| Human Molecular Genetics | Pages |
©1999 Oxford University Press |
Multipoint analysis of human chromosome 11p15/mouse distal chromosome 7: inclusion of H19/IGF2 in the minimal WT2 region, gene specificity of H19 silencing in Wilms' tumorigenesis and methylation hyper-dependence of H19 imprinting
Introduction
Results
Chromosome 11p15 LOH mapping in WTs
H19 expression and DNA methylation in the series of WTs
Multipoint analysis of chromosome 11p15.5 DNA methylation in WTs and WT-associated kidneys
Multipoint analysis of chromosome 11p15.5 gene expression in WTs
Methylation hyper-dependence of H19 imprinting
Discussion
Domain effects versus gene-specific alterations in the WT2 imprinted region
H19, IGF2 and the WT2 locus
Gene-specific contribution of DNA methylation to normal imprinting of the chromosome 11p15.5/mouse distal chromosome 7 domain
Mechanism of H19 epimutation
H19 as a model for gene-specific epigenetic silencing preceding tumorigenesis
Materials And Methods
Southern and northern blotting
LOH analysis
Semiquantitative RT-PCR
Mouse embryo DNA, RNA and reverse transcription
Allelic expression analysis
Acknowledgement
References
Multipoint analysis of human chromosome 11p15/mouse distal chromosome 7: inclusion of H19/IGF2 in the minimal WT2 region, gene specificity of H19 silencing in Wilms' tumorigenesis and methylation hyper-dependence of H19 imprinting
Received March 18, 1999; Revised and Accepted April 19, 1999
WT2 is defined by maternal-specific loss of heterozygosity (LOH) on chromosome 11p15.5 in Wilms' tumors (WTs). The imprinted H19 gene, in this region, is silenced and hypermethylated in most WTs, and this is linked to pathological biallelic expression of IGF2. However, H19 and IGF2 lie within a larger imprinted domain, and the gene specificity of H19 epimutation has been a persistent question. To address this, we assessed LOH, gene expression and DNA methylation at multiple sites in and around the imprinted domain. LOH mapping showed that the entire domain, including IGF2/H19, is within the minimal WT2 region. Genes within the domain, including IPL/TSSC3/BWR1C, IMPT1/ORCTL2/BWR1A/TSSC5, KvLQT1/KCNA9 and TAPA1/CD81, as well as the zinc finger gene ZNF195/ZNFP104 near the centromeric border, were expressed persistently in many WTs. DNA hypermethylation was not detected with 5[prime] upstream probes for IPL, IMPT1, KvLQT1 and ZNF195 in WTs or WT-associated kidneys. Fully developed WTs showed variable hypomethylation at an imprinted CpG island in a KvLQT1 intron, but this was only complete in the cases with LOH and was not observed in pre-neoplastic WT-associated kidneys with H19 epimutation. Analysis of the corresponding region of mouse chromosome 7 using methyltransferase-hypomorphic mice showed that the H19 imprint was fully erased, but that the allelic bias at Ipl, Impt1, p57 Kip2 and, to a lesser extent, Kvlqt1, persisted. Pre-existing massive allelic asymmetry for DNA methylation and hyper-dependence of transcription on methylation status may underlie the mechanism of gene-specific silencing of H19 in Wilms' tumorigenesis.
INTRODUCTION
Inactivation of H19 in Wilms' tumors (WTs) occurs by one of two pathways: 45% of WTs show loss of heterozygosity (LOH) for chromosome 11p15, reflecting loss of the maternal homolog and duplication of the paternal homolog, while another 30% of cases retain both homologs but show H19 inactivation by biallelic hypermethylation of the transcribed region and upstream sequences (1-4). H19 hypermethylation and silencing is always accompanied by biallelic activation of the linked and reciprocally imprinted IGF2 gene. These two genes interact in cis via a local mechanism of expression-competition (5), but they also lie within a larger chromosomal domain containing multiple other imprinted genes. This raises the question of whether H19 epimutations are gene specific or, alternatively, part of a domain-wide disruption of functional imprinting and/or DNA methylation. The latter possibility must be evaluated, since at least one human disorder, the Prader-Willi syndrome (PWS), is associated with coordinate alterations of DNA methylation and expression of multiple imprinted genes over >1 Mb of DNA (6).
Many interesting genes with known or suspected growth-related functions are located in or near the 11p15.5 imprinted domain. These include p57KIP2, encoding a cyclin-cdk inhibitor (7,8), IPL (TSSC3, BWR1C), encoding a protein of unknown function with sequence similarity to the apoptosis-related gene TDAG51 (9), IMPT1 (BWR1A, ORCTL2, TSSC5), encoding a polyspecific organic metabolite transporter (10), TAPA1 (CD81), encoding a membrane protein with growth-inhibiting activity in lymphocytes (11), and ZNF195 (ZNFP104), encoding a zinc finger transcription factor (12). In principal, all of these genes, as well as a large number of other genes and expressed sequence tags (ESTs) mapping to chromosome 11p15.5, are candidates for involvement in Wilms' tumorigenesis (WT). Related to the significance and gene specificity of H19 epimutation is the question of whether the minimal region of LOH in WTs defining the chromosome 11p15 WT2 locus includes all or only part of the known imprinted domain. Here we assess LOH, DNA methylation and RNA expression in WTs, in WT-associated kidneys and in control kidneys at multiple genes and markers in and around the chromosome 11p15.5 imprinted domain. We also use embryos deficient in DNA methyltransferase to test the relative methylation dependence of imprinting at multiple loci in the orthologous imprinted region of distal mouse chromosome 7. We discuss our findings in terms of the H19/IGF2 gene couple as a candidate for WT2, and as they relate to domain-wide versus gene-specific mechanisms for H19 silencing.
RESULTS
Chromosome 11p15 LOH mapping in WTs
A diagram of the chromosome 11p15.5 imprinted domain and flanking genes and markers is shown in Figure 1, and this region is shown in a genetic map in Figure 2. LOH specific for chromosome 11p15 in WTs was first observed a decade ago (13,14). Since then, a number of studies have produced medium-resolution LOH maps of chromosome 11p15 in WTs (15-17), but a denser map of both intragenic and microsatellite markers now permits a more complete analysis. We used DNAs from 54 primary untreated WTs from several different medical centers. Matched non-neoplastic DNAs were from kidney and/or blood. We used a panel of markers densely clustered in and around the known imprinted domain (15 distinct gene or marker loci, some of which contained more than one polymorphic site, spanning 17 cM from the telomere), as well as two more centromeric markers extending to 94 cM (Fig. 2). For each marker, the quality of the genotyping was good and, consistent with the low amount of non-neoplastic stromal admixture in WTs, in each case the interpretation of allelic losses was unambiguous (examples in Fig. 3).
Figure 1. The chromosome 11p15.5 imprinted domain. Genes in bold are imprinted in humans or mice; all of the genes in plain type except possibly ZNF195 (imprinting not characterized) are biallelically expressed in the human tissues examined to date (45,60,61). We have observed equal biallelic expression at the Nttp1 and H-ras loci in interspecific mouse crosses, supporting a telomeric border of the imprinted domain between the H19 and rpL23l genes (C.P. Walsh and B. Tycko, unpublished data for these genes and for Tapa1, retrievable at http://pathology.cpmc.columbia.edu/tycko/ ). Synonyms for genes are shown in the same boxes (see, for example, refs 62-65). The direction of transcription of the KvLQT1 sense (long arrow) and antisense (short arrow) transcripts are shown. Results are summarized by the shading of the rectangles. Divided rectangles indicate allele-specific DNA methylation. DNA methylation and expression data for L23MRP are from (27). a, WT-associated kidney with H19 hypermethylation; b, WT with biallelic H19 hypermethylation; c, pathological biallelic expression; d, reduced expression in many cases: no absolute correlation with H19 expression (33); e, data from Caspary et al. (31); m, maternal allele; p, paternal allele.
