a large Igf2/H19 YAC.

In order to study long-range regulation of transcription in the Igf2/H19 gene cluster, a 630-kb YAC (FEW.A12) containing both genes was isolated from the Massachusetts Institute of Technology mouse YAC library by PCR. To confirm that the YAC was a faithful representation of the ∼130 kb of the mouse genome that contains the Ins2, Igf2, and H19 genes, DNA from the yeast strain harboring the YAC was digested with both rare-cutting and frequent-cutting restriction enzymes followed by hybridization to probes within the Igf2/H19 locus (Fig. 1). The probes were derived from the Ins2, Igf2, and H19 genes as well as the 75-kb intergenic region (probes 3, 4, 5, and 6) and DNA 3′ of H19 (probe 10). These analyses revealed that the genes were situated roughly in the middle of the YAC on ∼125-kb EagI and SacII restriction fragments. Furthermore, no differences in the structure of the region were detected in a comparison with mouse genomic DNA (data not shown).

Expression of the Igf2/H19 YAC in Hep3B cells.

In order to transfer the Igf2/H19 YAC into tissue culture cells, a neomycin resistance (neo^R) selectable marker was introduced into the Ins2 gene located 15 kb upstream of Igf2 by using a yeast integration plasmid (Fig. 2A). The modified YAC, termed FEW.A12^neo, was transferred by lipofection into Hep3B cells, a human hepatoma cell line that expresses both the Igf2 and H19 genes at high levels. Colonies containing the YAC DNA were selected in the presence of G418 and screened for Igf2 and H19 by PCR, followed by restriction mapping by Southern blotting to verify that the genes had been transferred intact (Table 1).

FIG. 2.

Construction of modified Igf2/H19 YACs. Figures are not drawn to scale. (A) Generation of the FEW.A12^neo YAC. The top line shows the original FEW.A12 YAC, and the bottom line shows the structure of the FEW.A12^neo YAC after insertion of neo^R into Ins2. The genes are indicated by open rectangles, and the enhancers are indicated by ovals. Selectable markers within the integrated plasmid sequences and the prokaryotic origin of replication (ori) are black rectangles, and the 5′ and 3′ homologous sequences that were used for correct targeting are hatched rectangles. Restriction enzyme site used to identify transformants: H, HincII. (B) Generation of the FEW.A12^Δp YAC. The top line shows the FEW.A12^neo YAC, and the bottom line shows the structure of the FEW.A12^Δp YAC after the deletion of H19 promoter sequences and the subsequent insertion of plasmid sequences. Above the top line, a short horizontal line indicates the deleted region. Restriction enzyme site: R, EcoRI. (C) Generation of the FEW.A12^Δe YAC by two-step gene replacement. The top line shows the FEW.A12^neo YAC, the middle line shows the His⁺ Cyh^S intermediate, and the bottom line shows the FEW.A12^Δe YAC following excision of plasmid sequences, resulting in deletion of the H19 endoderm enhancers. Restriction enzyme sites: B, BglII; X, XmnI.

TABLE 1.

Summary of YAC transfections

YAC	No. of neoR colonies picked	No. of colonies positive for H19 and Igf2	No. of cell lines with:
			Intact H19	Intact Igf2	H19 and Igf2 linked on PFGE	H19 and Igf2 linked and both genes intact
FEW.A12neo	63	11	7	5a	6	3
FEW.A12Δp	71	15	9	12	4	3
FEW.A12Δe	148	7	6	6	5	4b

^aOnly 6 of 11 cell lines were analyzed at Igf2. 

^bTwo of the four cell lines are multicopy. 

Cell lines that satisfied these criteria were then analyzed by PFGE to assess whether Igf2 and H19 were linked at the integration site (Fig. 3 and data not shown). The two genes are contained on a 225-kb BssHII fragment and 125-kb EagI and SacII fragments in FEW.A12^neo. Occasionally, the two genes were linked on larger-sized DNA fragments, for example, the EagI fragment in FEW.A12^neo.I1 and the BssHII fragment in FEW.A12^neo.E6. We attribute these changes to methylation of one or more cleavage sites following transfer of the YAC to Hep3B cells. Alternatively, the YAC DNA may have sheared between the diagnostic restriction sites but outside the region containing the genes, leading to a junction fragment containing YAC DNA and Hep3B DNA at the site of integration. In 3 of 11 cell lines, the genes were verified to be present intact in single copy and linked on the same-sized DNA fragments (Table 1). Only these lines were used for expression studies.

