| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Medical Service, Veterans Affairs Palo Alto Health Care System, and Department of Medicine, Stanford University, Palo Alto, California 94304
Address all correspondence and requests for reprints to: Andrew Hoffman, Medical Service, Veterans Affairs Palo Alto Medical Center, 3801 Miranda Avenue, Palo Alto, California 94304. E-mail: arhoffman{at}stanford.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Loss of IGF2 or H19 imprinting commonly occurs in tumors and has been implicated in tumorigenesis (5, 6, 7, 8). The importance of IGF2 imprinting has recently been expanded, because it may provide a new biomolecular marker for cancer risk (9).
Extensive research on IGF2/H19 imprinting has revealed complicated regulatory mechanisms including multiple enhancer, silencer, and boundary elements, many of which are specific for different tissues (reviewed in Ref. 10). Some elements are within the IGF2 gene. For example, both the Reeve (11) and Feinberg (12) laboratories have identified separate regions upstream of IGF2 exon 3 in which loss of allele-specific methylation correlates with loss of IGF2 imprinting in Wilms’ tumors and colon cancer, respectively.
Strong evidence has been presented that IGF2/H19 imprinting depends upon an imprinting control region (ICR) upstream of H19 (13, 14). The ICR element displays parent-of-origin-dependent methylation in both mice and humans (15, 16, 17) and contains binding sites for the zinc-finger CCCTC-binding factor, known as CTCF. The mouse ICR contains four CTCF binding sites (18, 19, 20, 21), whereas the human ICR contains seven (19). In the mouse, at least the first three CTCF-binding sites are differentially methylated as evaluated by allele-specific PCR and methylation analysis (21A ). In the human, the sixth CTCF-binding site has been demonstrated to have allele-specific differential methylation (22), and the loss of differential methylation at this site correlates with loss of imprinting (LOI) in Wilms’ tumors (23), bladder cancer (22), colon cancer (24, 25), and osteosarcoma (26).
The human ICR is unable to function when introduced as a transgene into mice (27), demonstrating that the human and mouse ICRs are not interchangeable; however, they appear to share a similar mechanism of action. A current model hypothesizes that the maternally inherited allele maintains an unmethylated ICR, which is a suitable binding site for CTCF. CTCF binding establishes an insulator between IGF2 and downstream enhancers, which prevents IGF2 transcription and supports H19 transcription. Reciprocally, the paternally inherited allele maintains a methylated ICR, which prevents CTCF binding, leading to H19 transcription and IGF2 silencing. This insulator mechanism has been demonstrated both in vitro (18, 19) and in mouse models (28, 29).
The insulator model predicts that if the ICR elements of both alleles were unmethylated, then H19 would be biallelically expressed, whereas IGF2 is silenced. This model also predicts that if both ICR elements were methylated, then H19 would be silenced, whereas IGF2 is biallelically expressed. As predicted, this pattern of expression is seen in Wilms’ tumors (30, 31, 32). However, there are many examples in which expression of one gene is biallelic, with continued monoallelic expression of the other gene (26, 33, 34, 35), a scenario that is unexpected based on this model. We have proposed a modification to the insulator model (26), in which incomplete methylation changes in the ICR lead to partial insulation by CTCF, allowing for the various IGF2/H19 expression patterns observed in different tissues. If the insulator model were true, then we would expect to be able to discern predictable patterns of CTCF binding at the parental ICR elements based on observed IGF2/H19 expression.
This study tested whether CTCF binding to parental ICR elements accounts for the expression of IGF2 and H19 in normal human development and in osteosarcoma tumors. Both normal human tissues and osteosarcoma tumors provided excellent opportunities to study the relationship between CTCF binding at the ICR and IGF2/H19 expression, because both contain subsets of tissues in which expression is monoallelic and biallelic. We determined IGF2/H19 expression by RT-PCR, ICR methylation by bisulfite sequencing, and CTCF-binding at the ICR by chromatin immunoprecipitation (ChIP) and PCR. We show that CTCF consistently binds unmethylated ICR elements, but that this binding does not necessarily lead to regulation of IGF2/H19 expression in human tissues.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Fetal tissues were obtained from the Cooperative Human Tissue Network.