Figure 2. Chromosome 11p LOH mapping in WTs. Genetic distances are from the Cooperative Human Linkage Center (CHLC). All of the markers are either listed in public databases (http://www.resgen.com http://www.marshmed.org/genetics ) or have been described previously (1,4,10,21). Filled circles indicate LOH, open circles indicate retention of heterozygosity and dashes indicate non-informative markers. H19 RNA was scored as negative if it was <5% of fetal kidney levels by densitometry of northern blots. (+), expression between 5 and 10%; +, >10% (see Table 1 for a comparison with semi-quantitative PCR measurements). One additional case was not included in the detailed LOH mapping because of lack of normal DNA, but was included in the expression analysis; this was WT55. WT45 and WT50 (bracket) are bilateral tumors from one patient. Asterisks indicate partial LOH.
Figure 3. Examples of chromosome 11 genotyping in WTs. Results for representative cases and markers are shown. Markers are arranged from left to right in order of distance from the telomere. L, LOH; R, retention of heterozygosity; ni, non-informative. For each case and each marker, the scoring of allelic loss versus retention is unequivocal.
Of the 54 cases, 23 (43%) showed 11p15 LOH, one case (WT42) showed probable constitutional uniparental disomy, and the remaining 30 cases retained heterozygosity (Fig. 2). This rate of LOH is consistent with previous studies (14,15,17,18). The majority of the tumor cases were informative simultaneously at markers both within the known imprinted domain and within the more centromeric 8 cM of DNA. Significantly, in the interval spanning from the telomere 17 cM towards the centromere (tel-D11S1999) and including the entire known imprinted domain there were no examples of subregional allelic loss: all tumors which lost alleles at any one marker lost every other informative marker, and all tumors which retained heterozygosity at any one marker did so at all other informative 11p15 markers. Three cases with single mitotic crossovers at far centromeric locations, resulting in LOH of all of the more telomeric markers, were detected (Fig. 2; WT18, WT19 and WT25); all of the other LOH cases must represent outcomes of more centromeric crossovers or complete non-disjunctions.
Only one case, WT13, showed an internal allelic loss (a double recombination), with retention of all chromosome 11p15 markers but loss of the far centromeric marker D11S4966 (49 cM; corresponding to band 11p13) and again retention of both alleles at D11S2000 at 94 cM. This was a familial tumor, and it did not show H19 silencing. It seems probable that the relevant tumor suppressor (TS) gene target in this case was the chromosome 11p13 WT1 gene. A second case (WT32) showed LOH at the most centromeric marker tested (D11S2000) but retained heterozygosity for chromosome 11p15 markers. The primary conclusion from these data is that in this large series of cases the entire known imprinted domain is included in the region of distal chromosome 11p LOH that defines the WT2 locus.
This conclusion is consistent with a previous study that examined 28 WTs using a less dense panel of markers overlapping with ours (17). Like ours, that study found that all of the allelic losses were attributable to single mitotic crossover events, with loss of all of the markers telomeric to the crossover, or to complete non-disjunctions. In that study, two WTs were considered to have retention of heterozygosity at D11S988 but LOH at all of the more distal (telomeric) markers. This would define a somewhat smaller minimal region of loss (tel-D11S988), albeit one that still includes the entire imprinted domain. Our data are less concordant with a recent study by Karnik et al. (19). These investigators examined a smaller series of cases, 38 WTs, with a marker panel of the same density as ours and with many of the same markers. They reported that at least six WTs gave a pattern of LOH indicating double or quadruple mitotic crossovers within the small region between the telomere and D11S1331 (12.9 cM), while at least four additional cases apparently had single crossovers between this marker and the telomere. These data differ markedly from ours and it is difficult to accept that the recombination positions and frequencies would differ to this degree between two series of the same tumor type. The reasons for the differences in the two data sets are not clear, but the possibility should be considered that some in Karnik et al (19). may have been treated prior to surgery and may therefore have accumulated secondary chromosomal rearrangements. Secondary rearrangements of chromosome 11p in treated WTs have been reported previously (20).
H19 expression and DNA methylation in the series of WTs
Most WTs with 11p15 LOH showed markedly reduced H19 RNA relative to the fetal kidney controls, with 14/16 evaluable cases showing RNA <5% of fetal kidney levels by densitometry, and the remaining two cases showing reductions to ~10% of fetal kidney levels. In addition, of the 28 cases which retained both H19 alleles and for which sufficient RNA was available for northern analysis, 17 (61%) showed H19 silencing. Of the 30 cases which retained both H19 alleles and where sufficient DNA was available for Southern analysis, 20 (66%) showed biallelic H19 hypermethylation on Southern analysis with probes for the 5[prime] upstream region (Fig. 4a-d) and for the transcribed region (ref. 4, and data not shown). These results provide the basis for comparison with other chromosome 11p15.5 genes in the multipoint analysis of DNA methylation and gene expression, detailed below.
a
 |
b
 |
c
 |
d
 |
e
 |
Figure 4. DNA methylation analysis in WTs, WT-associated kidneys and fetal kidneys. (a) Southern blot of DNAs digested with EcoRI + HpaII, except lane M, EcoRI + MspI. The blot was hybridized first with the H19 5[prime] probe, then stripped and rehybridized with the IPLCpG island probe. The WTs and two of the WT-associated kidneys (Ki5 and Ki27) show biallelic hypermethylation of H19, complete for the WTs and partial for the WT-associated kidneys (compare band intensities for the silent versus active alleles within the same lane). Ki27a and Ki27b are two separate samples from different locations in the renal cortex. All of the DNA samples are biallelically unmethylated at the IPL CpG island. Chromosome 11p15 allelic status is indicated. The arrows indicate the direction of gene transcription. E, EcoRI; H, HpaII. (b) DNAs digested with RsaI (R), RsaI + CfoI (R/C) or RsaI + HpaII (R/H), comparing DNA methylation of H19 and a CpG island at the 5[prime] end of the ZNF195gene. The WTs and the WT-associated kidney show biallelic H19 hypermethylation, but all samples lack detectable DNA methylation within the ZNF1955[prime] region. (c) Blot hybridized first with the H19 5[prime] probe and then with probes for CpG island sequences at the 5[prime] ends of IMPT1 and IPL. The WTs and the WT-associated kidney show biallelic H19 hypermethylation and are clearly distinguishable from the control fetal kidneys with monoallelic hypermethylation; while the patterns show some variability, there are no tumor-specific differences in the methylation patterns with the IMPT1 probe. The IPL CpG island lacks detectable DNA methylation. (d) Southern blot hybridized first with the H19 5[prime] probe and then with the KvLQT1 5[prime] and intronic CpG island probes. There is partial hypomethylation of the KvLQT1 intronic CpG island in the WT DNAs, indicated by loss of the fully protected band (asterisk) in the R/C and R/H lanes, but not in the WT-associated kidney (Ki27). The presence or absence of the protected band in the R/C lanes varied among different kidney controls, but the difference between WTs (loss of protected band) and control kidneys (presence of protected band) was highly reproducible in the R/H digestions. (e) Southern blots of PstI (P) and PstI + SmaI (P/S) digested DNAs, hybridized with the KvLQT1 intronic CpG island probe. The cluster of SmaI sites in the core of the CpG island is markedly hypomethylated in the WTs with LOH, but not in the WTs which retain heterozygosity. ROH, retention of heterozygosity.
Multipoint analysis of chromosome 11p15.5 DNA methylation in WTs and WT-associated kidneys
In spite of its strong imprinting and in contrast to H19, the human p57KIP2 gene is hypomethylated biallelically both in normal tissues and in WTs (21). To ascertain whether this unexpected finding might pertain to other imprinted genes in the domain, we assessed CpG methylation in sequences near the transcriptional start sites of the imprinted genes IPL, IMPT1 and KvLQT1. We also assessed DNA methylation of the ZNF195gene. This gene lies within 1 Mb centromeric of the known imprinted domain (Fig. 1), is highly expressed in kidney and encodes a zinc finger transcription factor similar to WT1 (12). For these analyses, we initially focused on 12 WTs, five with 11p15 LOH and seven without LOH (WT5, WT7, WT8, WT12, WT15, WT23, WT25, WT27, WT28, WT30, WT31 and WT32; Fig. 2). Importantly, among the WTs without LOH were the three cases from our series which showed a high level of pathological biallelic H19 hypermethylation in the adjacent non-neoplastic kidney parenchyma, i.e. H19 epimutation prior to overt neoplastic transformation (Ki5, Ki8 and Ki27). As controls, we used DNAs from midgestation fetal kidneys (the appropriate stage-matched control tissue for WT), as well as normal kidneys from WT patients with tumor-specific 11p15 LOH whose non-neoplastic kidney parenchyma lacked H19 hypermethylation.
