Targeted Deletion of the Chicken β-Globin Regulatory Elements Reveals a Cooperative Gene Silencing Activity*

The chicken β-globin locus represents a well characterized system to study the role of both proximal and distal regulatory elements in a eukaryotic multigene domain. The function of the chicken βA/ϵ-intergenic enhancer and upstream regulatory elements 5′-HS1 and 5′-HS2 were studied using a gene targeting approach in chicken DT40 cells followed by microcell-mediated chromosome transfer into human erythroleukemia cells (K562). These regulatory elements all repressed expression of the ρ- and βH-chicken globin genes in the chromosome transfer assay. No ρ- or βH-globin gene expression was detected in K562 cells containing the chicken chromosome without deletions, whereas ρ- and βH-mRNA was activated in K562 cells containing chicken chromosomes with deletions of the intergenic enhancers, 5′-HS1 and 5′-HS2. Transcriptional activation of the ρ- and βH-globin genes correlated with hyperacetylation of histones H3 and H4, loss of histone H3 lysine 9 methylation, and binding of RNA polymerase II to the gene promoters. Surprisingly, the status of CpG dinucleotide methylation at the promoters did not correlate with the transcriptional status of the genes. Our results using a chromosomal transfer assay demonstrate an identical silencing function for these regulatory elements, which suggests they function as part of a common silencing pathway or complex.

The genes encoding the protein subunits of hemoglobin are among the first and most studied eukaryotic gene families. The chicken ␤-globin locus is a well studied gene locus with a number of unique and advantageous features. Developmentally, gene expression is well characterized, and unlike the human and mouse ␤-globin loci, the genes are not arranged in the order that they are expressed developmentally (1,2). There are four chicken ␤-like genes, 5Ј--embryonic, ␤ H -fetal, ␤ A -adult, and ⑀-embryonic-3Ј. In addition to the four structural genes, the locus possesses both intergenic and upstream regulatory elements (see Fig. 1). The intergenic enhancer has been shown to have properties of a locus control region (LCR) 1 element and has been shown to work cooperatively with 5Ј-HS1-4 (1, [3][4][5]. Patterns of histone acetylation, histone methylation, DNA methylation, DNase I sensitivity and DNA replication have been defined in this locus (6 -12). The first vertebrate insulator element described is located at 5Ј-HS4 in the chicken ␤-globin locus (13). The insulator element is located at the 5Ј-boundaries of DNase I sensitivity and histone acetylation (7). The functions of these regulatory elements have been studied in transient and stable transfection assays (14 -16) and in transgenic mice (3,5,17) but not in their endogenous chromosomal location like the human and mouse ␤-globin elements (18).
Chicken B cell lines such as DT40 cells have been described to be able to undergo homologous recombination at high frequency. They have been used to study both the functions of genes and regulatory elements (19,20). We have reported a somatic cell system using microcell hybrids where human chromosomes are transferred into DT40 cells for high frequency gene targeting and subsequently into mouse erythroleukemia cells for phenotypic analysis of mutations (20,21). We have used a variation of this system to conduct a deletion analysis of the chicken ␤-globin locus regulatory elements in their endogenous chromosomal location. In this report, we have deleted the intergenic ␤ A /⑀-enhancer, 5Ј-HS1, and 5Ј-HS2 in DT40 cells and analyzed the phenotype of these mutations in human erythroid K562 cells. Our results demonstrate that these regulatory elements possess identical gene silencing activity.

