We have identified a folate receptor gene upstream of the chicken β‐globin locus and separated from it by a 16 kbp region of silent chromatin. We find that this receptor is expressed only at a stage of erythroid differentiation (CFU‐E) preceding the activation of β‐globin genes, consistent with the role of folate receptors in proliferation. This discovery raises the question of how these two loci are regulated during erythropoiesis. Our data suggest that the folate receptor gene and the β‐globin locus are regulated independently. We show that a 3.3 kbp DNA region upstream of the folate receptor gene is sufficient to induce strong expression of a transgene in CFU‐E stage cells. We also find that the region between the β‐globin locus and the folate receptor gene is fully methylated and condensed at this stage of differentiation. Its 3′ boundary coincides with the 5′ β‐globin insulator. We speculate that the 5′ β‐globin boundary element might be important for the proper regulation of two adjacent domains activated at two different stages during differentiation.
Chromosomal position effects in mammals have been observed frequently in transgenic animals and in many diseases caused by natural mutations (Ton et al., 1991; Joos et al., 1992; Tommerup et al., 1993; Foster et al., 1994; Wagner et al., 1994; Fantes et al., 1995). These observations demonstrate that in addition to its role as a structural scaffold important for compaction, chromatin plays a major role in the utilization of genetic information. One attractive hypothesis is that chromosomes are subdivided into independent functional domains, which are either permissive or repressive toward gene expression. This model is reinforced by the observation that transcriptionally active genes are embedded in DNase I‐sensitive domains, extending many kilobases beyond the genes (Stalder et al., 1980; Lawson et al., 1982; Jantzen et al., 1986; Stratling et al., 1986; Levy‐Wilson and Fortier, 1989).
For a domain to be a truly independent unit, it should contain all of the regulatory elements necessary for the correct expression of the genes located within; some well characterized domains do maintain a higher level of tissue‐ and stage‐specific expression when integrated into the genome than constructs containing only proximal cis‐regulatory elements (Greaves et al., 1989; Bonifer et al., 1994; Mason et al., 1995; Dillon et al., 1997). DNA sequences that induce efficient gene expression independently of the site of integration have been identified in a number of different domains. Usually they are marked by a series of DNase I hypersensitive sites. Such clusters of regulatory elements required to induce copy‐number‐dependent expression have been designated as locus control regions (LCRs), and have been most thoroughly investigated in the human β‐globin gene locus (Forrester et al., 1987, 1989; Grosveld et al., 1987; Festenstein et al., 1996).
The β‐globin LCR, also studied in mouse and chicken (Moon and Ley, 1990; Hug et al., 1992; Mason et al., 1995), can activate genes located as far as 50 kbp away, and is responsible for the proper expression of the β‐globin genes (Epner et al., 1998; Reik et al., 1998). This raises the problem of how activation can be confined to a particular set of genes within the domain without affecting adjacent genes. It has been proposed that DNA sequences at the boundaries of an active chromatin domain may help in their establishment and maintenance (Geyer, 1997). The chicken β‐globin locus spans 33 kbp (Figure 1A), with boundaries identified by transitions to DNase I inaccessibility and low levels of histone acetylation (Hebbes et al., 1994). At the 5′ boundary, a DNA sequence element marked by a constitutive DNase I hypersensitive site (HS4) has been shown to possess the properties of an insulator; it blocks enhancer action, but only when it lies between the enhancer and a promoter (Chung et al., 1993).
We analyzed the region upstream of the well defined 5′ boundary of the chicken β‐globin locus, and found a region of condensed chromatin extending without interruption 16 kbp upstream of the β‐globin 5′ boundary. Immediately beyond this we discovered a novel locus that is activated during erythropoiesis. This locus encodes a folate receptor and has a different program of expression from the globin genes. The chicken folate receptor is expressed in erythroid precursor cells in the developing embryo and is repressed during terminal differentiation. We identify a strong regulatory element upstream of the folate receptor which by itself directs expression of a linked gene, fairly independently of the chromosomal site of integration in an erythroid cell line. We also show that at the time when the folate receptor gene is activated, the region between the β‐globin locus and the folate receptor locus is organized in a micrococcal nuclease‐resistant chromatin structure, and the DNA of this region is fully methylated. Our results indicate that the β‐globin and the folate receptor domains are regulated independently despite their proximity. We propose that the 5′ β‐globin insulator and the condensed region are important in assuring the proper expression of these two loci by establishing two units of independent gene activity.