To span large extents of CpG-rich sequences, we used a strategy in which multiple methylation-sensitive restriction sites were assessed in single lanes of digested DNA. Considering first the results for H19, after digestion with the non-methylation-sensitive enzyme EcoRI and the methylation-sensitive restriction enzymes HpaII or CfoI and hybridization of the blots with an H19 upstream region probe, a biphasic pattern of methylation was seen in the control kidneys (Fig. 4a, Ki19 and Ki28). Similar results were obtained after double digestion with RsaI and either CfoI or HpaII (Fig. 4b-d). A major population of DNA fragments was fully protected from digestion at every site extending over ~9 kb of DNA, starting from ~5 kb upstream of the transcriptional start site and including downstream DNA encompassing the entire body of the gene. By allele-specific PCR, these fragments correspond to the silent allele (22). In the control tissues, a second population of DNA fragments was digested completely or nearly completely at every CfoI or HpaII site spanned by the 5[prime] probe. This was indicated by the fact that every limit-digest fragment of >150 bp predicted from the restriction map could be visualized on the blots. These small fragments, corresponding to the transcriptionally active allele, were either reduced in intensity relative to the protected bands or were undetectable in the WTs. In addition, densitometry showed that the limit-digest fragments were relatively reduced in intensity by at least 5-fold in each of the three WT-associated non-neoplastic kidneys with H19 epimutation (Fig. 4b-d, and data not shown).
These results extend our previous observations of biallelic H19 hypermethylation in most WTs and in some WT-associated kidneys (1,4). We conclude that ~8% of WT patients (three of the 39 cases in our series for which sufficient kidney DNA was available for Southern analysis) have high-level mosaicism (>80% of cells) for biallelic H19 hypermethylation in their non-neoplastic kidneys. This frequency is roughly consistent with a study by Okamoto et al. (23), who followed up our original observations (1,4) by carrying out quantitative (densitometric) Southern analysis on their WT-associated kidneys. That study found tissue mosaicism for H19 epimutation at >50% in three of eight WT patients.
Turning to the data on the more recently identified imprinted genes, when the Southern blots were rehybridized with the IPL 5[prime] probe, spanning a CpG island, entirely different results were obtained. For this gene, every CfoI and HpaII site that could be assayed was cleaved fully on both alleles, both in the controls and in the WT DNAs. This was seen as complete conversion to every limit-digest fragment of >150 bp predicted by the restriction map (Fig. 4a and c). These results indicate that the restriction digests were complete, and that, both in control kidneys and in WTs, the CpG island of the imprinted IPL gene is hypomethylated biallelically.
As a second marker, we assessed a CpG-rich sequence overlapping the apparent transcriptional start site of the ZNF195 gene. This zinc finger family gene, which is located <1 Mb centromeric of the known imprinted domain, near NUP98 (24), is highly expressed in kidney and has sequence similarity to the WT1 TS gene. It is not known whether ZNF195 is imprinted. The Southern blots rehybridized with the ZNF195 probe showed complete digestion to the limit-digest fragments predicted from the CfoI and HpaII restriction maps, with no evidence of hypermethylation in the WT or WT-associated kidney samples (Fig. 4b, and data not shown).
We next examined CpG methylation of the imprinted IMPT1 gene, which lies between IPL and p57KIP2. We used a probe spanning a CpG-rich region overlapping the start site of the longest cDNA (10). Since the pattern of human IMPT1 transcripts on northern blots is complex, this may be only one of several potential start sites. The Southern blots showed that methylation at CfoI and HpaII sites in this region was variable, but no consistent differences were found between the WT, WT-associated kidney and normal fetal kidney samples (Fig. 4c, and data not shown). As a fifth marker, we assessed the KvLQT1 gene, located in the center of the imprinted domain. This gene contains a CpG island spanning the 5[prime] end of the longest cDNA sequence (GenBank accession no. AF000571). When the blots were rehybridized with a probe for this CpG island, complete conversion to limit-digest fragments was observed with CfoI and HpaII in all of the samples (Fig. 4d, and data not shown).
Recently, Higgins and co-workers (25,26) identified an intronic CpG island within the KvLQT1 gene, downstream of exon 10, which shows allele-specific DNA methylation. This CpG island is located near the origin of a number of ESTs (e.g. GenBank accession nos AA155639, AA155694, AA359588, AA533100 and AA701413) which lack open reading frames and which are in an antisense orientation relative to the KvLQT1 mRNA transcript. A probe for this CpG island showed a biphasic pattern of DNA methylation in normal kidney, and it revealed some degree of hypomethylation in all of the WTs (Fig. 4d and e, and data not shown). Among the WT cases that retained chromosome 11p15 heterozygosity, the demethylation was always partial, affecting flanking sites but not extending into the core of the CpG island (Fig. 4d, and data not shown). Loss of the hypermethylated bands was much more complete in the WTs with frank chromosome 11p15.5 LOH (Fig. 4e), an expected finding which is trivially explained by physical loss of the hypermethylated maternal allele. Hypomethylation of the KvLQT1 intronic CpG island was not observed in the three pre-neoplastic kidneys with H19 epimutation; these showed a pattern of methylation similar to that of normal fetal and juvenile kidney (Fig. 4d, and data not shown). We conclude that subtle hypomethylation at this KvLQT1 intron CpG island is characteristic of fully transformed WT cells, but that it is not a feature of the pre-neoplastic kidney cells of sporadic WT patients. In summary, among the six CpG-rich regions assessed in these five chromosome 11p15.5 genes, only the upstream region of the H19 gene was hypermethylated both in WTs and in a subset of WT-associated kidneys (Fig. 1).
Multipoint analysis of chromosome 11p15.5 gene expression in WTs
We next assessed RNA expression from five genes in or near the imprinted domain: ZNF195, IPL, IMPT1, KvLQT1, TAPA1 and H19. TAPA1 is located in the central area of the known domain, between the ASCL2/Mash2 gene, which is imprinted in mouse placenta (27), and the KvLQT1 gene, which is imprinted in several fetal tissues of humans and mice (28). TAPA1 encodes a widely expressed cell surface protein which mediates growth inhibition of some cell types on antibody cross-linking (11). Homozygous deletion of Tapa1 from the germline of mice produces a hyperproliferative phenotype in lymphocytes, but hemizygous deletion has not been reported to produce an obvious parent-of-origin-dependent effect (29,30). Nonetheless, this gene is subject to a definite parent-of-origin-dependent allelic expression bias early in mouse development (31; L. Yuan, C.P. Walsh and B. Tycko, unpublished data, retrievable at http://pathology.cpmc.columbia.edu/tycko/ ).
For each gene, RNA expression in WTs was compared with the most nearly histologically matched control tissue, midgestation fetal kidney. In fetal kidneys, ZNF195, TAPA1 and H19 RNAs are all expressed at high levels and could therefore be compared on northern blots. IPL, IMPT1 and KvLQT1 are expressed at much lower levels and so these genes were analyzed by semi-quantitative RT-PCR. The northern blot results for TAPA1 and ZNF195 are shown in Figure 5. Both of these genes are expressed at equal levels in fetal kidney and WTs. Specifically, in WTs with H19 silencing, there is persistent high expression of TAPA1 and ZNF195.
Figure 5. Northern blot analysis of ZNF195 and TAPA1 mRNA in WTs and fetal kidneys. There are two major ZNF195 transcripts (arrows) with variable ratios of intensity in the different samples (12). WTs with H19 silencing show persistent expression of ZNF195 and TAPA1. A G3PDH probe was used as a loading control.
The results for KvLQT1, IMPT1 and IPL are shown in Figure 6 and Table 1. To allow assessment of possible domain-like effects on gene expression, the WTs for this analysis were enriched deliberately in H19-expressing cases (12 cases with H19 silenced and six cases with persistent H19 expression). The cDNAs were serially diluted and PCR was carried out with primers for the gene of interest together with primers for a control gene, glyceraldehyde 3-phosphate dehydrogenase (G3PDH). All of the RT-PCR products spanned at least one intron, so there was no possibility of artifactual detection of genomic PCR products. Conditions were established which allowed both RT-PCR products to be visualized in the non-saturated range in single-tube reactions, and the expression was measured by phosphorimaging. Control reactions omitting reverse transcriptase were negative, both for the gene of interest and for G3PDH (Fig. 6, and data not shown). As a second control, we used an identical RT-PCR approach to measure H19 RNA in the fetal kidneys and WTs (Fig. 6 and Table 1). By this method, there was good concordance with results previously obtained by northern analysis (Table 1).
