Control of Organ-specific Demethylation by an Element of the T-cell Receptor-α Locus Control Region*

DNA methylation is important for mammalian development and the control of gene expression. Recent data suggest that DNA methylation causes chromatin closure and gene silencing. During development, tissue specifically expressed gene loci become selectively demethylated in the appropriate cell types by poorly understood processes. Locus control regions (LCRs), which are cis-acting elements providing stable, tissue-specific expression to linked transgenes in chromatin, may play a role in tissue-specific DNA demethylation. We studied the methylation status of the LCR for the mouse T-cell receptor α/δ locus using a novel assay for scanning large distances of DNA for methylation sites. Tissue-specific functions of this LCR depend largely on two DNase I-hypersensitive site clusters (HS), HS1 (T-cell receptor α enhancer) and HS1′. We report that these HS induce lymphoid organ-specific DNA demethylation in a region located 3.8 kilobases away with little effect on intervening, methylated DNA. This demethylation is impaired in mice with a germline deletion of the HS1/HS1′ clusters. Using 5′-deletion mutants of a transgenic LCR reporter gene construct, we show that HS1′ can act in the absence of HS1 to direct this tissue-specific DNA demethylation event. Thus, elements of an LCR can control tissue-specific DNA methylation patterns both in transgenes and inside its native locus.

The process of reversible DNA methylation is highly active during mammalian embryogenesis and somatic cell differentiation. Studies of mice deficient in the DNA methylation machinery have shown that this process is required for successful mouse development (1). Mammalian DNA methyltransferase (2) catalyzes the addition of a methyl group at the 5 position of the cytosine ring in the CpG dinucleotide. The recently identified DNA demethylase catalyzes the reversal of this process (3). Differential methylation of genomic loci is implicated in cancerous transformation, genome stability, and tissue-specific gene expression. The effect of DNA hypermethylation has been implicated in the inactivation of tumor suppressor genes (4 -7), and DNA undermethylation has been shown to lead to elevated mutation rates (8). During embryogenesis, a post-fertilization period of DNA demethylation removes the methyl groups from these C residues in the genome, thus erasing the germ cell DNA methylation patterns. Between uterine implantation and gastrulation, methyl groups are added back to the genome. This resets the overall methylation pattern to one that is characteristic of most somatic cells (9,10). Later in development, DNA demethylation mechanisms remove the methyl groups at specific loci, creating tissue-specific DNA methylation patterns which are proposed to affect gene expression (10,11).
DNA methylation is thought to regulate gene expression by three possible mechanisms. The first is direct interference with the binding of transcription factors by alteration of C nucleotides in their recognition sequences (12)(13)(14). The second mechanism involves the binding of specific factors to methylated DNA which then block the binding of other factors required for gene induction (15)(16)(17). The third mechanism proposes that DNA methylation indirectly regulates gene expression via alteration of chromatin structure (11,18). Recent studies have indicated that the process of histone deacetylation may link the latter two mechanisms. Factors that specifically bind to methylated DNA have been reported to recruit histone deacetylases which act to close chromatin structure, rendering genes inaccessible to transcriptional machinery (19,20). Thus, the processes late in development that allow the appropriate cell typespecific demethylation of tissue specifically expressed gene loci are important for preventing these genes from being silenced by global repression mechanisms (18,21). However, only a small handful of tissue-specific demethylation phenomena have been studied (22,23). Although this process is still poorly understood, evidence from these studies indicates that it may depend on the action of tissue-specific transcriptional control elements. We hypothesized that the action of the locus control region (LCR), 1 which is thought to regulate chromatin structure (24 -26), may be involved in establishing tissue-specific DNA methylation patterns. An LCR forms a tissue-specific open chromatin domain for a linked gene regardless of its position in the genome. These LCR induced chromatin changes could aid in directing DNA methylation and demethylation enzyme complexes to specific locations at specific stages in development.
We have studied the LCR for the T cell receptor (TCR) ␣/␦ locus. This locus contains the multiple gene segments that are rearranged by the V-D-J recombinase complex to create functional TCR␣ and TCR␦ genes and proteins. These proteins are subunits of the receptors that the ␣␤ and ␥␦ T-cells, respectively, use to recognize foreign antigens (27). The rearrangement and expression of these genes are limited to T cells, which reside mainly in lymphoid organs such as thymus and spleen. Just downstream of these genes is the Dad1 gene, which in contrast to the TCR genes, is ubiquitously expressed (28).