Nucleic acid collection
Genomic DNA was collected from tissues, using a solution of 4 M guanidinium thiocyanate, 25 nM sodium citrate, 1% 2-mercaptoethanol, and 0.5% Sarkosyl as described previously (7). The homogenate was extracted with phenol/chloroform and then precipitated with 2-propanol. The DNA pellet was washed with ethanol and dissolved in distilled water. RNA was collected from tissue using Tri-reagent (Sigma, St. Louis, MO).
Genotyping IGF2 and H19 polymorphisms
Genotypes for fetal tissues were determined by PCR of genomic DNA. Each PCR was performed in 6 µl volume under liquid wax. Reactions contained 400 ng DNA, 0.1 µM appropriate primers, 50 µM deoxynucleoside triphosphate, and 0.4 U Klen Taq I (Ab Peptides, St. Louis, MO). Primers for analysis of each polymorphism are listed in Table 1
. One primer from each pair was end-labeled with 
-32P ATP (Fig. 1A). PCR conditions were 95 C for 60 sec, followed by 29 cycles of 95 C for 25 sec, 66 C for 30 sec, and 72 C for 45 sec.
|
|
Products were electrophoresed on 5% polyacrylamide-urea gel and visualized by a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Expression of IGF2 and H19 alleles
Samples with heterozygous genotypes were analyzed for allelic expression by RT-PCR. RNA samples were treated with DNase I (Invitrogen) in the presence of RNase inhibitor (Invitrogen) to eliminate contaminating genomic DNA. Then, 0.4 µg of total RNA, random hexamer, and Moloney murine leukemia virus reverse transcriptase (Invitrogen) were used for cDNA production. cDNA samples were PCR amplified with primers and conditions used for genotyping, but for only 27 cycles. Products were electrophoresed on 5% polyacrylamide-urea gel and visualized by a PhosphorImager.
Bisulfite treatment of genomic DNA
Bisulfite treatment of genomic DNA efficiently converts unmethylated cytosines to uracil, whereas 5-methylcytosine remains unchanged (37). Two micrograms of genomic DNA were denatured in 20 µl 0.3 M NaOH for 20 min at 37 C and then placed on ice. A total of 220 µl of 3.5 M sodium bisulfite containing 1 mM hydroquinone was added, and the solution was covered with liquid wax. The solution was incubated at 0 C for 12 h, then 50 C for 8 h. Resulting bisulfite-treated DNA was purified using QIAEX II Extraction Kit (Qiagen, Valencia, CA) and Centri-Spin20 Columns (Princeton Separations, Adelphia, NJ). From 50 µl final volume in water, 2 µl was used for each subsequent analysis.
Cloning and sequencing of bisulfite-treated DNA
To determine the methylation status of cytosines in genomic DNA, DNA was bisulfite treated, amplified by PCR (see Table 1
for primers), cloned, and sequenced. Two microliters of bisulfite-treated DNA were PCR amplified in a total volume of 6 µl covered with liquid wax. One primer from each pair was end-labeled with -32P ATP. PCR conditions were 95 C for 60 sec, followed by 30 cycles of 95 C for 20 sec, 61 C for 30 sec, and 70 C for 90 sec, and finally 70 C for 10 min. PCR products were resolved on 5% acrylamide gels. Small pieces of gel containing the desired product were cut out and eluted in 100 µl water for 10 min at 99 C. One microliter of this elution was subjected to a second round of PCR with similar conditions, except with nonlabeled primers and only 20 cycles. One-microliter samples of the second amplification were resolved on 1% agarose gels with ethidium bromide to verify that products were of the anticipated size. Successful PCR products were then cloned into the pCR2.1-TOPO vector (Invitrogen) and transformed into TOP10 One Shot Escherichia coli (Invitrogen), and plasmid DNA was collected by QIAprep Spin Mini-prep kit (Qiagen). Automated sequencing of DNA was performed using Big-Dye (Perkin-Elmer, Wellesley, MA).