Figure 6. RT-PCR analysis of H19, KvLQT1, IMPT1 and IPL RNA in WTs and fetal kidneys. Serial 5-fold dilutions of cDNAs were subjected to PCR with primers specific for the indicated genes. Representative cases are shown. Data from phosphorimaging of lanes in the non-saturated range are in Table 1.
Table 1. RNA from the H19, KvLQT1, IMPT1 and IPL genes in WTs compared with midgestation fetal kidneys
| Tissue | H19 | 11p15 | H19 | H19 | KvLQT1 | IMPT1 | IPL |
| Me-CpG | alleles | northern | RT-PCRa | RT-PCRa | RT-PCRa | RT-PCRa | |
| FKi1 | +/- | 1.0 | 1.3 (0.17) | 1.2 (0.27) | 0.89 (0.05) | ||
| FKi2 | +/- | 0.78 | 0.52 (0.06) | 0.49 (0.12) | 1.1 (0.18) | ||
| FKi4 | +/- | 1.3 | 1.8 (0.25) | 1.1 (0.27) | 0.86 (0.18) | ||
| FKi6 | +/- | 0.97 | 0.45 (0.16) | 1.1 (0.36) | 1.1 (0.35) | ||
| WT29 | +/+ | ROH | <0.05 | 0.00 | 0.59 (0.17) | 0.37 (0.11) | 0.51 (0.20) |
| WT32 | +/+ | ROH | <0.05 | 0.00 | 0.63 (0.24) | 0.12 (0.05) | 0.19 (0.13) |
| WT23 | +/+ | ROH | <0.05 | 0.02 | 0.75 (0.15) | 1.1 (0.12) | 0.60 (0.15) |
| WT18 | +/+ | LOH | <0.05 | 0.03 | 0.51 (0.10) | 0.54 (0.24) | 0.34 (0.02) |
| WT20 | +/+ | ROH | <0.05 | 0.03 | 0.60 (0.12) | 0.38 (0.18) | 0.20 (0.12) |
| WT21 | +/+ | LOH | <0.05 | 0.05 | 0.53 (0.08) | 0.57 (0.07) | 0.14 (0.04) |
| WT15 | +/+ | LOH | <0.05 | 0.06 | 0.70 (0.13) | 0.49 (0.14) | 0.42 (0.12) |
| WT17 | +/+ | LOH | <0.05 | 0.06 | 0.40 (0.03) | 1.5 (1.1) | 0.79 (0.22) |
| WT27 | +/+ | ROH | <0.05 | 0.07 | 0.70 (0.06) | 0.41 (0.17) | 0.41 (0.21) |
| WT30 | +/+ | LOH | <0.05 | 0.07 | 0.53 (0.21) | 0.23 (0.16) | 0.97 (0.38) |
| WT32 | +/+ | ROH | <0.05 | 0.15 | 0.54 (0.03) | 0.29 (0.05) | 0.16 (0.04) |
| WT28 | +/+ | LOHb | 0.08 | 0.15 | 0.64 (0.26) | 0.29 (0.04) | 0.67 (0.12) |
| WT31 | +/- | ROH | nd | 0.94 | 2.7 (0.84) | 1.3 (0.37) | 0.57 (0.06) |
| WT22 | +/- | ROH | 0.8 | 1.6 | 1.9 (0.57) | 0.90 (0.07) | 2.7 (1.1) |
| WT24 | +/- | ROH | >1.0 | 3.4 | 0.62 (0.16) | 0.42 (0.13) | 0.96 (0.15) |
| WT55 | +/- | ROH | >1.0 | 1.1 | 1.2 (0.21) | 0.07 (0.01) | 0.24 (0.06) |
| WT13 | +/- | ROH | >1.0 | 1.0 | 0.75 (0.17) | 0.21 (0.05) | 0.05 (0.01) |
| WT16 | +/- | ROH | >1.0 | 0.90 | 0.60 (0.10) | 0.24 (0.07) | 0.11 (0.01) |
aValues are normalized by setting the average of the four fetal kidneys = 1.0.
bPartial LOH.
For IMPT1, 16 of 18 WTs examined showed mRNA expression at levels [ge]20% of the average fetal kidney value, and this was true for IPL mRNA in 12 of the cases. For KvLQT1, the variability in expression levels among the four different control fetal kidneys suggested that the expression of this gene might vary markedly according to metabolic state or the stage of cell differentiation. Nonetheless, all of the WTs showed persistent KvLQT1 mRNA at levels [ge]40% of the mean fetal kidney value. IMPT1 and IPL also showed some variability of expression among the four control fetal kidneys, and for these genes there are previous data suggesting changes in expression with tissue maturation (9,10). The strongest conclusions from these data are that there is no obvious correlation of KvLQT1, IPL or IMPT1 expression with the chromosome 11p15 allelic status of the tumors, and that none of these genes is silenced uniformly in WTs with H19 inactivation. There is also no absolute correlation between expression of any of these three mRNAs and H19 RNA (Table 1 and Fig. 6).
The only indication of a domain effect, albeit a weak and variable one, was that the average level of KvLQT1 mRNA in the WTs which expressed H19was somewhat higher than the average in the WTs which did not express H19 (Table 1). This trend is similar to our previous findings for p57KIP2 mRNA, which when measured by RNase protection assays did not show an absolute correlation with H19 expression but was highest on average in the H19-expressing WTs (21,32).
Methylation hyper-dependence of H19 imprinting
In view of the gene specificity of biallelic H19 hypermethylation in WT patients, and the qualitatively unique extent of its allelic asymmetry for DNA methylation, we were curious about whether this gene might be more dependent on DNA methylation for its normal imprinting, compared with other genes in the extended domain. The allelic expression bias at some, though not all, imprinted genes is erased by the demethylating agent 5-aza-deoxycytidine (AzaC) (21,33,34), and previous studies have used mice deficient in the Dnmt1 methyltransferase to show a role for DNA methylation in imprinting (31,35). However, the relative sensitivities of genes in the mouse distal chromosome 7 domain to erasure of imprinting by DNA demethylation have not been assessed completely. To examine this, we used mice carrying a targeted mutation in the DNA methyltransferase gene, which reduces methylation levels to ~30% of wild-type levels in homozygotes (36). To obtain homozygous knockout mice carrying marked (Mus m. castaneus) alleles, we first intercrossed C57BL/6J Dnmt1N/+ mice with CAST Dnmt1+/+ mice and screened for Dnmt1N/+ heterozygotes in the F1 generation. These were then backcrossed to the C57BL/6J heterozygotes and embryos collected at 9.5 days post-coitum (d.p.c.), a day before the homozygous embryos would normally die. DNA prepared from fragments of Dnmt1N/N embryos showed demethylation of IAP endogenous retroviral sequences (37) by Southern blotting (data not shown) and strong IAP up-regulation was seen on northern analysis (Fig. 7a). Embryos were also typed for the presence of a CAST chromosome 7 using Impt1 and Rpl23 genomic PCR and restriction fragment length polymorphism (RFLP) analysis (data not shown). We then tested for the reactivation of the silent allele of the H19 gene in Dnmt1N/N embryos carrying a CAST and C57BL/6J chromosome 7. By single strand conformation polymorphism (SSCP) analysis of cDNA PCR products from the individual 9.5 d.p.c. embryos, the H19 paternal allele was found to be fully reactivated in the Dnmt1N/N homozygotes, while remaining strongly imprinted in their heterozygous and wild-type littermates (Fig. 7a).
a
 |
b
 |
Figure 7. Multipoint analysis of imprint stability in Dnmt1 hypomorphic mice. (a) Activation of endogenous retrovirus transcription and erasure of the H19 imprint in Dnmt1N/N embryos. The IAP analysis is a northern blot with the 28S RNA probe as a loading control. The H19 analysis is an SSCP gel. The numbers refer to individual conceptuses. The genotypes are indicated below each lane: +, wild-type or N/wt. Duplicate determinations from independent aliquots of cDNA are indicated by primes. (b) Lack of erasure of the p57Kip2, Impt1 and Ipl imprints and partial erasure of the Kvlqt1 imprint in Dnmt1N/N conceptuses. The p57Kip2analysis is by SSCP, the Impt1 and Kvlqt1 analyses are by RFLP and the Ipl analysis is by slot blotting. The Ipl-A allele-specific oligonucleotide (ASO) reproducibly gives a weaker signal than the Ipl-C ASO on genomic PCR products from heterozygotes (data not shown), presumably because of the reduced G/C content due to the C->A substitution. This does not compromise the interpretation of the imprinting analysis, which involves side-by-side comparisons of the different embryos. The ratio of CAST to BL/6 allelic expression (C/B ratio) was determined for Kvlqt1 by phosphorimaging. The asterisks indicate that conceptus 2 is a homozygous control, showing that the RFLP digestions are complete. Note the stronger Kvlqt1 imprinting in conceptus 11, which was taken at 8.5 d.p.c. rather than 9.5 d.p.c. This is consistent with the labile imprinting of this gene in development (66).