Thus, strict regulation of the accessibility of this locus in chromatin must be maintained to achieve the differential expression of these juxtaposed genes. The TCR␣ LCR is located downstream of the TCR␣ gene and is 3 kilobases (kb) upstream of the Dad1 exons (Fig. 1A). It contains nine DNase I-hypersensitive sites (HS) (29,30). HS1 maps to the TCR␣ transcriptional enhancer (31). HS1Ј (located less than 1 kb 3Ј of HS1) is an important element involved in tissue-specific functions of the LCR (32). HS2-6 are downstream of HS1Ј while HS7 and 8 lie 5Ј of HS1. This LCR has been found to confer high-level, position-independent, copy number-dependent, and T-cell specific expression to linked transgenes in mice (29,30). Furthermore, functional subdomains of this LCR have been identified which control its tissue specificity and chromatin opening functions (30). The major tissue-specific functions of this LCR have been mapped to HS1 and HS1Ј (32).
We have investigated the DNA methylation status of this LCR using a novel method for scanning large regions of DNA to directly detect methylated CpG dinucleotides. Here we report the identification of a lymphoid organ-specific demethylation event in a localized region of the TCR␣ LCR. Studies in structural gene knockout animals indicate that this demethylation mechanism operates independently of TCR cell surface expression or gene rearrangement. In contrast, germline deletion of the HS1/HS1Ј region of the LCR disrupts this demethylation event which occurs at a site 3.8 kb downstream of HS1Ј. The HS1Ј element of the LCR has previously been shown to have a chromatin based (i.e. non-classical enhancer) activity affecting the tissue distribution of transgene expression (32). Experiments using transgenic mice containing various 5Ј deletion mutants of an LCR reporter construct showed that HS1Ј can act in the absence of HS1 to direct this novel demethylation mechanism. Thus, elements of an LCR can play a role in establishing tissue-specific DNA methylation patterns.

EXPERIMENTAL PROCEDURES
Mice-C57BL/6 mice, Rag1 deficient mice (33), and TCR␣ constant region deficient mice (34) (Jackson Laboratory, Bar Harbor, ME) were used for the study of the endogenous TCR␣ locus. TCR␣ enhancer knockout mice (E␣Ϫ/Ϫ) were described previously and kindly provided by B. P. Sleckman and F. W. Alt, Harvard (35). All immunocompromised mice were housed under strict microisolator conditions. LCR transgenic mice and constructs used in this study were described previously (30,32) and are referred to in the text.
Organ Genomic DNA Preparation-Thymus and spleen were disrupted with frosted microscope slides to obtain single cell suspensions. Kidney and heart were finely diced. Single cells, or organ pieces, were washed in phosphate-buffered saline, resuspended and homogenized in 3 ml of homogenizing buffer (HB) 60 mM KCl, 15 mM NaCl, 15 mM Tris, 5 mM MgCl 2 , 300 mM sucrose, 5% glycerol. 1/10 volume of 5% SDS, 100 mM EDTA was added, and the suspensions were treated with 200 g/ml proteinase K at 55°C overnight. Samples were extracted twice with 1 volume of phenol, and once with 1 volume of 24:1 mixture of chloroform and isoamyl alcohol. After ethanol precipitation, genomic DNA was resuspended in TE (10 mM Tris, 1 mM EDTA) buffer.
McrBC Titration Assay-For the endogenous locus, 50 g of genomic DNA were digested with SacI restriction enzyme, which generates a TCR␣ locus fragment of 17 kb, and resuspended in 55 l of 10 mM Tris (pH 8.0). For the transgenic samples, MfeI restriction enzyme was used to generate an 11-kb parent fragment of the transgene. The DNA samples were then divided into five 10-l aliquots and digested with increasing amounts of the DNA methylation-dependent McrBC restriction enzyme (New England Biolabs) (0 -40 enzyme units) for 15 min at 37°C in 20 l final volume. Samples were run in 0.6% agarose (Life Technologies, Inc.) at 2 volts/cm overnight, and analyzed by Southern blot. To detect the endogenous LCR, the blot was probed with an 800-bp SmaI fragment that detected the 3Ј end of the SacI generated parent fragment. Transgenic LCR blots were probed with an MfeI/EheI fragment that detected the 3Ј end of the parent fragment. These probes were labeled with ␣-[ 32 P]ATP using Klenow DNA polymerase (Roche Molecular Biochemicals) by random oligonucleotide-primed synthesis.