ChIP
A total of 150 mg of each tissue was minced with forceps and a razor blade and homogenized with a bounce homogenizer. Tissue that was not fully homogenized was removed, and the remaining homogenate was fixed in 10 ml of PBS containing 1% formaldehyde. After 12 min in PBS/1% formaldehyde, the fixation process was stopped by addition of 0.5 ml of 2.5 M glycine. Fixed tissue was washed twice in ice-cold PBS with protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 µg/ml aprotinin, and 1 µg/ml pepstatin), then incubated in ice-cold lysis solution (1% sodium dodecyl sulfate and 10 nM EDTA with protease inhibitors) for 10 min. Fixed tissue was sonicated with a Branson 250 Sonifier set at 40% output and 90% duty cycle. Samples of sonicated tissue were removed after 10, 15, 20, and 25 cycles of 10 sonication pulses; DNA was purified with MinElute Purification Kit (Qiagen); and DNA was visualized by agarose electrophoresis. Samples with DNA sheared to between 100 and 1,200 bp (typically requiring 15–20 sonication cycles) were used for subsequent immunoprecipitation reactions.
Separate ChIP reactions were performed to isolate nucleosomes bound to CTCF, dimethyl histone 3 lysine (H3K)4, and dimethyl H3K9 using antibodies (Upstate, Lake Placid, NY; nos. 06-917, 07-030, and 07-212) and protocols from Upstate, Inc. (found at www.upstate.com). After ChIP, immune complexes were collected with Protein A Agarose beads (Upstate), washed, and eluted according to Upstate protocols, and fixation was reversed by addition of NaCl to a final concentration of 0.2 M and incubation at 65 C overnight. ChIP-enriched DNA was purified with MinElute Purification Kit (Qiagen). Eluted DNA was amplified by PCR with primers specific for the IGF2/H19 ICR region (Table 1). One primer was end-labeled with -32P ATP. PCR conditions were 95 C for 60 sec, followed by 35 cycles of 95 C for 20 sec, 66 C for 30 sec, and 72 C for 45 sec. One microliter of PCR products was digested with 4 U of DraIII enzyme in a total volume of 10 µl. Digested products were resolved on 5% acrylamide gels, then visualized and quantified by a PhosphorImager.
Reamplification of ChIP PCR products
Small pieces of acrylamide gel containing the desired PCR products were cut out and eluted in 100 µl water for 10 min at 99 C. One microliter of this elution was subjected to a second round of PCR with similar conditions, except for only 20 cycles. One microliter of PCR products was digested with 2 U of NlaIII enzyme in a total volume of 10 µl. Undigested and digested products were resolved on 5% acrylamide gels, then visualized and quantified by a PhosphorImager.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. Two polymorphisms involve the creation of a restriction enzyme site, the other involves a 19-bp deletion in exon 9 of IGF2. PCR of genomic DNA with specific primers (Table 1), followed by restriction digestion when appropriate, allowed for the determination of informative samples (Fig. 1B). RT-PCR of total RNA allowed for analysis of expressed alleles. Note that in both fetuses, imprinting of IGF2 and H19 is lost in the brain, while it is maintained in liver and kidney.
Methylation of the IGF2/H19 ICR in fetal tissues
The methylation status of the IGF2/H19 ICR in fetal tissues was determined by cloning and sequencing of bisulfite-treated DNA. After bisulfite treatment of DNA, the primer pair used amplifies a 452-bp segment containing 26 CpG dinucleotides, five of which are contained within a CTCF-binding site (Fig. 2A). This is the sixth CTCF-binding site within the human IGF2/H19 ICR (18, 19). The parental alleles can be distinguished by the presence of two single nucleotide polymorphisms. In all fetal tissues examined, bisulfite sequencing shows that one allele is methylated, whereas the other is unmethylated, particularly within the CTCF-binding site (Fig. 2B).
|
). All tissues used in this study are informative for this polymorphism.
|
, input row) results in nearly equal visualization of the two parental alleles. Enrichment of DNA by anti-CTCF antibodies was highly selective for one allele (Fig. 3B, CTCF row).
To confirm that the enriched allele was the unmethylated allele, we eluted the anti-CTCF-enriched PCR products from fetus 1, re-PCR amplified them, and digested them with NlaIII. As can be seen in Fig. 2A, bisulfite DNA sequencing of fetus 1 revealed an A/G polymorphism at base 6325. The allele containing the A polymorphism was methylated, whereas the allele containing the G polymorphism was unmethylated. The G polymorphism creates an additional NlaIII site at base 6325 within our amplified fragment (Fig. 4A). Because the anti-CTCF-enriched PCR products were cut at this NlaIII site (Fig. 4B), we can show that ChIP with anti-CTCF antibodies selectively enriched the unmethylated allele.
|
, H3K4 row). Antidimethyl H3K9 antibodies enriched the opposite allele, except in brain where H3K9 was equally present on both alleles (Fig. 3B, H3K9 row).