We next analyzed the Kvlqt1, p57Kip2, Impt1 and Ipl genes. We used the same cDNAs from the individual embryos with known genotypes that previously had been assessed for H19 allelic expression. We generated gene-specific RT-PCR products and distinguished the paternal and maternal alleles using RFLP and SSCP analysis for Kvlqt1, RFLP analysis for Impt1 (10), SSCP analysis for p57Kip2and slot blotting for Ipl. The Kvlqt1 gene showed a partial reactivation of the imprinted allele in the Dnmt1N/N homozygotes, but the degree of reactivation of the maternal allele at this locus was much less than that observed for H19 in the same embryos (Fig. 7b). Since a recent study of imprinting in the more severely demethylating Dnmt1 S-allele background reported complete loss of expression (rather than loss of imprinting) of Kvlqt1 (31), we examined the total expression of this gene in the Dnmt1N/N embryos and the normal controls. RT-PCR under non-saturating conditions and normalized to an Rpl23 control showed persistent expression in all of the embryos, regardless of genotype (C.P. Walsh and B. Tycko, unpublished data, retrievable at http://pathology.cpmc.columbia.edu/tycko/ ). In even stronger contrast to H19, each of the three other genes, p57Kip2, Impt1 and Ipl, remained strongly imprinted in multiple Dnmt1N/N homozygous embryos (Fig. 7b). We conclude from these data that among these five imprinted genes, only H19 is susceptible to full erasure of functional imprinting in the Dnmt1N/N hypomethylating background.
In Figure 1, we tabulate our findings together with previous data from Caspary et al. (31). That study used a related but more severe demethylating background (the Dnmt1 S-allele) to examine allelic expression of the H19, Mash2, Kvlqt1 and p57Kip2 genes. This genetic background also uncovered different degrees of methylation dependence of the functional imprint among the different genes, and the combined results clearly indicate that among the genes in the domain the maintenance of imprinting at H19 is most dependent on DNA methylation.
With reference to Caspary et al. (31), it appears that the mouse p57Kip2 gene is susceptible to partial, but not complete, erasure of imprinting under more severely demethylating conditions. This may be a direct effect of demethylation of critical sites in or near this gene which are resistant to demethylation in the Dnmt1N/N embryos. Alternatively, it may be an indirect effect due to altered expression of other trans-acting genes capable of modifying p57Kip2 expression from one or both alleles in the more severely developmentally perturbed Dnmt1 S-allele homozygotes. A comparison of the findings regarding Kvlqt1 is also interesting. In the Dnmt1 S-allele mutants, there was extinguishing of expression of this gene, while in the less severe Dnmt1 N-allele mutants, there was partial loss of imprinting without loss of expression. This `biphasic' response to demethylation may reflect at least two distinct types of cis-acting methylated regulatory sequence in the large Kvlqt1 locus.
DISCUSSION
Domain effects versus gene-specific alterations in the WT2 imprinted region
With increasing evidence for clustering of imprinted genes in domains, an obvious question has been whether diseases caused by abnormal expression or silencing of imprinted genes might be contiguous multigene syndromes. This possibility has received the most support from studies of the chromosome 15q11-q13 imprinted domain and its associated genetic syndromes, PWS and Angelman syndrome (AS). In the PWS/AS region, the CpG islands in the 5[prime] regions of the ZNF127, NDN and SNRPN genes are normally markedly hypermethylated on the inactive maternal alleles (reviewed in ref. 6). DNA from non-deletion cases of PWS shows an abnormal `bimaternal' methylation pattern on Southern blotting, with multiple gene-associated probes spaced over ~1.5 Mb, and this is associated with coordinate loss of expression of ZNF127, NDN, SNRPN and IPW (reviewed in ref. 6). It is thought that loss of expression of more than one of these genes, or other imprinted genes in the domain, may be necessary for the complete PWS phenotype.
The current study focuses on the other well-studied imprinted domain in the human genome, on chromosome 11p15.5. This region harbors genes involved in two related disorders, Beckwith-Wiedemann syndrome (BWS) and WT. BWS, a somatic overgrowth syndrome, is known to be `multigenic', but not at all in the same sense as may be true for PWS. BWS occurs by at least two different pathways: either by overexpression of the imprinted IGF2 gene through its biallelic activation (38) or, in a different subset of cases, by inactivation of the p57KIP2 gene through germline mutation of the expressed maternal allele (32,39,40). The IGF2 activation in BWS is due in some cases to H19 inactivation and hypermethylation (41,42) and in others to an H19-independent mechanism, perhaps involving disruption of a regulatory element in the nearby KvLQT1 gene (28,43). Thus, while different classes of BWS can occur by lesions in different genes, there is no evidence for BWS as a contiguous multigene syndrome.
Here we have addressed this question as it relates to WTs. Gene expression analysis at ZNF195, IPL, IMPT1, KvLQT1, H19 and TAPA1 and DNA methylation analysis at CpG-rich sequences in the first five of these genes indicate that among these genes, extensive DNA hypermethylation and transcriptional silencing is restricted to H19. While there is case-to-case variability in expression of IPL, IMPT1 and KvLQT1 mRNAs, this is also seen among the fetal kidney controls and is likely due in part to varying cell differentiation, not to tumor-specific mechanisms. A similar situation may pertain for the p57KIP2 gene. For this gene, many, though not all, WTs have significant reductions of mRNA relative to fetal kidney when assessed by RNase protection, but there is no absolute correlation with H19 silencing (21,32). In situhybridization has suggested that some of the variability of p57KIP2 mRNA in WTs may be accounted for by the differentiation of the neoplastic cells, with highest expression in epithelial-predominant areas (44). H19 silencing, in contrast, is not dependent on the relative amounts of epithelial, blastemal and stromal components in WTs (1), and this conclusion is reinforced by the fact that biallelic DNA hypermethylation of this gene affects nearly 100% of the cells in the tumors.
Our finding that H19 hypermethylation is gene specific extends previous data showing lack of hypermethylation of the chromosome 11p15.5 genes L23MRP (45), p57KIP2 (21) and HRAS (3). There was also lack of detectable DNA hypermethylation in WTs and WT-associated kidneys with four anonymous probes within 50 kb upstream and downstream of H19, outside of the 9 kb core hypermethylated region (1). Consistent changes in CpG methylation have been found in WTs using IGF2 probes (e.g. ref. 3), but these changes may be secondary to the inactivation of H19. This conclusion is based on the established dependence of mouse Igf2 imprinting and DNA methylation on H19 promoter activity (5,46) as well as by recent findings in which the presence or absence of the active H19 transcription unit affected Igf2 allelic expression to some extent independently of Igf2 DNA methylation (47).
The partial hypomethylation of the imprinted KvLQT1 intron CpG island observed in the fully developed WTs, but not in the pre-neoplastic WT-associated kidneys, is similar to previous findings at the chromosome 11p15.5 L23MRP locus (45), and also similar to results with a probe for the H-RAS variable number of tandem repeats (VNTR) (B. Tycko, unpublished data). It suggests that, as in other tumor types, widespread DNA hypomethylation accompanies more localized and gene-specific DNA hypermethylation in WTs. Whether this broad partial hypomethylation contributes to the tumor phenotype or, alternatively, is simply a consequence of full neoplastic transformation will be an interesting topic for future research. While we have not observed complete demethylation of this region in WTs which retain 11p15.5 heterozygosity, the proximity of the intronic KvLQT1 CpG island to BWS-associated translocation breakpoints is intriguing, and its possible role in that disorder is under investigation (26).