Cleavage with Methylation-sensitive Restriction Enzymes-15 g (for the endogenous locus) or 5 g (for transgenic) genomic DNA were completely digested with MseI restriction enzyme to generate a 1.2-kb parent fragment containing the LCR HS4 region. The LCR HS2 region MseI parent fragment was 700 bp. Samples were ethanol precipitated and resuspended in 10 mM Tris (pH 8.0), and then digested with MspI, HpaII, or HhaI. Samples were run in 1.2% agarose (Life Technologies, Inc.), and analyzed by Southern blot. For analysis of the HS4 region, a 48-bp single-stranded synthetic oligonucleotide probe recognizing the 3Ј end of the HS4 region MseI fragment was used (GCAGCCAAGCAA-CACTGACAGTGGGAAACATTCTTCCCCAGGGAGAAG). This probe was labeled with [␥-32 P]ATP using T4 polynucleotide kinase (New England Biolabs). These blots were washed to a final stringency of 2 ϫ SSC, 0.1% SDS at 65°C. For the analysis of HS2 methylation, the blot was probed with a 200-bp NcoI/AvrII fragment that recognized the 3Ј end of the parent fragment. These blots were washed to a final stringency of 0.2 ϫ SSC, 0.1% SDS at 65°C. The signals from the parent fragments and subfragments on these Southern blots were quantified by Phospho-rImager analysis (Molecular Dynamics). Background signals were obtained for each lane. The background readings were then subtracted from the parent fragment and subfragment signals in the same lane. Percent demethylation was calculated by dividing the subfragment signals by the sum of the parent and subfragment signals. These percentages indicated the extent of demethylation.

RESULTS
In addition to the study of the endogenous TCR␣ LCR, we use a transgenic model for TCR␣ LCR activity which has been previously described (30,32). In this system, a human ␤-globin genomic fragment (24, 36 -38) is linked as a reporter gene to the full-length (all 9 HS) LCR to generate the ␤:1-8 transgene (Fig. 1B). This LCR dominates over the endogenous ␤-globin control elements resulting in position-independent, high-level T cell-specific expression of the ␤-globin gene (30). The chromatin structure of the LCR sequences in this transgene (as defined by the DNase I hypersensitivity assay (39)) is similar to that seen in the endogenous locus. Therefore this system is a good physiological model for LCR study. Three deletion mutants of this construct were previously described and are also used later in our study (32).
Tissue Differential Methylation Patterns Exist in TCR␣ LCR DNA-We initially wanted to determine the general DNA methylation patterns over the length of this LCR in organs that differentially express the genes of the locus. The enzyme McrBC was used in a novel titration assay similar to the DNase I hypersensitivity assay (39). McrBC is a methylation-dependent restriction enzyme that recognizes a pair of methylated cytosine residues in the sequence 5Ј-Pu m C (N 40 -2000 ) Pu m C-3Ј, and cleaves within 30 bp from one of the methylated residues. Titration of this enzyme onto genomic DNA allows the scanning of large regions of DNA without regard to the presence of particular restriction enzyme recognition sites. The relatively loose sequence specificity allows, in principle, the detection of up to 50% of the CpG dinucleotides occurring in a given DNA sequence. McrBC treatment of isolated genomic DNA generates different fragment sizes depending on the location of methylated residues in the LCR fragment. These cleavage products can be detected by indirect end labeling of a parent restriction fragment of interest by Southern blot. Their positions can also be mapped relative to the fragment end recognized by the labeled probe. We analyzed the TCR␣ LCR in both endogenous and ␤:1-8 transgene contexts. For our transgenic analysis, high-copy transgenic lines were used. Short exposure of Southern blots to film allows selective detection of the strong signals from transgene cleavage products when compared with the low signals from the endogenous locus. First, parent fragments were generated by restriction enzyme digestion of thymus and kidney genomic DNA from C57BL/6 wild-type mice and full-length LCR reporter construct (␤:1-8) transgenic mice. Fig. 2A shows the analysis of the transgenic LCR sequences. As increasing amounts of McrBC were added, the parent fragments became less evident, and smaller fragments progressively appeared (marked A-F). The detected McrBC cleavage indicated a high degree of DNA methylation in the TCR␣ LCR sequences. The cleavage patterns were roughly similar from kidney and thymus (sites A-C and E-F). The only major difference was the stark appearance of a 1.5-kb fragment in the kidney sample (site D). This indicated an area of hypermethylated CpG dinucleotides in kidney DNA which mapped close to the region of HS4 (Fig. 2B). This area appeared selectively unmethylated in thymus DNA. McrBC analysis of spleen and heart DNA from the same animals showed that the HS4 region was hypermethylated in heart DNA, but not in spleen DNA (data not shown). The results obtained from analysis of the endogenous TCR␣ LCR in the four organs were similar to those seen in the transgenic model (data not shown).