MATERIALS AND METHODS
Cell Culture and Electroporation-Chicken B cell line DT40 (19,20) and targeted clones E-DT 29, HS1-DT 22, HS1ϩDT 2, HS2-DT 44, and FRDT 13 were maintained in Dulbecco's modified Eagle's media supplemented with 10% fetal bovine serum and 5% chicken serum. Human cell line K562 and transfer clones E-K1, HS1-K3, HS2-K2, HS1ϩK2, FRK1, and FRK2 were maintained in RPMI 1640 media supplemented with 10% fetal bovine serum. For stable transfections into DT40 cells, 1 ϫ 10 7 exponentially growing cells were harvested and suspended in serum-free Dulbecco's modified Eagle's medium and 20 g of PvuIlinearized (p␤⑀-neo R , pHS1-neo R , pHS1ϩneo R , pHS2-neo R and pFRneo R constructs) plasmid DNA was added. The cells were electroporated at 25 F, 550 volts (20) using a Bio-Rad Gene Pulser. The cells were incubated in nonselective medium for 48 h, then plated into 8 ϫ 96-well plates in complete medium containing 1 mg/ml G418 (VWR). After 10 days, individual clones were picked and expanded. For transient transfections into K562 cells, 1 ϫ 10 7 exponentially growing cells were harvested and suspended in serum-free RPMI 1640 medium, and 20 g of supercoiled p␤⑀ϩneo R (16) or pHS1ϩneo R plasmid DNA was added. The cells were electroporated at 960 F, 250 volts. After incubation at 37°C for 48 h, total RNA was isolated and used as template for positive control RT-PCR of the chicken ␤ A -, ⑀-, and -globin genes.
Targeting Constructs-Positive (neo R or his R ) selectable marker genes were inserted into the deletion targeting plasmids p␤⑀- (22). For negative selection, a 1.6-kb dipA fragment (21) was inserted at the 5Ј-end of the targeting sequence. 5Ј-HS1 targeting vectors were constructed by amplifying ϳ8-kb fragment from DT40 genomic DNA, a 2-kb loxpgkneo R fragment (23) was inserted into the unique Eco47III site to get pHS1ϩ vector, and Eco47III plus BlnI sites to get pHS1Ϫ vector. 5Ј-HS2 targeting vectors were constructed by amplifying an ϳ7.6-kb fragment from DT40 genomic DNA, a 2-kb loxpgkneo R fragment (23) was inserted into SspBI plus BamHI sites to get pHS2Ϫ vector. A 1.6-kb dipA fragment (21) was inserted at the 5Ј-end of the targeting sequence as the negative control. The folate receptor targeting vector was constructed by amplifying a 4.6-kb fragment from DT40 genomic DNA. A 2-kb loxpgkneo R fragment was inserted into the unique NcoI site of FR4.6. For negative selection a 1.6-kb dipA fragment (21) was inserted at the 5Ј-end of the targeting sequence.
Single Cell RT-PCR-K562 and transfer clones were counted with a hemocytometer and diluted in sterile 1ϫ phosphate-buffered saline, such that a theoretical given cell number was present in a 5-l volume.
The cells were then added to 0.5-ml Eppendorf tubes containing 5 l of 2ϫ RNasin ribonuclease inhibitor freeze medium. The cells were quickly frozen on dry ice and stored at Ϫ70°C. The entire 10 l of cell lysate was added directly to a 50-l RT-PCR using the Access RT-PCR System (Promega enotes applications, Promega) (25) and amplified using identical primers and conditions as detailed above.
Real Time PCR-The abundance of human and chicken sequences in various clones was analyzed by real time PCR on an ABI 7900. Reversetranscribed cDNA samples were quantified by OliGreen (Molecular Probes), and 8 ng of cDNA was amplified using human and chicken primer/probe combinations. The primers and probes are listed in supplemental Table I. The abundance of each sequence was calculated using a standard curve of amplified sequences ranging from 0.005 to 50 ng/reaction. Each measurement was performed in triplicate, and each experiment was repeated twice on two independently derived cDNA samples.
Southern Blotting and Hybridization-10 g of each DNA sample was digested with the indicated restriction endonucleases, and agarose gel electrophoresis, Southern blotting, hybridization, and autoradiography were performed as described (26,27).
Hybridization Probes-A map of the chicken ␤-globin locus and upstream folate receptor locus is shown in Fig. 1. Probes ␤ H , ⑀, , MN3 and chicken 3Ј were amplified from DT40 genomic DNA. The primers are listed in supplemental Table II. ␤ H -, ⑀-, and -primers are the same as chromatin immunoprecipitation (ChIP) primers. A neo R fragment was excised from plasmid loxpgkneo R by NotI (23). To obtain the HS23 probe, a 5-kb BamHI fragment was subcloned from a phage DNA clone (6), and a 1.2-kb KpnI/HindIII fragment was excised. Probe Enh J (200 bp) was excised from plasmid p⌬CAT with NcoI plus EcoRI (17).