H19, IGF2 and the WT2 locus
Our LOH mapping places both H19 and IGF2 in the minimal region defining WT2. The H19 gene satisfies several criteria for WT2: (i) it is silenced in all cases with LOH as well as in more than half of the cases which retain both alleles; (ii) its expression is extinguished by a gene-specific mechanism (DNA hypermethylation) which is recognized as a mechanism for elimination of TS gene activity; (iii) and the biological read out of this inactivation (a somatic `gain of imprinting' on the maternal allele of H19) is both loss of H19 RNA and, via disruption of a cis-acting regulatory circuit, a reciprocal biallelic activation of IGF2 (somatic `loss of imprinting' of IGF2). Both of these are predicted to be permissive factors for subsequent tumorigenesis, since IGF2 protein can be anti-apoptotic (48) and H19 RNA can be growth suppressive or anti-tumorigenic in some cell types, including 293 transformed human kidney cells (49,50) (L. Yuan and B. Tycko, unpublished data). H19 inactivation and IGF2 biallelic activation in WT precursor cells as the molecular counterpart of WT2 can account for previous genetic data indicating that WT2 is a permissive, rather than rate-limiting, TS locus. Specifically, frank 11p15 LOH is sometimes found in non-neoplastic kidney cells and in peripheral blood of WT patients with unifocal tumors (51). Indeed, an imprinted TS locus is not expected to have a `gatekeeper' function, since if it did there would be an unacceptably high rate of tumor formation. Since the minimal WT2 region remains large, other TS genes may be included in it, but none has yet been reported.
Gene-specific contribution of DNA methylation to normal imprinting of the chromosome 11p15.5/mouse distal chromosome 7 domain
According to our data, H19 is qualitatively different from the other imprinted genes in the chromosome 11p15.5 domain in terms of the extent of allele-specific DNA methylation in normal tissues. In contrast to the complete methylation of >200 restriction sites over 9 kb of DNA upstream and including most of the transcribed region of the H19 gene, allele-specific DNA methylation is not detected by Southern blotting in the 5[prime] regions of IPL, IMPT1, p57KIP2 and KvLQT1 in normal kidney. KvLQT1 does, however, show allele-specific methylation at the site of initiation of an antisense transcript in one of its introns (25,26; this study). These findings raised the possibility that H19 imprinting might be hyper-dependent on DNA methylation. Our analysis of allele-specific mRNA expression in mice with the N-hypomorphic allele of Dnmt1 supports this conclusion, since in that genetic background there is full reactivation of the imprinted H19 allele, partial loss of imprinting of Kvlqt1, but no reactivation of Ipl, Impt1 or p57Kip2. Imprinting highly dependent on DNA methylation correlates with developmental stability of the imprint in somatic cells, since among these genes the H19 imprint is very stable while the imprints at Kvlqt1, Ipl, Impt1 and p57KIP2 are either `leaky' or absent in many adult tissues (9,10,21,29).
Mechanism of H19 epimutation
The mechanism of inactivation of the maternal H19 allele in WT precursor cells is unknown. However, the massive pre-existing allelic asymmetry in DNA methylation at this locus, together with the gene specificity of chromosome 11p15.5 DNA hypermethylation in WTs, specific and perhaps restricted to H19, and the hypersensitivity of H19 allelic expression to DNA methylation, add to the appeal of gene-specific models for H19 inactivation. For example, methylation transfer from the imprinted (hypermethylated) to the non-imprinted allele might occur via heteroduplex formation and action of the nuclear DNA methyltransferase enzyme on the resulting hemimethylated sites (52). In the Ascobolus `methylation-induced premeiotically' (MIP) phenomenon, where interallelic methylation transfer has been proven to occur, greater sequence similarity of the two alleles is associated with higher rates of methylation transfer (53). The model therefore makes a prediction that there might be an excess of H19 homozygotes among WT patients whose tumors show epigenetic H19 silencing.
In our previous series, we observed that constitutional homozygosity at H19 RFLP markers was in fact more frequent in WT patients whose tumors showed H19 epimutation than in WT patients whose tumors showed frank 11p15 LOH (1). In the current series, this is still true. For each patient, an H19 heterozygosity index (HI) can be scored as (number of heterozygous markers/total number of markers examined). Among patients whose tumors show 11p15 LOH, the average HI is 0.51 and among patients whose tumors show retention of heterozygosity with no abnormality of H19 DNA methylation it is 0.49. In contrast, among patients whose tumors show epigenetic silencing of H19 with retention of both alleles, the average HI is 0.21. In addition, all three patients (WT5, WT8 and WT27) with widespread H19 hypermethylation in the non-neoplastic kidney parenchyma were homozygous for all H19 polymorphisms tested. These trends are intriguing, but since this is only a single data set they do not allow a strong conclusion. The model also raises the question of whether genetic syndromes with high rates of homologous recombination might predispose to WT, and in fact there have been four reported cases of WTs in Bloom syndrome (54,55).
H19 as a model for gene-specific epigenetic silencing preceding tumorigenesis
Inactivation of H19 in WT is now among the best-characterized examples of epigenetic gene silencing in cancer. A number of TS genes are subject to this phenomenon, and these genes are reactivated easily with demethylating agents (for H19, see refs 21,34; for other examples, see refs 56,57). In spite of this, there has been persistent scepticism concerning the biological importance of epigenetic silencing in cancer. This consists of doubts as to whether the silencing event is primary, as opposed to a secondary consequence of the neoplastic state, and as to whether the silencing is locus specific. Most studies of DNA methylation have indeed been restricted to late-stage tumors, and very few have assessed DNA methylation at multiple linked genes. This deficit is slowly being filled: for example, a recent study from Belinsky et al. (58) used methylation-sensitive PCR to detect p16/INK4a hypermethylation in pre-neoplastic bronchial epithelial hyperplasia. The current study has addressed these issues by extending the evidence that H19 inactivation is a very early event in the kidneys of WT patients, often preceding the overt neoplastic phenotype, and by showing in an analysis of multiple linked genes that silencing of H19 by DNA hypermethylation is locus specific.
MATERIALS AND METHODS
Southern and northern blotting
Genomic DNAs and total RNA were prepared by standard proteinase K/phenol and Trizol reagent (Gibco BRL, Gaithersburg, MD) methods, respectively. For Southern blotting, 4 µg of genomic DNA was digested overnight with 25 U of the indicated restriction enzymes. Electrophoresis was on 1.0% agarose gels. Hybridization was at 42.5°C in 50% formamide-containing solution (Hybrisol-I; Oncor, Gaithersburg, MD). All of the radiolabeled probes were denatured in the presence of 100 µg of autoclaved salmon sperm DNA prior to hybridization, and the KvLQT1 probes were quenched with Cot1 DNA (Gibco BRL) for 15 min at 65°C after denaturation. For northern blotting, 10 µg of denatured total RNA was electrophoresed on formaldehyde-containing 1.0% agarose gels. With the exception of the H19 5[prime] probe, all probes were generated by PCR using primers based on sequences in the GenBank database. Probes for northern blots were as follows: G3PDH, a 0.5 kb RT-PCR product generated using primers described below (semi-quantitative RT-PCR); H19, a 1 kb genomic PCR product generated from H19 exon 1 (1); ZNF195, a 2 kb RT-PCR product generated with primers ZNF 5[prime], 2 (CCA GAA GTG AAA CGC CAG G) and ZNF 3[prime], 2 (CTC TTG TGT GCT CTG GAG AC); and TAPA1, a 1.2 kb RT-PCR product generated with primers TAPA1 5[prime], 1 (ACC TGC TCT TCG TCT TCA A) and TAPA1 3[prime], 2 (GAC GGA GTC AGG ATG TTG T). Probes for Southern blots were as follows: H19 5[prime] region, a 3 kb genomic phage subclone depicted in Figure 4a; ZNF195 CpG island, a 1.1 kb genomic PCR product corresponding to nucleotides 68 109-69 213 of the PAC clone pDJ1173a5 genomic sequence (GenBank accession no. AC000378); IPL CpG island, a 1 kb PCR product corresponding to nucleotides 108 893-109 893 of a 244 kb chromosome 11p15.5 sequence contig (GenBank accession nos AC001228 and U90582); IMPT1 5[prime] region, a 2.3 kb PCR product corresponding to nucleotides 80 910-83 213 of the same 244 kb sequence contig; and KvLQT1 CpG island, a 1.6 kb genomic PCR product corresponding to nucleotides 58 451-60 091 of the PAC clone DJ915f1 genomic sequence (GenBank accession no. AC003693). For quantitation of mosaicism for the H19 epimutation, densitometry was done by flatbed scanning and analysis with NIH Image software (written by Wayne Rasband, US National Institutes of Health, available from the Internet by anonymous FTP from zippy.nimh.nih.gov ).