CTCF-binding at the IGF2/H19 ICR in osteosarcoma
In a previous manuscript (26), we noted that IGF2 or H19 imprinting was lost in a subset of osteosarcomas, the most common primary bone tumor. Tumors with H19 LOI displayed IGF2 maintenance of imprinting (MOI) and hypomethylation of the sixth CTCF-binding site on both alleles. Tumors with IGF2 LOI displayed H19 MOI and biallelic methylation of this CTCF-binding site. Tumors that maintained imprinting of both genes contained one methylated and one unmethylated allele in this region. Thus, osteosarcomas provided an additional opportunity to examine the relationship between IGF2/H19 imprinting, ICR methylation, and CTCF-binding of the ICR.
PCR of genomic osteosarcoma DNA before enrichment with anti-CTCF antibody results in nearly equal visualization of the two parental alleles (Fig. 3C, input row). In two osteosarcomas with IGF2/H19 MOI, anti-CTCF antibody enriched one allele (Fig. 3C, CTCF row). There was a dramatic decrease in visualization of both alleles in tumors with IGF2 LOI after anti-CTCF enrichment. Both alleles continued to be visualized proportionally in H19 LOI tumors after anti-CTCF enrichment.
As with the fetal samples above, to prove that the CTCF-enriched product from osteosarcoma no. 1 was the unmethylated allele, this band was eluted, reamplified, and NlaIII digested. The methylated allele of this tumor has the G polymorphism at 6325, whereas the unmethylated allele has the A polymorphism (26). NlaIII failed to digest at this site, showing that this is the allele with the A polymorphism. Thus, the anti-CTCF antibody again enriched the unmethylated allele.
Histone 3 methylation at the IGF2/H19 ICR in osteosarcoma
In osteosarcoma tumors with IGF2/H19 MOI, ChIP reactions targeting antidimethyl H3K4 enriched the same allele as anti-CTCF, whereas antidimethyl H3K9 enriched the opposite allele (Fig. 3C). However, in tissues with IGF2 or H19 LOI, this pattern was abrogated.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We have previously shown that altered methylation of CTCF-binding site 6 in the ICR reliably predicts IGF2/H19 MOI or LOI in osteosarcomas (26). In the present manuscript, we used ChIP to evaluate whether CTCF binding at this site is predictive of IGF2/H19 imprinting, as proposed in the CTCF-boundary model. Because this model has provided a significant advance in our understanding of IGF2/H19 imprinting in mice, identifying the strengths and weaknesses of this model are of critical importance. A weakness of the model exposed in this study is that CTCF binding in the ICR did not necessarily dictate the imprinting status of IGF2 and H19 in human tissues.
In brain tissue, it has previously been noted that IGF2 and/or H19 expression in specialized central nervous system (CNS) structures is biallelic (39, 40). Our results confirm biallelic expression in the brain and show that biallelic expression occurs despite similar patterns of CTCF binding at the ICR in CNS and non-CNS tissues. Thus, LOI in the brain occurs despite the presence of the CTCF boundary. One potential explanation for this would be that brain-specific enhancers may exist upstream of the ICR, where they would not be blocked by a CTCF boundary (41). Another explanation would be that CTCF is not by itself sufficient to establish the boundary, and that the required additional factors are absent in the fetal brain.
Our osteosarcoma tissues provided a series of tissues in which IGF2/H19 imprinting status and ICR methylation were already defined (26). Subsets of tumors were available with MOI and LOI, allowing for comparisons with CTCF binding. In osteosarcomas with MOI, there is allele-specific methylation of the ICR and allele-specific CTCF binding, as predicted in the CTCF boundary model. Noting this, we expected to see CTCF-binding patterns in tumors with LOI that would explain their IGF2/H19 expression. Altered ICR methylation did appear to cause altered CTCF binding in tumors with LOI. When ICR elements on both alleles are unmethylated, as in tumors with H19 LOI, CTCF binds to both alleles. And when ICR elements on both alleles are methylated, as in tumors with IGF2 LOI, CTCF binding is absent. However, altered CTCF binding did not explain IGF2/H19 expression. For example, IGF2 is still expressed in tumors with biallelic CTCF binding. This is contrary to the predictions of the CTCF-boundary model.