FIG. 3.

PFGE analysis of FEW.A12^neo YAC-containing cell lines. (A) High-molecular-weight DNA from FEW.A12^neo cell lines (E6, G3, and I1), Hep3B cells, C57BL/6 (BL/6) mouse spleen cells, and the AB1380 strain containing FEW.A12^neo was digested with BssHII, EagI, or SacII and subjected to PFGE. The blot was hybridized to probe 8 (Fig. 1). Migration of ligated λ DNA is shown on the left. The BssHII-digested fragment in E6 is too large to be resolved in this gel. (B) The identical blot was hybridized to probe 2.

The three cell lines were assayed for H19 and Igf2 RNAs by RNase protection assays that were specific for the transfected mouse H19 and Igf2 genes. All three expressed mouse H19 RNA at levels that were higher than in neonatal liver, a tissue that expresses high levels of the gene (Fig. 4). This finding was not unexpected, given that previous studies had shown high-level expression of the H19 gene after transient transfection into Hep3B cells (60). The fact that all three lines exhibited very similar levels suggested that the H19 gene was relatively resistant to any negative position effects caused by the sites of integration of the YAC. The same samples were monitored for human H19 and IGF2 RNAs by Northern blotting to control for RNA integrity (Fig. 4 and data not shown).

FIG. 4.

Expression of H19 and Igf2 in YAC-containing cell lines. Total RNA from FEW.A12^neo, FEW.A12^Δp, and FEW.A12^Δe lines (2 to 5 μg), as well as from neonatal mouse liver and parental Hep3B cells, was assayed by RNase protection for mouse H19 RNA (top panel) and Igf2 RNA (middle panel). The size of the undigested probe is shown in the lefthand lanes. As shown in the bottom panel, the same total RNAs (10 μg) were analyzed by Northern blotting with a probe specific for the human H19 gene (hH19).

The three lines also expressed high levels of Igf2 mRNA (Fig. 4), a finding that was unexpected if transcription of H19 is sufficient to preclude transcription of Igf2. Like the levels of H19 RNA, the levels of Igf2 RNA were consistent among the lines. It was apparent from this result that the YAC was not recapitulating the transcriptional status of the maternal chromosome.

Effect of deletion of the H19 promoter on expression of Igf2.

Although H19 and Igf2 were both expressed from the transfected YAC, it is possible that promoter competition was occurring, albeit with a relaxation in the strong preference for H19 transcription. If this were true, however, the level of Igf2 expression would be expected to increase upon mutation at the H19 locus. Alternatively, there might be no mechanistic link between the expression of the two genes on the YAC, in which case mutations at H19 would have no impact on Igf2. To discriminate between these alternatives, we modified the YAC by deleting an 800-bp XbaI-DraIII fragment spanning the 5′ flank and promoter of H19. In its place, the yeast LEU2 gene was integrated (Fig. 2B).

In order to accomplish this modification, it was necessary to move the FEW.A12^neo YAC to another yeast host that harbored additional auxotrophic mutations like leu2 and his3. The kar1 mutant strain YPH925 is defective in normal karyogamy; therefore, matings between the kar1 mutant strain and the YAC-containing KAR1⁺ AB1380 strain result in “cytoductants,” daughter cells with one haploid parental genotype but a mixed parental cytoplasm. A minority (0.1%) of cytoductants are rare “chromoductants” that have gained chromosomal information from the other parent. By using a mating scheme that selects for transfer of the YAC as well as maintenance of the recipient haploid YPH925 genome, we selected for “YACductants” that had transferred the FEW.A12^neo YAC to YPH925 (50). Once established, the new strain was transfected with the integration plasmid and the proper integrants were selected and confirmed by Southern blotting.

The modified YAC, called FEW.A12^Δp, was transferred to Hep3B cells, and 15 neo^R colonies that were positive for both Igf2 and H19 by PCR were identified. Three lines, all containing a single copy of the YAC, were shown by both conventional electrophoresis and PFGE and hybridization to contain the two intact genes linked on the same large restriction fragment (Table 1 and data not shown).