Since differential methylation seemed to be specific to the HS4 region of the LCR, we focused on this area to confirm the results of the McrBC assay with a more widely used method. We utilized restriction enzymes whose cleavage activities were differentially sensitive to DNA methylation. An HS4 regionspecific probe was used to detect enzyme cleavage by Southern blot analysis. MspI and HpaII restriction enzymes recognize the same sequence 5Ј-CCGG-3Ј. MspI cuts DNA regardless of its methylation status, but HpaII only cuts the sequence if the internal cytosine residue is unmethylated. We also used HhaI, which recognizes the sequence 5Ј-GCGC-3Ј and only cuts unmethylated DNA. C57BL/6 genomic DNA from thymus, spleen, kidney, and heart were digested with the MseI restriction enzyme to generate a 1.2-kb parent fragment spanning the HS4 region, and then further digested with MspI, HpaII, or HhaI to assess DNA methylation in the region (Fig. 3A). As expected, treatment with MspI resulted in cleavage of the parent fragment in DNA from all organs (Fig. 3B, top). Treatment with HpaII or HhaI resulted in extensive parent band cleavage in thymus and spleen but not in kidney or heart. The signals present in the parent fragment and enzyme cleavage subfragments (in the HhaI cleavage samples) were quantified by Phos-phorImager analysis. After subtracting the lane background signal, the percent demethylation (subfragment signal divided by the sum of the subfragment and parent signals) was calculated and graphed (Fig. 3C). A high percentage indicates relative demethylation in the region, while a low percentage implies methylated DNA. Such PhosphorImager analysis of the C57BL/6 DNA Southern blot showed nearly 100% demethylation in the thymus and spleen and below 30% demethylation in kidney and heart. These same results were reproduced in ␤:1-8 transgenic genomic DNA subjected to the same restriction enzyme treatment (data not shown). Since HpaII and HhaI only cleave unmethylated DNA, this result confirmed that the HS4 region of the TCR␣ LCR was hypermethylated in nonlymphoid organs, and undermethylated in lymphoid organs.
The HS1/HS1Ј Region of the LCR Is Involved in Lymphoid- specific Demethylation-Germline deletion of TCR␣ enhancer region (HS1 and HS1Ј) results in a deficiency in TCR␣ gene rearrangement and transcription (35). To assess the DNA methylation status of this impaired TCR␣ locus, we obtained the TCR␣ enhancer knock-out mice and performed the restriction enzyme cleavage assay. Fig. 3B (middle) shows that this mutation resulted in the disruption of the lymphoid organspecific DNA demethylation mechanism operating on the HS4 region of the endogenous locus. This is evidenced by the persistence of some HS4 region parent fragment equivalently in all organs upon treatment with HpaII and significantly reduced parent fragment cleavage by HhaI. This indicates equivalent partial DNA methylation of this region in all organs rather than the clear tissue-specific patterns seen in the wildtype locus. PhosphorImager analysis confirmed this result (Fig.  3C). These HS1/HS1Ј knockout mice also have a defect in thymic development due to the absence of cell surface TCR (34).
To determine if the DNA demethylation defect was secondary to impaired thymic maturation, the TCR␣ constant region knock-out mice were also analyzed. These mice have the identical block in thymic development present in the regulatory region mutant, but have an intact HS1/HS1Ј region. The lymphoid organ-specific DNA demethylation activity in these mice was similar to that of the wild type counterpart (Fig. 3, B, bottom, and C). This indicated that the mutation of the HS1/ HS1Ј region was indeed directly involved in the lymphoid organ-specific demethylation of HS4 DNA.