ChIP-ChIP was performed according to the protocol included with the anti-acetyl H3 or H4 antibodies (Upstate Biotechnology, Lake Placid, NY). Control DNA was recovered from the sonicated supernatant by the same procedure as for the ChIP DNA. The average size of the control DNA recovered was 600 -1000 bp. We used antibodies antidiacetylated histone H3 (Upstate, number 06-599), anti-tetracetylated histone H4 (Upstate, number 06-866), anti-dimethyl-Histone H3 (Lys-9) (Upstate, number 07-212), pol II (Santa Cruz Biotechnology, number sc-899), and purified mouse IgG (Sigma). Approximately 20% of input DNA was used as template for PCR. No antibody and mouse IgG were used as negative controls. PCR primer sequences are listed in supplemental Table II. A quantitative analysis of results was performed using densitometric analysis (Bio-Rad) and determination of the ratio of signal/input (28). Each experiment was performed 2-3 times, and representative data are shown.
Microcell-mediated Chromosome Transfer-The E-K, HS1-K, HS1ϩK, HS2-K, and FRK series microcell hybrids were prepared as described (21). Approximately 3 ϫ 10 7 microcells were fused with 1 ϫ 10 7 K562 (his R -by stable transfection with a loxpgkhis R plasmid) cells. After fusion, the material was rinsed at least three times with serum-free medium to remove PEG. The fusion mixtures were incubated in nonselective medium for 48 h and then plated clonally into 8 ϫ 96-well microtiter plates in 0.2 ml/well of medium containing 1 mg/ml G418 plus 1 mg/ml histidinol. After 4 -8 weeks, hybrid clones were expanded and maintained in medium containing 0.7 mg/ml G418. Microcell hybrids derived from E-DT 29 donor are designated as E-K series clones, HS1-DT 22 are designated as HS1-K series clones, HS1ϩDT 2 are designated as HS1ϩK series clones, HS2-DT 44 are designated as HS2-K series clones, and those derived from FRDT 13 donor are designated as FRK series clones.
DNA Methylation Analysis-5 g of genomic DNA extracted from exponentially growing cells was digested with 100 units of either MspI or HpaII for Ͼ5 h at 37°C. The digested DNA was extracted with phenol-chloroform and precipitated with ethanol. 50 ng of undigested, MspI-digested, or HpaII-digested DNA was used as a template for PCR. Primer sequences are listed in supplemental Table II. Amplification products were separated by gel electrophoresis and stained with ethidium bromide and photographed under ultraviolet light. For Southern analysis, 10 g of genomic DNA was digested with KpnI plus either MspI or HpaII at 37°C for Ͼ5 h. Southern blotting was performed as described (27).

RESULTS
Targeting vectors were constructed to delete the endogenous ␤ A /⑀-intergenic enhancer and upstream regulatory elements 5Ј-HS1 and -HS2. Control targeting vectors were constructed to insert selectable marker genes without deletions for the upstream folate receptor locus as well as 5Ј-HS1 (HS1ϩ) (Fig. 1, A and C). The folate receptor gene is located upstream of the ␤-globin locus and is inactive in adult red blood cells but active in primitive erythroid cells. It is separated from the chicken ␤-globin locus by an insulator element and constitutively closed chromatin (12). The enhancer deletion construct (ϳ100-bp deletion, Fig. 1B) was identical to one previously used in transient transfection and transgenic analyses except for the addition of positive and negative selectable marker genes (15,16). Targeting vectors were electroporated into DT40 cells, and after selection, colonies were screened by Southern blot analysis. Targeted clones were obtained at frequencies between 2-15%. DT40 cells containing targeted chromosomes were transferred into human (K562) cells using microcell-mediated chromosome transfer. Clones containing transferred chicken chromosomes were obtained. These hybrid clones were then analyzed by Southern blot analysis using conventional and field inversion gel electrophoresis to determine the integrity of the transferred chicken chromosome. The selectable marker gene was deleted by transient expression of Cre recombinase (20). Probes from the upstream folate receptor locus and 3Ј-HS1 and internal probes within the chicken ␤-globin locus, demonstrated that the locus was intact and not rearranged (Fig. 2). Chicken chromosomes with a selectable marker gene inserted into the folate receptor gene or 5Ј-HS1 were used as marked, wild type control chromosomes. Interestingly, Southern analysis of the transferred folate receptor and 5Ј-HS1ϩ clones in K562 cells indicated the presence of the normal, unmarked chicken ␤-globin locus as well (Fig. 2, A and B). Cre recombinase-mediated removal of the selectable marker gene did not alter the expression phenotype of these clones, indicating that these clones could be used as normal controls for comparison to the phenotype of our targeted clones.