Identification of an active erythroid‐specific locus 16 kbp upstream of the β‐globin locus
It has been demonstrated previously that for at least 4 kbp upstream of the chicken β‐globin locus, chromatin is DNase I‐resistant and its histones are hypoacetylated (Hebbes et al., 1994). This study defined the beginning of a new condensed chromatin region. The first step in understanding how adjacent loci are organized on the chromosome was the identification of the 5′ boundary of this condensed region. We searched for local perturbation in chromatin structure that might indicate the presence of active loci by mapping DNase I hypersensitive sites (DHS). The proximal DHS (designated HSA) was found 16 kbp upstream of the insulator in 11‐day embryonic erythrocytes, or red blood cells (RBCs) (Figure 1B and C, 11 Day RBC), which are transcriptionally active and expressing the βA and βH globin genes (Hesse et al., 1986; Mason et al., 1995). To determine whether HSA is developmentally regulated, DHSs were mapped in other cells. In contrast with all the sites observed within the β‐globin locus, HSA is not present in terminally differentiated adult erythrocytes (Figure 1C, Ault RBC) or embryonic brain (Figure 1C, Day Brain). Thus, in contrast to HS4, the hypersensitive site that marks the 5′ boundary of the β‐globin locus, HSA is not a constitutive DHS.
The regulation of HSA was also analyzed in two virally transformed chicken erythroid cell lines: HD24 cells are arrested at the multipotent erythroid‐myeloid precursor cell (MEP) stage; 6C2 cells are arrested at the colony forming unit erythrocyte stage cell line (CFU‐E) (Figure 2A). These cell lines are model systems that have been exploited in numerous studies to examine chromatin structure and the regulation of erythroid‐specific genes at early developmental stages (see Discussion). HSA is present in both of these cell lines (Figure 2C). The location of HSA was determined more precisely by utilizing a different probe and restriction digest combination (Figure 2D). These observations define a 16 kbp region which is bordered in erythrocytes by two hypersensitive sites: HSA and the insulator, HS4 (Figure 1B). The 5′ end of this region is delimited during erythropoiesis by HSA. We thought that its presence could signal the beginning of a new active domain, suggesting that it could be a regulatory element for a gene important during erythroid differentiation. We sequenced a cloned 6.3 kbp fragment that contains HSA near its 3′ end (Figure 1B, BamHI clone), and found that it has considerable identity (see below) to known human genes of the folate receptor family (Shen et al., 1994). It is very likely, therefore, that these sequences encode a chicken folate receptor. The orientation of this novel coding region is opposite to that of the β‐globin genes.
To confirm erythroid‐specific expression of this folate receptor gene, we performed wholemount in situ hybridization on 1‐ to 3‐day‐old chick embryos. The first erythrocyte precursors are detectable in the blastoderm after ∼24 h (BFU‐E stage), later developing into colonies of rapidly proliferating cells, known as blood islands, that precede terminal differentiation (Sabin, 1920; Lucas and Jamroz, 1961; Samarut et al., 1979). Expression of folate receptor mRNA was restricted to yolk sac blood islands, clearly recognizable by their characteristic cobblestone appearance when stained with Giemsa (Sabin, 1920) (Figure 3). No significant mRNA expression was detected in either pre‐blood island blastoderm or circulating embryonic red cells.
Using Northern blot analysis, we found that transcripts of this gene are detectable only in 6C2 cells (Figure 4A), consistent with the observation in embryos and erythrocytes that the expression of the folate receptor gene is tissue‐ and stage‐specific. In order to characterize this coding region fully, we identified the site of initiation of transcription by RNase mapping. We used two different overlapping probes to analyze transcripts in the erythroid lines HD24 and 6C2 and in erythrocytes (Figure 4A, 11 Day RBC), loading a quantity of RNA for each sample corresponding to the same number of nuclei (see Materials and methods). RNase mapping with the two probes identified the same site for transcriptional initiation and also confirmed that expression of this gene is only detectable in 6C2 cells (Figure 4B), in contrast to the presence of HSA in both cell lines. The alignment of the chicken DNA sequence with the three human known isoforms of the folate receptor and the use of the GT–AG rule allowed a prediction of the exon–intron junctions (Figure 4C). These junctions were confirmed by reverse transcriptase polymerase chain reaction (RT–PCR) on the total RNA of 6C2 cells (data not shown). We found a polyadenylation signal downstream of the stop codon of the coding sequence. The sum of the lengths of the four exons leads to a predicted 1.9 kb transcript, which is the size observed in the Northern blot analysis (Figure 4A and C). The protein is composed of three regions: a conserved central region, and variable N‐ and C‐terminal regions. The identity found in the conserved region between the different human isoforms and the chicken gene is between 65 and 70% (Figure 5A). One of the human isoforms corresponds to a truncated form of the folate receptor (FRγ), which is not bound to the cellular membrane. The other two isoforms (FRα and FRβ) possess a signal for attachment to a glycosylphosphatidylinositol (GPI) membrane anchor at their C‐termini (Shen et al., 1994). As can be seen in Figure 5B, the C‐terminal region of the chicken folate receptor contains such an attachment signal sequence, suggesting that it is anchored to the extracellular membrane.
In summary, we have identified a gene 16 kbp upstream of the 5′ boundary of the β‐globin locus that is expressed at a specific stage of erythroid development. Terminal differentiation is accompanied by the activation of globin genes with the repression of this folate receptor gene. The proximity of two loci and their activation in the same developmental pathway but at two different stages of differentiation suggests that complex controls of gene regulation are employed which are not yet understood.