RNase protection assays were used to assess expression of the imprinted genes in FEW.A12^Δp cell lines. As expected, H19 expression was nearly undetectable in the absence of the H19 promoter (Fig. 4). Northern analysis revealed that the very faint signal observed in RNase protection was derived from a larger-sized transcript that was most likely initiated at a cryptic promoter(s) within the plasmid sequences upstream of the H19 structural gene (data not shown). Despite the loss of the H19 promoter, Igf2 RNA was expressed in these lines at levels comparable to those obtained with the FEW.A12^neo YAC (Fig. 4). Thus, in contrast to previous findings that had demonstrated that the loss of the H19 gene and its promoter increased the level of expression of Igf2 in vivo (32), expression of Igf2 in the YAC-containing cell lines was independent of the H19 promoter.

Independence of Igf2 expression on the H19 enhancers.

The lack of dependence of Igf2 transcription on the status of H19 transcription in Hep3B cells could be explained if the Igf2 gene did not require the H19 enhancers for its expression. Leighton et al. (33) have shown that, in vivo, endodermal expression of the Igf2 gene from the paternal chromosome is absolutely dependent upon two enhancers in the 3′ flank of the H19 gene. To delete the H19 enhancers from the YAC, we constructed a yeast integration plasmid containing the yeast markers CYH2 and HIS3. Two-step gene replacement resulted in a YAC, designated FEW.A12^Δe, bearing a 6.2-kb deletion that encompassed the H19 endoderm enhancers and was identical to the germ line deletion constructed by Leighton et al. (33) (Fig. 2C).

In the absence of the enhancers, expression of both H19 and Igf2 was nearly identical to that when the enhancers were present (Fig. 4). Mouse H19 RNA was detected in three of the four FEW.A12^Δe lines but was absent from a multicopy line, FEW.A12^Δe.X3. Furthermore, all four lines expressed Igf2, although at slightly different levels (Fig. 4). Interestingly, the Igf2 gene was expressed in the FEW.A12^Δe.X3 line, where H19 expression was repressed. These data suggest that Igf2 expression is independent of sequences at the H19 locus within the context of the transfected YAC.

Methylation of H19 and Igf2.

One confounding factor in these experiments that could account for the difference between the behavior of the genes in Hep3B cells and in their normal context could be inappropriate DNA methylation of the YAC. On the maternal chromosome in vivo, no allele-specific methylation has been detected at Igf2 or H19 although there is extensive DNA methylation on both chromosomes in the region between the two genes (29). At the point of gene transfer, the YACs were completely unmethylated; however, it is possible that they acquired methylation once they were integrated into Hep3B cells.

To determine whether this was the case, we investigated the methylation status of the H19 gene in cell lines containing FEW.A12^neo. The CpG sites that are thought to constitute the gametic imprint in the 5′ flank of the H19 gene are contained within a 3.7-kb SacI fragment immediately upstream of the promoter. We examined the methylation of this fragment by digestion with SacI alone or in combination with the methylation-sensitive enzyme HpaII or its methylation-insensitive isoschizomer, MspI. The 3.7-kb SacI fragment was completely digested by HpaII in the three cell lines containing FEW.A12^neo (E6, G3, and I1 [Fig. 5A]), indicating that the region of the gametic imprint was unmethylated on the YAC, as it is on the maternal chromosome. Likewise, the SacI fragment, which is present in its entirety in FEW.A12^Δp, was completely digested in the three FEW.A12^Δp lines (K7, L8, and O7 [Fig. 5A]), demonstrating that the lack of H19 transcription does not lead to DNA methylation.

FIG. 5.

DNA methylation of H19 and Igf2. (A) Position of the 3.7-kb SacI fragment immediately upstream of the H19 transcriptional start site (arrow). The hatched box represents the probe; horizontal bars beneath the diagram represent major digestion products. Restriction enzyme sites: R, EcoRI; H, HpaII; Sc, SacI. DNA was prepared from FEW.A12^neo (E6, G3, and I1), FEW.A12^Δp (K7, L8, and O7), and parental Hep3B cell lines. BL/6, C57BL/6 mouse liver DNA. DNA (15 μg) was digested with SacI alone or with SacI plus HpaII or MspI. Arrows to the left of the gel indicate the sizes of the 3.7- and 2.1-kb digestion fragments. (B) Structure of the 1.5-kb EcoRI-HindIII fragment from Igf2 DMR 1. DNA (15 μg) was digested with EcoRI and HindIII alone or with EcoRI and HindIII plus HpaII or MspI. Arrows to the left of the gel indicate the sizes and positions of the digestion products, in kilobases.