Expression of the chicken ␤-globin genes , ␤ H , ⑀, and ␤ A in K562 cells was determined by RT-PCR. Expression patterns before and after Cre-mediated deletion of the selectable marker gene were identical (Fig. 3). K562 cells express fetal and embryonic human globin genes but no adult ␤-globin (Ref. 29 and Fig. 3). Surprisingly, K562 cells containing the control chicken chromosome expressed both chicken adult (␤ A ) and embryonic (⑀) globins, but notand ␤ H -globin mRNA (Fig. 3). In contrast to the wild type chromosome, Fig. 3 shows that in early passage K562 cells that contained a deletion of the intergenic enhancer, 5Ј-HS1 or 5Ј-HS2, the chicken ␤ H -globin gene was expressed at levels comparable with the endogenous human ␥-globin gene. In addition to activation of the ␤ H -globin gene, early passage K562 clones containing mutant chicken chromosomes no longer expressed the ␤ A -or ⑀-globin genes. Interestingly, with extended time in culture (several months or longer), transcription of -, ␤ A -, and ⑀-globin genes in the mutant chicken chromosome clones was stably up-regulated (Fig. 3A), whereas ␤ Hglobin gene expression remained unchanged. There was no evidence of genomic rearrangement in late passage K562 hybrid cells (Fig. 2C) and a dilutional single cell RT-PCR analysis showed no evidence of heterogeneity in mRNA expression patterns (Fig. 3C). Quantitative real time PCR analysis of gene expression patterns in control and mutant chicken chromosomes in K562 cells demonstrated results similar to those shown using standard RT-PCR (Fig. 3B). One characteristic of active genes is the presence of acetylated histones H3 and H4 at their promoters (30). The patterns of histone acetylation in the chicken ␤-globin locus in normal and mutant chromosomes were studied using ChIP assays with antibodies to diacetylated histone H3 and tetracetylated histone H4 (Fig. 4). Hyperacetylated histones H3 and H4 were present at the promoter regions of chicken ␤-like globin genes in both mutant and control chromosomes and correlated with gene expression. At the ␤ H -globin gene promoter, acetylated histones H3 and H4 correlated with expression and were present at the ␤ H promoter in early and late passage K562 cells containing the mutant chicken ␤-globin loci. In early passage mutant cells, there was no acetylated histone H3 and H4 found at the -globin gene promoter, consistent with the lack of expression of this gene. In late passage mutant cells where the -globin gene was expressed, acetylated histones H3 and H4 were present at the -globin gene promoter (Fig. 4A). At the ␤ Aand ⑀-globin gene promoters, binding of acetylated histones H3 FIG. 3. RT-PCR analysis of chicken ␤-globin gene expression in K562 microcell hybrids containing control or mutant chicken chromosomes. A, K562 clones containing the control chromosomes (FRK and HS1ϩK) clones express chicken ␤ A -and ⑀-globin genes but no ␤ H -or -globin mRNA. Targeted early passage cells (E) express ␤ H -globin mRNA and no detectable ␤ A -, ⑀-, or -chicken globin mRNA. As a positive expression control (PC), plasmids including the chicken ␤ A -and ⑀-globin gene genomic regions or the -globin gene genomic region were electroporated into K562 cells, and ␤ A -, ⑀-, and -globin mRNAs were transiently expressed. Human ⑀-, ␥-globin and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression served as endogenous controls for RNA quantity and integrity and were coamplified with chicken ␤-globin primers when technically compatible (see two lower panels) with similar results. Cre recombinase was transiently expressed to remove the neo R gene with no change in gene expression patterns. Late passage (L) K562 mutant clones transcribe ␤ A -, ⑀-, -, and ␤ H -globin mRNAs. B, quantitative real time PCR analysis of gene expression patterns in control and mutant chicken chromosomes in K562 cells demonstrated results similar to those shown in A. Human ⑀-globin was used as control. C, late passage mutant K562 cells have a homogenous expression pattern. A dilutional RT-PCR analysis was used to examine the patterns of gene expression in single cells. Identical RNA expression patterns were seen in five independent single cell dilution series from late passage mutant K562 cells, suggesting that there were no detectable cell subpopulations with different expression patterns to explain the differences observed in early versus late passage cells. and H4 was observed in control and late passage deleted cells, consistent with expression patterns.