We also used ChIP to evaluate histone modifications in the ICR. In tissues with IGF2/H19 MOI, dimethyl H3K4 is associated with the unmethylated allele, whereas dimethyl H3K9 is associated with the methylated allele. Because the ICR element is 2 Kb upstream of the H19 transcription start site, this may reflect the local histone state near the H19 promoter. H3K9 methylation of the methylated ICR would be expected to recruit histone deacetylase complexes and repress the expression of the nearby H19 gene (38). H3K4 methylation of this region would be expected to prevent histone deacetylase complex recruitment and promote H19 expression (38). This pattern is lost in osteosarcomas with LOI and may represent disruption of epigenetic mechanisms in these tumors.
In summary, ChIP and PCR demonstrated that CTCF consistently binds unmethylated ICR elements in fetal tissues and osteosarcoma tumors; however, CTCF binding did not result in IGF2/H19 expression patterns predicted by the CTCF-boundary model. These data suggest that CTCF binding at the ICR is by itself not sufficient to regulate IGF2/H19 imprinting in many human tissues.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Abbreviations: ChIP, Chromatin immunoprecipitation; CNS, central nervous system; CTCF, CCCTC-binding factor; H3K, histone 3 lysine; ICR, imprinting control region; LOI, loss of imprinting; MOI, maintenance of imprinting; OS, osteosarcoma.
Received May 30, 2003.
Accepted for publication July 1, 2003.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  X. Qiu, T. H. Vu, Q. Lu, J. Q. Ling, T. Li, A. Hou, S. K. Wang, H. L. Chen, J. F. Hu, and A. R. Hoffman A Complex Deoxyribonucleic Acid Looping Configuration Associated with the Silencing of the Maternal Igf2 Allele Mol. Endocrinol., June 1, 2008; 22(6): 1476 - 1488. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Mitalipov, L. Clepper, H. Sritanaudomchai, A. Fujimoto, and D. Wolf Methylation Status of Imprinting Centers for H19/IGF2 and SNURF/SNRPN in Primate Embryonic Stem Cells Stem Cells, March 1, 2007; 25(3): 581 - 588. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Tost, H. Jammes, J.-M. Dupont, C. Buffat, B. Robert, T.-M. Mignot, F. Mondon, B. Carbonne, U. Simeoni, G. Grange, et al. Non-random, individual-specific methylation profiles are present at the sixth CTCF binding site in the human H19/IGF2 imprinting control region Nucleic Acids Res., November 14, 2006; 34(19): 5438 - 5448. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. K. Murphy, Z. Huang, Y. Wen, M. A. Spillman, R. S. Whitaker, L. R. Simel, T. D. Nichols, J. R. Marks, and A. Berchuck Frequent IGF2/H19 Domain Epigenetic Alterations and Elevated IGF2 Expression in Epithelial Ovarian Cancer Mol. Cancer Res., April 1, 2006; 4(4): 283 - 292. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Li, T. H. Vu, G. A. Ulaner, E. Littman, J.-Q. Ling, H.-L. Chen, J.-F. Hu, B. Behr, L. Giudice, and A. R. Hoffman IVF results in de novo DNA methylation and histone methylation at an Igf2-H19 imprinting epigenetic switch Mol. Hum. Reprod., September 1, 2005; 11(9): 631 - 640. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. X. Fu, S. R. Schwarze, M. L. Kenowski, S. LeBlanc, J. Svaren, and D. F. Jarrard A Loss of Insulin-like Growth Factor-2 Imprinting Is Modulated by CCCTC-binding Factor Down-regulation at Senescence in Human Epithelial Cells J. Biol. Chem., December 10, 2004; 279(50): 52218 - 52226. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Klenova, A. C. Scott, J. Roberts, S. Shamsuddin, E. A. Lovejoy, S. Bergmann, V. J. Bubb, H.-D. Royer, and J. P. Quinn YB-1 and CTCF Differentially Regulate the 5-HTT Polymorphic Intron 2 Enhancer Which Predisposes to a Variety of Neurological Disorders J. Neurosci., June 30, 2004; 24(26): 5966 - 5973. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