HS4 Region Demethylation Is Local and Occurs Independently of VDJ Recombination-
The results from the above studies of mice with mutant TCR␣ loci raised the possibility that demethylation of the HS4 region DNA may precede the onset of VDJ recombination or gene expression. To confirm this, recombinase-activating gene Rag1 deficient mice were analyzed to determine the requirement of VDJ recombination for the lymphoid organ-specific demethylation activity. The Rag1 gene product is required early in T cell development for TCR gene rearrangement. Thymocytes in these mutant mice are arrested at a very early stage of development (33). Using the same methylation sensitive restriction enzymes used previously, we found that HS4 demethylation occurs normally in these mice (Fig. 3C). PhosphorImager analysis of Southern blots (HhaI samples) of Rag1Ϫ/Ϫ, in comparison to those experiments shown in Fig. 3B, confirmed that Rag1Ϫ/Ϫ and wild-type DNA are similar with respect to the methylation status of the HS4 region of the LCR. The only mutation that affects the wild-type DNA demethylation pattern is the E␣Ϫ/Ϫ deletion. This shows that unrearranged TCR␣ alleles can undergo demethylation in the lymphoid system. This also suggests that the demethylation mechanism acts early in thymocyte development prior to gene rearrangement.
The results of the McrBC assay strongly suggested that the DNA demethylation event was strictly localized to the HS4 region of the LCR. To confirm this, we analyzed intervening regions of the LCR in wild-type and mutant mice, with MspI, HpaII, and HhaI enzymes (Fig. 4). We found that the HS2 region is heavily methylated in lymphoid organ DNA despite the demethylated state of the neighboring HS4 region. In nonlymphoid organs, HS2 region DNA appears to be partially methylated. In contrast to the strong effect of the E␣Ϫ/Ϫ mutation on the tissue-specific methylation status of HS4 region DNA, this mutation has only a very minor effect on the wild- type HS2 DNA methylation pattern in these organs (Fig. 4, B  and C). DNA derived from lymphoid organs is still heavily methylated and non-lymphoid organ DNA remains partially methylated.
HS1Ј Induces Lymphoid Organ-specific DNA Demethylation in the Absence of HS1-As previously mentioned, two HS clusters are missing from the LCR in HS1/HS1Ј knockout mice. We employed our transgenic model to determine the contribution of each HS cluster to the control of HS4 region DNA demethylation in lymphoid organs (Fig. 5A). Genomic DNA from mice transgenic for the ␤:1-6, ␤:1Ј-6, and ␤:2-6 constructs were analyzed. As mentioned before, these constructs are the previously described deletion mutants of the parent ␤:1-8 transgene (Fig. 1A) (30,32). Once again, we used the methylation sensitive restriction enzymes to determine DNA methylation status of the transgenes. As shown in Fig. 5B, the cleavage pattern of the ␤:1-6 and ␤:1Ј-6 transgenes were generally similar to those of the ␤:1-8 transgene (data not shown) and the endogenous LCR (Fig. 3B, top). These data show that HS7, HS8, and HS1 were not critical for lymphoid organ-specific HS4 demethylation. In contrast, analysis of ␤:2-6 transgenic mice showed that the HS4 region parent fragment of this construct was resistant to cleavage by HhaI and HpaII in all four organs analyzed. This indicated that the ability to demethylate this DNA segment in lymphoid organs is lost upon deletion of the HS1Ј region. This result was confirmed in other independent lines transgenic for the ␤:1Ј-6 and ␤:2-6 constructs (data not shown). Therefore, the difference in DNA methylation status between these two transgenes is not due to the position of transgene integration. PhosphorImager analysis confirmed these results (Fig. 5c). DISCUSSION Recent literature suggests that DNA methylation can precede, and indeed may guide the modification of chromatin structure and transcription. Thus, it is important to understand the control elements regulating the differential DNA methylation machinery and its targets. This is especially true for the tissue-specific DNA demethylation events that prevent particular genes from being silenced by global mechanisms in the specific cells in which they should be active. Here we introduce a novel assay for the detection and mapping of sites of DNA methylation. The major existing technique detects methylation indirectly and usually employs frequent cutting restriction enzymes (e.g. HpaII and HhaI). This makes them difficult to use in the analysis of large regions of DNA. We have used an McrBC titration assay to detect methylated CpG dinucleotides over longer stretches of DNA without regard to the presence of particular restriction sites. This assay also enables the detection and mapping of regions of tissue-differential DNA methylation without specific prior knowledge of their likely location. Once a region of difference is found, more traditional assays can be used to examine that particular, and perhaps unexpected region. We used this combination of assays to identify a novel, lymphoid organ-specific DNA demethylation activity that operates in a region of the TCR␣/Dad1 locus. Furthermore, we demonstrate that a distant cis-acting element of the LCR is involved in the control of this tissue-specific demethylation both in transgenic constructs as well as in its endogenous locus.