ChIP assays were also used to examine the distribution of lysine 9 methylation at histone H3 (H3-meK9) that has been shown in several different systems to be associated with transcriptional repression (8,30,31). ChIP assays using an antibody to dimethylated H3 K9 showed the presence of H3-meK9 at the ␤ H -globin gene promoter in control cells but not in mutant cells, consistent with gene expression patterns (Fig. 5). The promoter of the expressed ␤ A -and ⑀-globin genes were devoid of H3-meK9 in control cells. However, in the early passage mutant cells, H3-meK9 was present at the ␤ A -and ⑀-globin gene promoters. In late passage cells where expression of the ␤ A -and ⑀-globin gene had reactivated, H3-meK9 was absent from the ␤ A -and ⑀-globin gene promoters.
The transcriptional activity of vertebrate genes correlates with the presence of hypomethylated CpG nucleotides at the gene promoters (30). To examine the DNA methylation pattern of CpG dinucleotides at the chicken ␤-globin gene promoters, assays were performed using methylation-sensitive restriction enzymes in both PCR and Southern analyses. Both techniques yielded similar results, although the Southern analysis is more quantitative in nature (Fig. 6, A and B). The promoter of the ␤ H -globin gene was methylated in both wild type and deleted chromosomes regardless of expression status. However, the DNA methylation status of promoters of the ⑀and ␤ A -globin genes differed in normal and mutant chromosomes. The ⑀and ␤ A -globin gene promoters were CpG hypomethylated in cells containing the wild type ␤ A -and ⑀-globin expressing chromosome, whereas in mutant K562 cells, the ⑀and ␤ A -globin gene promoters were CpG methylated. The methylation status at the ⑀and ␤ A -globin gene promoters did not change in early or late passage cells (Fig. 6), even though these genes were transcriptionally activated in late passage cells, suggesting another epigenetic mechanism was operative.
RNA polymerase II (pol II) is present at the promoters of actively transcribed genes, and this observation has been confirmed in the ␤-globin locus using the ChIP assay (32)(33). Recent evidence has suggested that the mouse ␤-globin LCR may regulate transcription at the level of transcriptional elongation (33). To determine whether the changes in gene expression patterns we observed were transcriptional or posttranscriptional in nature, ChIP assays were performed using antibodies to pol II. Fig. 7 showed that the presence of pol II at the chicken ␤-globin promoters correlated with the transcriptional status of the chicken ␤-globin genes. In cells containing the control chicken chromosome, pol II was present at the ␤ Aand ⑀-globin gene promoters but not theor ␤ H -globin gene promoters. In early passage K562 cells containing the mutant chromosomes, pol II was present at the promoter of the ␤ Hglobin gene but was absent at the -, ⑀and ␤ A -globin gene promoters. However, in late passage cells, where -, ⑀-, and ␤ A -globin gene transcription has been reactivated, pol II was associated with all four globin gene promoters. These data are consistent with the chicken intergenic enhancer and upstream regulatory elements (5Ј-HS1 and 5Ј-HS2) regulating gene expression at the level of transcriptional initiation. The results are summarized in Table I. DISCUSSION The chicken ␤-globin locus represents a well characterized model system where the relationship of chromatin structure, transcription, and DNA replication can be studied. The availability of the recombination-proficient chicken B cell line DT40 has allowed the development of a system where mutations in the regulatory elements can be introduced into the endogenous chicken ␤-globin locus, and the erythroid phenotype can be analyzed after microcell-mediated chromosome transfer into human erythroleukemia cells.