For the Igf2 gene, methylation has been detected on the expressed paternal allele at two different sites, one that is approximately 1.5 kb 5′ of the most distal promoter (DMR 1) and a second that is within an intron of the gene itself (DMR 2) (6, 15, 45). Unlike the methylation at the H19 gene, DMR methylation is acquired after fertilization and therefore cannot constitute a gametic mark. Nevertheless, it has been proposed that the methylation prevents the binding of a repressor of the Igf2 gene and thereby promotes its expression (58).

The methylation status of the 1.5-kb region containing Igf2 DMR 1 was analyzed by digestion with EcoRI and HindIII in combination with HpaII or MspI. The 1.5-kb fragment represents the fully methylated fragment, characteristic of the expressed paternal allele, whereas the lower-molecular-weight fragments are partially unmethylated and characteristic of the maternal allele (15). Both FEW.A12^neo and FEW.A12^Δp cell lines displayed no fully resistant, methylated fragments (E6, G3, I1, K7, L8, and O7 [Fig. 5B]), indicating that although expression of Igf2 in these lines was comparable to the levels in neonatal liver, DNA methylation was not required in this context for expression.

DISCUSSION

The maternal chromosome encoding the Igf2/H19 gene cluster has been viewed as the “default” chromosome, that is, the chromosome without allele-specific epigenetic modifications such as DNA methylation. As such, we would predict that when an unmethylated YAC containing the Igf2 and H19 genes was introduced into tissue culture cells competent to express both genes, H19 would be expressed to the exclusion of Igf2, as observed in vivo. Contrary to expectations, the YAC-containing cell lines expressed both genes at levels comparable to or even higher than those observed in neonatal liver. Furthermore, no evidence for transcriptional interaction between the genes could be demonstrated from the modified YACs in which the H19 promoter or its enhancers had been deleted. These results suggest that DNA introduced into a differentiated cell is missing an epigenetic modification that is required to silence the Igf2 gene on the maternal chromosome.

One caveat to this conclusion is that the neo^R marker was inserted just 15 kb away from the Igf2 gene. The insertion of neo^R into the tightly linked Ins2 gene was intended to maximize the likelihood that neo^R lines would contain the genes of interest. It is possible, however, that by imposing a requirement for neo^R gene expression, we selected for integration sites that were also conducive to Igf2 expression. On the other hand, Sasaki et al. (45) have shown that the maternal Igf2 promoter is in an open, hypersensitive state, yet its transcription is repressed. Therefore, although the presence of the neo^R marker may have selected for open chromatin in the Igf2 region, the situation is not unlike that found in vivo.

Another possible explanation for the coexpression of Igf2 and H19 from the YAC is the requirement for species-specific trans-acting factors for imprinting. To date, the acquisition of an imprinted state of a human gene in a mouse cell, or vice versa, has not been demonstrated. In this regard, it is intriguing that the sequences of the candidate gametic mark regions of both H19 and Igf2r are not conserved in primary sequence between mouse and human cells, although each is subject to allele-specific methylation (28, 48, 55, 56). Furthermore, although Hep3B cells express both genes, it is unknown whether the genes are imprinted. This would be difficult, if not impossible, to determine, because the cells are no longer diploid.

It is formally possible that coexpression of the genes on the integrated YAC DNA reflects an absence of cis-acting sequences rather than trans-acting factors. That is, the YAC DNA could be missing sequences required to establish the Igf2 imprint. We view this as unlikely based on findings that 14-kb H19 transgenes are imprinted in heterologous chromosomal locations (3, 13, 38), and the deletion of a subset of those sequences in the germ line prevents Igf2 imprinting (32).