Elements Controlling Specific DNA Methylation Status-Other cis-acting elements affecting DNA methylation have been previously identified in only a few gene loci. Transgenic models have been used to detect an exogenous DNA element targeting a transgene locus for methylation by the mouse strain-specific methylation-modifier gene product, Ssm1 (40). Using transient reporter gene transfection assays and cultured cell lines, cis-acting elements controlling DNA demethylation events have been described in the rat ␣-actin promoter (22) and in the immunoglobulin light chain enhancer (Ig) (23,41). The Ig demethylation was recently confirmed in vivo in B-cells of transgenic mice. It appears to occur in a mono-allelic fashion prior to VDJ recombination of the locus early in B-cell development (42). The TCR␣ HS4 region demethylation described here also seems to occur prior to VDJ recombination of the TCR␣/␦ locus. Rag1Ϫ/Ϫ mice are not able to carry out VDJ recombination and only contain immature lymphocytes in their lymphoid organs. Yet, HS4 is still demethylated in these immature cells. The HS4 demethylation seen in thymus and spleen DNA from wild-type mice is complete and, therefore, is most likely not mono-allelic. Unlike the Ig locus, VDJ recombination at the TCR␣/␦ locus does not display allelic exclusion (43,44). As the Ig demethylation is postulated to have a role in allelic exclusion, mono-allelic demethylation would not be expected in the TCR␣/␦ locus. Furthermore, the complete demethylation of HS4 seen in spleen DNA indicates that this process occurs in both B-cells (70 -75% of splenocytes) and T-cells (25-30% of splenocytes). Similar complete demethylation of this region occurs in Rag1-deficient spleen. These data further suggest that HS4 demethylation may also occur in non-lymphoid hematopoietic cell types (e.g. monocytes and granulocytes). These blood cell populations can, variably, make up as much as 50% of a Rag1-or Rag2-deficient spleen (33,45). Further study of these populations in wild-type mice is necessary to confirm this possibility. Nevertheless, demethylation of HS4 DNA seems to occur at an early stage either in lymphoid precursors, or possibly even earlier in hematopoietic development.
HS1Ј can control the lymphoid-specific demethylation of HS4 region DNA, yet it is present in the nuclei of all organs examined (28,30,32). This is consistent with our previous data demonstrating an important role for HS1Ј in tissue-specific functions of the LCR. The data we present here add strength to the hypothesis that distinct sets of nuclear factors act via sequences in HS1Ј to accomplish different functions in lymphoid and non-lymphoid tissues.
DNA Methylation, Chromatin Structure, and LCR Activity-The correlation between DNA methylation, chromatin changes, and transcription has been known for a long time. Methylated DNA generally, but not absolutely (46), correlates with inactive genes and closed chromatin (18). The recent finding of the histone deacetylase complex interaction with MeCP2, and its effect on transcription, has provided part of a biochemical basis for this perceived correlation. However, there are conflicting data regarding which changes first, the methylation patterns, or the chromatin structure (47,48). Our data raises the possibility that LCR induced chromatin changes may first guide DNA demethylation at specific points in their loci to prevent subsequent histone deacetylase-mediated chromatin closure. In this model, the state of chromatin has two distinct roles. The postulated early chromatin changes mediated by LCRs would be an "enabling" step for subsequent gene expression while the histone deacetylase mediated chromatin modification would constitute more of a "maintenance" step ensuring the propagation of the correct chromatin state to the progeny of cell division. If this is true, it may explain why it has been difficult to put DNA methylation and chromatin changes in clear-cut order. Since the demethylation event we describe appears to be localized, our model would also raise the interesting possibility that specific, site-directed, rather than global, demethylation is enough to prevent chromatin closure by histone deacetylases. pattern of ␤:2-6 transgenic mice is not. C, PhosphorImager analysis of the HhaI lanes of the above Southern blots. Percent demethylation is calculated as described under "Experimental Procedures." T, thymus (stippled bars); S, spleen (striped bars); K, kidney (solid bars); H, heart (open bars).