K562 cells were permissive for expression of all the chicken ␤-like globin genes in the chicken chromosomal context. Previous studies showed that transfected chicken globin genes could be expressed in murine erythroleukemia cells, suggesting conservation of the basic transcriptional machinery between mice and chickens (34). K562 cells express the human embryonic and fetal ␤-globin genes but no endogenous or transfected adult ␤-globin genes (29,35); therefore, the expression of the chicken adult ␤-globin gene in this environment was initially surprising. However, K562 cells are reported to express the minor adult ␦-globin gene (36,37). K562 variants that express the adult ␤-globin gene have been described, and constructs containing the ␤-globin gene were active when transiently transfected into those variant K562 cells (36). Wild type K562 cells have been hypothesized to be defective in a trans factor re-quired for human adult ␤-globin gene expression (29,(35)(36)(37), which is clearly not required for chicken adult ␤-globin (␤ A ) gene expression. One candidate for such a factor is EKLF, an erythroid transcription factor reportedly not expressed in K562 cells (38). However, exogenous expression of EKLF in K562 cells was unable to restore transcription of the human ␤-globin gene (38). The ability of all the chicken ␤-like globin genes to be expressed in K562 cells suggests EKLF is not required for their transcription in this cell line.
Expression of the chicken ␤ H -and -globin genes occurred only in the chromosomal context when regulatory elements were deleted. Transient transfection and transgenic mouse studies have both suggested the intergenic enhancer as well as the upstream regulatory elements function in promoting chicken ␤-like globin gene transcription (3,14,39). In transient transfection assays, only the intergenic enhancers 5Ј-HS2 and 5Ј-HS3 showed enhancer activity. However, transient transfection assays, in contrast to transgenic mouse studies, provided evidence for a competitive mechanism of the intergenic enhancer between the ␤ A -and ⑀-globin genes (15,16). Experiments in transgenic mice also suggested that this element possessed the properties of an LCR in directing position-independent, copy number-dependent transgene expression, although the level of expression was only 2-3% of the endogenous mouse gene (17).
When the intergenic enhancer 5Ј-HS1 or 5Ј-HS2 is deleted,and ␤ H -globin gene transcription are activated, suggesting that these elements function as silencers in a chromosomal context. Silencer activity of these elements was not assayed in transient transfection assays, but silencing elements contained within the -globin gene have been described using these assays (39) FIG. 5. Analysis of lysine 9 histone H3 methylation at the promoter regions in control and mutant chicken ␤-globin loci. A, in DT40 cells, which do not express chicken ␤-globin genes, H3-meK9 was found at all ␤-like gene promoters. In K562 clones FRK1 and HS1ϩK, H3-meK9 is found at the ␤ H -andbut not at the ␤ A -and ⑀-globin gene promoters. In early passage K562 cells containing mutant chicken chromosomes, H3-meK9 is enriched at the ␤ A -, ⑀and -globin gene promoters. In late passage K562 cells containing mutant chicken chromosomes, all ␤-like gene promoters are devoid of H3-meK9, which correlates with the activation of all ␤-like gene expression in late passage cells. 5Ј-Condensed chromatin (C.C.) served as control. B, quantitative analysis of data from A (see text and the legend to Fig. 4 for details).
Transcriptional activation of the ␤ H -and -globin genes is correlated with hyperacetylation of histones H3 and H4 at the promoter but no changes in the status of methylated CpGs at the promoter. Because the methylation patterns of CpG dinucleotides in the chicken ␤-globin locus have been well studied during normal avian erythropoiesis and shown to cor-relate with gene expression (6,10,40), the presence of CpG methylation at the gene promoters is likely because of the epigenetic history of the chicken chromosome in DT40 lymphoid cells. Upon transfer of chicken chromosomes to the erythroid K562 cell environment, potential expression of all the chicken ␤-like globin genes can occur without promoter CpG hypomethylation. In this setting, the presence of acetylated histones H3 and H4 at the promoter is sufficient for transcription of these genes despite methylated CpGs at the promoter. Observations (31) reported in stably integrated immature chicken erythroid (6C2) cells demonstrated that the transcription of transgenes using the chicken ␤ A -promoter was accompanied by histone acetylation and absent H3-meK9 binding at the promoter. Similar to our results, transcription still occurred in the presence of methylated CpGs at the promoter. Collectively, these results suggest that at some promoters and chromosomal loci, epigenetic histone modifications can be dominant to DNA methylation. Treatment of K562 hybrid cells containing control or early passage enhancer-deleted chicken chromosomes with inhibitors of histone deacetylases trichostatin A and/or DNA methyltransferases was unable to change the pattern of chicken ␤-globin gene expression (not shown). These observations suggest involvement of a trichostatin A-resistant histone deacetylase such as sir2 (41) or an alternative, dominant histone modification such as H3-meK9 (8,42).