The explanation we favor is that some germ line modification is required to set up the silencing of Igf2; therefore the maternal chromosome is not a true “default” state. Since the YAC DNA was transfected into a differentiated cell line, epigenetic modifications that are acquired during either passage through the germ line or differentiation would be missing. Support for a requirement for germ line passage comes from the work of Tucker et al. (57). Those authors showed that in DNA methyltransferase (Dnmt)-deficient ES cells, which had lost almost all genome methylation, reintroduction of Dnmt cDNA restored proper methylation and expression of nonimprinted genes but failed to do so for imprinted genes. Only after germ line transmission of these ES cells were the proper methylation and expression patterns of the imprinted genes H19, Igf2, and Igf2r restored (57). In that study, however, the lack of DNA methylation resulted in bi-allelic H19 expression and silencing of Igf2, not the coexpression we observed. Therefore, one must conclude that the epigenetic modification that is missing in the YAC experiments may not be DNA methylation itself but possibly a chromatin conformation that cannot be assembled in a differentiated cell.

A similar requirement for germ line transmission in order to establish appropriate expression is provided by the human β-globin cluster, where developmental switching is thought to proceed via a competition between genes. In this example, genes compete for the locus control region, a large chromatin region consisting of multiple DNase I hypersensitive sites that sequentially engages the embryonic, fetal, and adult genes (12, 20). When YACs containing the region are transferred through the mouse germ line, the appropriate temporal regulation of the gene cluster is recapitulated (5, 14, 23). Transfer of a globin YAC into mouse erythroleukemia cells, however, resulted in coexpression of fetal and adult globin genes shortly after transfer of the YAC despite the adult environment of the mouse erythroleukemia cells (37). Ten to 20 weeks in culture were required for exclusive adult β-globin expression to occur.

It is striking that for the modified YACs, Igf2 expression appears to be independent of sequences at the H19 locus. That is, the overall activity of the Igf2 promoter was unaffected by a deletion of the H19 promoter, arguing strongly that the genes are not competing in this context for access to transcriptional elements. Even more surprising, neither gene showed any dependence upon the H19 enhancers for high-level expression. By both transient transfection into Hep3B cells and stable expression in transgenic mice, the H19 gene has been shown to be dependent upon the 3′ enhancers (13, 60). That dependence was also demonstrated at the endogenous locus, by showing that mice with a germ line deletion of the enhancers had reduced levels of expression of both H19 and Igf2 (33). Thus, the behavior of the transfected FEW.A12^Δe YAC is difficult to understand. One resolution would be the utilization of enhancers for which the genes do not compete.

A possible precedent for this comes from the observation that there is coexpression of H19 and Igf2 on the maternal chromosome in the choroid plexus and leptomeninges, two tissues in the brain in which the imprinting of Igf2 is absent but the imprinting of H19 is maintained (11, 53). Although the basis for this has not been determined, it is possible that it reflects the action of enhancers for which the genes do not compete. In fact, mice carrying a deletion of the 3′ H19 endoderm enhancers maintain choroid plexus expression of both genes, arguing for a separate, yet-to-be-identified enhancer(s) for that tissue (33). It is conceivable that these enhancers are contained on the integrated YAC DNA and inappropriately activated, thereby explaining both the coexpression and the lack of dependence on the H19 enhancers.

In a recent study we observed coexpression of the two genes in mice by supplying the maternal chromosome with a second set of enhancers approximately equidistant from the two genes. In that instance, the requirement for competition was eliminated by the duplication of the enhancers (59). Furthermore, in pathological conditions such as Beckwith Wiedemann syndrome and in transformed cells, it has been shown that there is relaxation of Igf2 imprinting (22, 36, 39, 41). In some instances this is accompanied by a loss of expression of maternal H19, a finding that is readily explained by the competition model (35, 40, 51). In other instances, however, it appears that the two genes are coexpressed on the maternal chromosome (7), much as we have observed in these studies. Although it has been assumed that the relaxation is cis mediated, it is conceivable that it is due to the loss of a trans-acting factor during transformation that is required for gene silencing. If the latter is the case, the coexpression of the transfected genes in transformed Hep3B cells would be explained.

In conclusion, we have shown that expression of the H19 and Igf2 genes appears to be unlinked in an in vitro system whereby modified YACs are transfected into differentiated tissue culture cells. The YACs remained hypomethylated at the Igf2 and H19 loci, however, suggesting that they are mimicking the maternal chromosome, at least in terms of hypomethylation. These data suggest that the hypomethylated maternal chromosome might not be a default state and that an active process during either germ line passage or differentiation is required in vivo to achieve the silencing of the Igf2 gene.