LOH analysis
All but one of the chromosome 11 markers used in this study have been described previously by us (1,4,10,21) or are described in public databases (http://www.resgen.com/ ; http://www.marshmed.org/genetics ). The KvLQT1 SmaI RFLP marker has not been described previously and it reflects a C->A nucleotide change located 32 bp downstream of the splice donor site of exon 13. Genomic PCR for this marker was carried out with primers Ke13F (act gtc act gcc tgc act ttg) and Ke13R (ggt tga gag gca aga act c) to generate a 291 bp product which was digested with SmaI and analyzed on 1.5% agarose gels. PCR conditions were an initial denaturation at 94°C for 5 min followed by 30 cycles of denaturation at 94°C for 45 s, annealing at 60°C for 30 s, and extension at 72°C for 1 min, with a final extension for 5 min. For most of the other markers, PCR products were radiolabeled by end-labeling the primers with [[gamma]-32P]ATP and T4 polynucleotide kinase (Promega, Madison, WI). For microsatellite markers, the radiolabeled PCR products were separated on standard denaturing 6% polyacrylamide gels. For RFLP analyses, the PCR products were separated either on non-denaturing acrylamide gels or (for non-radiolabeled products corresponding to the H19 polymorphisms and the IGF2 ApaI RFLP) on 1.2% agarose gels. The EJ-RAS VNTR polymorphism was assessed by Southern blotting of MspI digests.
Semiquantitative RT-PCR
A 2 µg aliquot of total RNA was annealed with oligo(dT) primer and reverse transcription was carried out with MLV reverse transcriptase (Superscript; Gibco BRL). Gene-specific primers for PCR were as follows: G3PDH upstream (acc aca gtc cat gcc atc a), downstream (tcc acc acc ctg ttg ctg t); H19 upstream (aac acc tta ggc tgg tgg g), downstream (tcg gag ctt cca gac t); IPL upstream (cac cct ccg agc cct cgg a), downstream (agt tct tct gct gca ggg); IMPT1 upstream (ggc tgt ctc cac ctc gga c), downstream (agt tta ttg cca gtc tgt g); and KvLQT1 upstream (agc aag cgc gga agc ctt ac), downstream (ggc ctc ccc tct cac tca gg). PCRs used standard total primer concentrations of 6 ng/µl. PCR conditions were as follows: G3PDH/H19 primer ratio (1:1), initial denaturation at 94°C for 4 min followed by 10 cycles of denaturation at 94°C for 1.3 min, annealing at 60°C for 1 min, and extension at 72°C for 1.3 min, and then followed by 18 cycles of denaturation at 94°C for 1.3 min, annealing at 56°C for 1 min, and extension at 72°C for 1.5 min; G3PDH/IPL primer ratio (1:6.7), initial denaturation at 94°C for 5 min followed by 30 cycles of denaturation at 94°C for 45 s, annealing at 61°C for 30 s, and extension at 72°C for 1 min; G3PDH/IMPT1 primer ratio (1:10), initial denaturation at 94°C for 4min followed by 27 cycles of denaturation at 94°C for 1 min, annealing at 56°C for 45 s, and extension at 72°C for 45 s; G3PDH/KvLQT1 primer ratio (1:10), initial denaturation at 94°C for 4 min followed by 30 cycles of denaturation at 94°C for 1 min, annealing at 56°C for 45 s, and extension at 72°C for 1 min. All PCRs were completed by a final extension at 72°C for 7-10 min. All PCRs were done in standard PCR buffer (Boehringer Mannheim/Roche, Indianapolis, IN), supplemented with dimethyl sulfoxide (DMSO) at 5% (v/v). The ratios of control (G3PDH) to experimental signals were then determined by phosphorimaging (Storm System; Molecular Dynamics, Sunnyvale, CA) and analysis with ImageQuant software (Molecular Dynamics). For each data point in each experiment, the results of two cDNA dilutions (two successive lanes on the blot) in the non-saturated range of the PCR were averaged.
Mouse embryo DNA, RNA and reverse transcription
Total RNAs were prepared from individual embryos by extraction with Trizol. For genotyping, a small fragment of fetal tissue was separated at the time of dissection and was used for DNA preparation by proteinase-K/SDS lysis and phenol-chloroform extraction. Reverse transcription of ~0.5 µg of RNA from each conceptus was done with oligo(dT) priming and Supercript MuLV reverse transcriptase.
Allelic expression analysis
PCRs used a standard cycling program with an initial denaturation at 94°C for 4 min, followed by 30 cycles of annealing at 60°C for 45 s, extension at 72°C for 1 min and denaturation at 94°C for 1 min, with a final extension at 72°C for 7 min. The standard PCR buffer (Boehringer Mannheim/Roche) contained dNTPs at 0.1 mM and was supplemented with 5% (v/v) DMSO. Either standard Taq polymerase or long-template optimized Taq polymerase (Expand; Boehringer Mannheim/Roche) were used, as indicated below. For radiolabeling, aliquots of gel-isolated PCR products were subjected to an additional eight cycles of PCR in the same reaction mixture with dNTPs reduced 1:50 and supplemented with 6 µCi/ml [32P]dCTP. SSCP was done essentially according to the original protocol (59), in all cases at room temperature, with the specific conditions given below. The primers for each gene and the conditions for allelic analysis are also given below.
PCR primers for amplification of cDNAs were mH19-FOR (TGA ATC AAG AAG ATG CTG CA) and mH19-REV (CCT GGT GAG GAG GGG CAA AG). For SSCP analysis, the radiolabeled products were digested with AluI, diluted in SDS/EDTA (0.1%/10 mM), denatured and electrophoresed at 300 V overnight on non-denaturing 6% acrylamide gels.
cDNA PCR with primers mKvlqt1-FOR (GAT CAC CAC CCT GTA CAT TG) and mKvlqt1-REV (CCA GGA CTC ATC CCA TTA TC). The radiolabeled products were digested with PvuII for RFLP analysis on non-denaturing 6% acrylamide gels. For SSCP, the radiolabeled PCR products were digested with HinfI, diluted in SDS/EDTA (0.1%/10 mM), denatured and electrophoresed at 500 V for 6 h on non-denaturing 6% acrylamide gels at room temperature. For measuring Kvlqt1 mRNA levels by semi-quantitative RT-PCR, single-tube PCRs were carried out for a total of 30 cycles using a reaction mixture containing Kvlqt1 and Rpl23 primers in a 1:1 ratio. The products were blotted and the membrane was hybridized with a Kvlqt1/Rpl23 probe consisting of a 1:1 mixture of the gel-isolated RT-PCR products. Phosphorimaging was done (Storm System; Molecular Dynamics) and analysis was carried out with ImageQuant software (Molecular Dynamics).
PCR primers for amplification of cDNAs were mKip2-FOR (ATG GAG GTG GAC AGC GAG TC) and mKip2-REV (GAG AGA GGC TGG TCC TTC AG). For SSCP analysis, the radiolabeled PCR products were digested with BglII, diluted in SDS/EDTA (0.1%/10 mM), denatured and electrophoresed at 300 V overnight on non-denaturing 6% acrylamide gels.
The PCR primers for amplification of cDNAs have been described previously (10). The radiolabeled products were digested with KspI for RFLP analysis on non-denaturing 6% acrylamide gels (10).