Our study illustrates the feasibility of a gene targeting approach in DT40 cells to delete regulatory elements from the chicken ␤-globin locus. The mutant phenotype was assayed in human erythroid cells by transferring the targeted chromosome into human erythroleukemia cells (K562). A similar approach has previously been used to study regulatory regions in the human ␤-globin loci, where the group of 5Ј-DNase I hypersensitive sites is collectively referred to as the ␤-globin LCR. Deletions of the individual endogenous murine and human hypersensitive sites produced modest (20 -40%) alterations in globin gene transcription (23,43,44). Larger deletions of the LCRs produced complete or severe reductions in transcription of the human and mouse ␤-like genes, respectively (21,26,45).
Recently, analysis of mice containing a targeted deletion of the ␤-globin LCR has revealed evidence for the function of the LCR in regulating ␤-globin gene expression via transcriptional elongation (32). Our analysis of the targeted deletion of the chicken intergenic enhancer, 5Ј-HS1 or -HS2 did not demonstrate that these regulatory elements regulate chicken ␤-globin gene expression via transcriptional elongation. ChIP assays using antibodies to pol II show a direct correlation of the presence of pol II at chicken ␤-globin promoters and transcriptional activation of these genes, suggesting that an effect of these elements on transcriptional elongation is unlikely. GATA-1 has been shown to recruit pol II to the mouse ␤-globin LCR, whereas NF-E2 is involved in recruiting pol II to the ␤-globin promoters. Both GATA-1 and NF-E2 act through a pol II initiation mechanism (32,46), which is compatible with our observations in the chicken ␤-globin locus. We hypothesize that K562 cells provide a permissive environment and appropriate protein factors to allow transcription of all the chicken ␤-globin genes. Which subset of genes is transcribed is dependent on the presence of cis elements on the chicken chromosome. Because we have shown K562 cells are capable of expressing all the chicken ␤-like globin genes, these cells provide, in the absence of an accessible animal model, a useful erythroid in vitro system to assay the phenotype of targeted mutations in chicken chromosomes. Individual deletions of the intergenic enhancer or upstream 5Ј-HS1 and -HS2 yielded identical phenotypes, suggesting a common gene silencing pathway or complex. Because there was no compensation/complementation between FIG. 6. Analysis of DNA methylation in control and mutant chicken ␤-globin loci. A, methylation-sensitive PCR analysis of chicken ␤-globin promoters. Genomic DNA was digested either with HpaII or MspI, undigested (U), HpaII-digested (H), or MspI-digested (M). DNA was amplified by PCR with primers from the ␤ H , ␤ A , , and ⑀ promoter regions. Methylated CpG sequences will not be cut by HpaII and therefore be amplifiable. Unmethylated sequences will be cut into smaller fragments and are not amplifiable. Chicken 5Ј-HS4 (HS4) and the 5Ј-HS4 upstream region (condensed chromatin (C.C.)) were used as controls for known hypomethylated and methylated sequences in erythroid cells, respectively (11). The promoters of all the chicken ␤-globin genes contained methylated CpG dinucleotides in mutant K562 clones regardless of transcriptional status. In control clones FRK1 and HS1ϩK2, methylated CpG dinucleotides were detected in the ␤ H -,but not the ␤ A -and ⑀-globin gene, promoters that correlates with transcriptional status. In DT40 cells, all ␤-globin sequences except 5Ј-HS4 were CpG-methylated. B, Southern blot analysis of DNA methylation using restriction endonuclease isoszchizomers HpaII (H) and MspI (M). Southern analysis confirms the PCR analysis shown in A, indicating that the promoters of theand ␤ H -globin genes are methylated in both control and mutant chromosomes, whereas the ⑀ promoter is CpGhypomethylated in cells containing the control chromosome and CpGmethylated in K562 cells with the mutant chromosome. The small DNA fragments produced by complete cleavage (plus KpnI, K) are shown (27). If a band observed in the KpnI plus HpaII (KϩH) lane, that sequence is unmethylated. In contrast, if there is no band observed, then that sequence is methylated and a higher molecular weight band was seen (not shown). The presence of numerous HpaII sites in the ␤ A -promoter rendered the Southern analysis for this region difficult to interpret (not shown). Probes from the chicken HS23 and 3Ј-␤-globin regions yielded similar DNA methylation patterns in control versus mutant K562 clones compatible with complete or near complete hypomethylation. C, map of genes, probes, and position of MspI restriction sites (vertical lines). the deletions, the likely possibility was that these elements were involved in the same pathway/complex.