The primers for amplifying the mouse Ipl cDNAs were described previously, and long-Taqpolymerase was used (9). Allele-specific oligonucleotides were Ipl-A (TTC CAG GTA TGG AAG AA) and Ipl-C (TTC CAG GTC TGG AAG AA). The loading control was the forward primer used in the original PCR. Oligonucleotides were labeled using T4 kinase and [[gamma]-32P]ATP. Slot blots were pre-hybridized and hybridized at 42°C in a solution containing 6× SSC/0.5% SDS/1% dried milk/3% formamide/10 mM sodium pyrophosphate, and were washed in 6× SSC, twice at room temperature for 5 min and once at 47°C for 15 min.
ACKNOWLEDGEMENT
This work was supported by grants to B.T. and T.H.B. from the NIH.
REFERENCES
*To whom correspondence should be addressed. Tel: +1 212 305 1287; Fax: +1 212 305 5498; Email: bt12{at}columbia.edu
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:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Gabory, M.-A. Ripoche, A. Le Digarcher, F. Watrin, A. Ziyyat, T. Forne, H. Jammes, J. F. X. Ainscough, M. A. Surani, L. Journot, et al. H19 acts as a trans regulator of the imprinted gene network controlling growth in mice Development, October 15, 2009; 136(20): 3413 - 3421. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Yoshimizu, A. Miroglio, M.-A. Ripoche, A. Gabory, M. Vernucci, A. Riccio, S. Colnot, C. Godard, B. Terris, H. Jammes, et al. The H19 locus acts in vivo as a tumor suppressor PNAS, August 26, 2008; 105(34): 12417 - 12422. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. W. Brown, F. Power, B. Moore, A. K. Charles, and K. T.A. Malik Frequency and Timing of Loss of Imprinting at 11p13 and 11p15 in Wilms' Tumor Development Mol. Cancer Res., July 1, 2008; 6(7): 1114 - 1123. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. T. Bjornsson, L. J. Brown, M. D. Fallin, M. A. Rongione, M. Bibikova, E. Wickham, J.-B. Fan, and A. P. Feinberg Epigenetic Specificity of Loss of Imprinting of the IGF2 Gene in Wilms Tumors J Natl Cancer Inst, August 15, 2007; 99(16): 1270 - 1273. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. C. Blackburn, L. Z. Hill, A. L. Roberts, J. Wang, D. Aud, J. Jung, T. Nikolcheva, J. Allard, G. Peltz, C. N. Otis, et al. Genetic Mapping in Mice Identifies DMBT1 as a Candidate Modifier of Mammary Tumors and Breast Cancer Risk Am. J. Pathol., June 1, 2007; 170(6): 2030 - 2041. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Monk, P. Arnaud, S. Apostolidou, F. A. Hills, G. Kelsey, P. Stanier, R. Feil, and G. E. Moore From the Cover: Limited evolutionary conservation of imprinting in the human placenta PNAS, April 25, 2006; 103(17): 6623 - 6628. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Yuan, C.-M. Li, D. J. Yamashiro, J. Kandel, H. Thaker, V. V. Murty, and B. Tycko Genomic Profiling Maps Loss of Heterozygosity and Defines the Timing and Stage Dependence of Epigenetic and Genetic Events in Wilms' Tumors Mol. Cancer Res., September 1, 2005; 3(9): 493 - 502. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. A. Melnikov, R. B. Gartenhaus, A. S. Levenson, N. A. Motchoulskaia, and V. V. Levenson (Chernokhvostov) MSRE-PCR for analysis of gene-specific DNA methylation Nucleic Acids Res., June 8, 2005; 33(10): e93 - e93. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. G. Alleman, R. L. Tabios, G. V. R. Chandramouli, O. N. Aprelikova, C. Torres-Cabala, A. Mendoza, C. Rodgers, N. A. Sopko, W. M. Linehan, and J. R. Vasselli The In vitro and In vivo Effects of Re-Expressing Methylated von Hippel-Lindau Tumor Suppressor Gene in Clear Cell Renal Carcinoma with 5-Aza-2'-deoxycytidine Clin. Cancer Res., October 15, 2004; 10(20): 7011 - 7021. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Sandovici, M. Leppert, P. R. Hawk, A. Suarez, Y. Linares, and C. Sapienza Familial aggregation of abnormal methylation of parental alleles at the IGF2/H19 and IGF2R differentially methylated regions Hum. Mol. Genet., July 1, 2003; 12(13): 1569 - 1578. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Weksberg, A. C. Smith, J. Squire, and P. Sadowski Beckwith-Wiedemann syndrome demonstrates a role for epigenetic control of normal development Hum. Mol. Genet., April 2, 2003; 12(90001): R61 - 68. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Satoh, T. Nakagawachi, H. Nakadate, Y. Kaneko, Z. Masaki, T. Mukai, and H. Soejima Significant Reduction of WT1 Gene Expression, Possibly Due to Epigenetic Alteration in Wilms' Tumor J. Biochem., March 1, 2003; 133(3): 303 - 308. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Ishizaki, M. Yoshie, Y. Yaginuma, T. Tanaka, and K. Ogawa Loss of Igf2 Imprinting in Monoclonal Mouse Hepatic Tumor Cells Is Not Associated with Abnormal Methylation Patterns for the H19, Igf2, and Kvlqt1 Differentially Methylated Regions J. Biol. Chem., February 14, 2003; 278(8): 6222 - 6228. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Goldberg, M. Wei, B. Tycko, I. Falikovich, and D. Warburton Identification and expression analysis of the human {micro}-protocadherin gene in fetal and adult kidneys Am J Physiol Renal Physiol, September 1, 2002; 283(3): F454 - F463. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Weksberg, J. Nishikawa, O. Caluseriu, Y.-L. Fei, C. Shuman, C. Wei, L. Steele, J. Cameron, A. Smith, I. Ambus, et al. Tumor development in the Beckwith-Wiedemann syndrome is associated with a variety of constitutional molecular 11p15 alterations including imprinting defects of KCNQ1OT1 Hum. Mol. Genet., December 1, 2001; 10(26): 2989 - 3000. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Bliek, S. M. Maas, J. M. Ruijter, R. C.M. Hennekam, M. Alders, A. Westerveld, and M. M.A.M. Mannens Increased tumour risk for BWS patients correlates with aberrant H19 and not KCNQ1OT1 methylation: occurrence of KCNQ1OT1 hypomethylation in familial cases of BWS Hum. Mol. Genet., March 1, 2001; 10(5): 467 - 476. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. H. Bestor The DNA methyltransferases of mammals Hum. Mol. Genet., October 1, 2000; 9(16): 2395 - 2402. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. Hartmann, A. Waha, A. Koch, C. G. Goodyer, S. Albrecht, D. von Schweinitz, and T. Pietsch p57KIP2 Is Not Mutated in Hepatoblastoma but Shows Increased Transcriptional Activity in a Comparative Analysis of the Three Imprinted Genes p57KIP2, IGF2, and H19 Am. J. Pathol., October 1, 2000; 157(4): 1393 - 1403. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S.-i. Horike, K. Mitsuya, M. Meguro, N. Kotobuki, A. Kashiwagi, T. Notsu, T. C. Schulz, Y. Shirayoshi, and M. Oshimura Targeted disruption of the human LIT1 locus defines a putative imprinting control element playing an essential role in Beckwith-Wiedemann syndrome Hum. Mol. Genet., September 1, 2000; 9(14): 2075 - 2083. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. S. Doherty, M. R.W. Mann, K. D. Tremblay, M. S. Bartolomei, and R. M. Schultz Differential Effects of Culture on Imprinted H19 Expression in the Preimplantation Mouse Embryo Biol Reprod, June 1, 2000; 62(6): 1526 - 1535. [Abstract] [Full Text]
|
|
|
|
|
|
V. Juan, C. Crain, and C. Wilson Evidence for evolutionarily conserved secondary structure in the H19 tumor suppressor RNA Nucleic Acids Res., March 1, 2000; 28(5): 1221 - 1227. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Prawitt, T. Enklaar, G. Klemm, B. Gartner, C. Spangenberg, A. Winterpacht, M. Higgins, J. Pelletier, and B. Zabel Identification and characterization of MTR1, a novel gene with homology to melastatin (MLSN1) and the trp gene family located in the BWS-WT2 critical region on chromosome 11p15.5 and showing allele-specific expression Hum. Mol. Genet., January 22, 2000; 9(2): 203 - 216. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||