Only a few reports have suggested a similar gene silencing activity in other enhancers and regulatory sequences. Reports of stage-specific gene expression and repression in the human ␤-globin locus have been reported in yeast artificial chromosome constructs with deletions of the 5Ј-LCR HS sites 2, 3, and 4 (47)(48)(49). However, no reported regulatory mutations in the endogenous mouse or human ␤-globin loci display a phenotype similar to that reported here. A recent report using a proteomics-based approach to analyze DNA-protein interactions at human LCR 5Ј-HS2 obtained evidence for a repressive role of c-maf complexes during early erythropoiesis, prior to globin gene expression (50). A silencer 5Ј to the human ⑀-globin gene has been reported using a transgenic mouse assay (51), but no confirmatory mutations in the endogenous mouse or human loci have been reported.
The precedent for colocalization of vertebrate tissue-specific enhancer and silencer activities in a 100-bp sequence is unique. In another well studied enhancer in the immunoglobulin heavy chain locus (IgH E), a repressive function has been reported in non-B cells, localized to the 3Ј-E4 region and the mE5 box. Octamer and basic helix-loop-helix proteins were found to bind to these sequences (52)(53)(54). In the human and mouse CD4 gene loci, separate enhancer and silencing activities have been mapped, the enhancer located ϳ10-kb upstream, and the silencer located within the first intron of the CD4 gene (55-57). The CD4 silencer is stage-and lineage-specific and, similar to FIG. 7. Analysis of RNA pol II binding in vivo to chicken ␤-globin gene promoters in control and mutant clones. Pol II binding correlated with expression status in early and late passage K562 clones. A, early passage (E) mutant K562 cells. Pol II binding at the ␤ H -globin promoter region was demonstrated in enhancer, 5Ј-HS1-and 5Ј-HS2-deleted clones but not in normal chromosome clones, correlating with gene expression. There was no pol II binding to the -, ␤ A -, and ⑀-globin gene promoters in the deleted clones, which also correlates with gene expression status. Control (FRK and HS1ϩK) K562 clones had pol II associated with the ␤ A -and ⑀-promoters, consistent with gene expression status. Late passage (L) mutant K562 cells. As in early passage cells, pol II binding at the ␤ H -globin gene promoter region was demonstrated in K562 clones containing the deleted chicken chromosome but not in K562 clones containing the control chromosome. However, coincident with the activation of -, ␤ A -and ⑀-globin transcription in late passage cells, there is now pol II binding at their promoters. B, quantitative analysis of data from A (see text and the legend to Fig. 4 for details).

TABLE I
Summary of results of the intergenic enhancer, 5Ј-HS1 or 5Ј-HS2 deletion on the chicken ␤-like globin genes FRK and HS1ϩK denote control chicken chromosome, and E-K E and E-K L, HS1-K E and HS1-K L, HS2-K E and HS2-K L denote early and late passage K562 containing the enhancer-deleted, 5Ј-HS1-and 5Ј-HS2-deleted chromosomes.
our results in the chicken ␤-globin locus, the epigenetic inheritance of the silenced state was not affected by inhibitors of DNA methyltransferases or histone deacetylases (58). The somatic cell system we have described will allow further investigation of these novel chromosomal silencers and their role in the regulation of the chicken ␤-globin domain.