A 3.3 kbp region upstream of the folate receptor gene contains regulatory elements sufficient to induce strong expression in 6C2 cells
The proper developmental regulation of the chicken β‐globin locus (33 kbp) necessitates a complex combination of upstream regulatory elements marked by upstream hypersensitive sites (HS1‐3) and the β/ϵ enhancer in addition to globin gene promoters (Mason et al., 1995). This model implicates regulatory elements that function over long distances (e.g. the distance between HS3 and the ϵ promoter is 20 kbp; see Figure 1A). The identification of a gene located 16 kbp upstream of the β‐globin domain and activated just before β‐globin expression during erythropoiesis raises the question of whether the folate receptor shares regulatory elements with β‐globin genes. It is equally possible that the insulator and condensed chromatin lying between these regions may generate a physical shield, causing the two loci to be regulated independently. The identification of potent regulatory elements located proximally to the folate receptor gene would suggest that these, and not the β‐globin elements, regulate folate receptor expression. We searched for additional regulatory elements downstream of the folate receptor gene, but no DHSs were identified in 6C2 cells within 5.5 kbp downstream of the polyadenylation signal (data not shown). Thus, the only detectable hypersensitive site near the folate receptor gene is HSA.
We decided, therefore, to fuse only the upstream region of the gene, which contains HSA, to a luciferase reporter gene. The fusion contained 3.3 kbp upstream of the folate receptor transcriptional start site and 0.4 kbp downstream (Figure 6). The linearized plasmid containing the promoter and the luciferase reporter gene was stably transfected into 6C2 cells. The number of integrated copies and the luciferase activity were determined for each resulting independent clone (see Materials and methods; Figure 6). All of the clones analyzed expressed high levels of luciferase activity, suggesting that most of the regulatory elements required for transcription of the folate receptor gene were included in the construction.
We asked whether the level of expression obtained was directly related to the number of copies integrated into the genome of 6C2 cells. A good correlation between gene copy number and luciferase expression is observed. The regression line produced is shown in Figure 6 and has a correlation coefficient of 0.85, which is statistically significant at a confidence level of >99%. This compares favorably with similar assays for copy‐number‐dependence of expression in hematopoietic cell lines (e.g. see May and Enver, 1995). The copy‐number‐dependent expression induced by this sequence suggests that it contains powerful regulatory elements, although it should be pointed out that in all stable transformation experiments of this kind, the construct may have preferred sites of integration within the genome that support greater activity. Furthermore, we obviously may have excluded additional regulatory elements that may be located much further downstream of the gene. With these reservations, the results support the conclusion that this upstream region could be sufficient to direct proper expression of the folate receptor in situ.
A 16 kbp region located between the β‐globin insulator and the folate receptor is condensed in 6C2 cells
Overexpression of the folate receptor has been observed in several human tumor cells (Campbell et al., 1991; Coney et al., 1991; Weitman et al., 1992; Ross et al., 1994). It has been proposed that the folate receptor plays a role in cell proliferation. We find (Figure 3) that a folate receptor gene is expressed at a stage of cell proliferation which precedes the final differentiation and expression of globin genes during erythropoiesis in chicken embryos. The role of the folate receptor in proliferation suggests that proper repression of its gene may be crucial for differentiation. Because the establishment of a condensed region between the β‐globin locus and the folate receptor gene could be necessary for the repression of the folate receptor gene, we investigated the structural properties of the region between these two loci.
The whole region between HS4 and HSA was cloned (Figure 1B). DNA sequencing revealed that 30% of the DNA contained in the 9.9 kbp HindIII clone is repeated sequence, principally the CR1 repeat commonly found in the chicken genome (Stumph et al., 1981). We did not identify any known open reading frames. Previous studies have demonstrated that the 3′ end of this 16 kbp region possesses the characteristics of repressed chromatin; the 4 kbp located immediately upstream of the insulator (HS4) contain hypoacetylated histones and the chromatin in this region is more resistant to DNase I digestion than the chromatin contained in the β‐globin locus in immature erythrocytes (Hebbes et al., 1994). The absence of DHSs and the high concentration of repeated sequence in this 16 kbp region suggests that a repressed chromatin structure might extend further upstream, thus defining a 16 kbp condensed domain at this stage of differentiation.
Further structural analysis of this domain was carried out in 6C2 cells. The nuclease sensitivity of regions flanking HSA, inside and downstream of the folate receptor gene, and toward the center of the 16 kbp region were examined using micrococcal nuclease (MNase). Genomic DNA was purified after MNase digestion of 6C2 nuclei, blotted and hybridized successively with four different probes (probes MN1, MN2, MN3 and P1; Figure 7A). Inspection of the digestion patterns shows that the DNA surrounding the folate receptor gene (probes MN1, MN2 and MN3) is more sensitive than the DNA located towards the center of the 16 kbp region (probe P1) (Figure 7B). Sensitivity is greatest within the active folate receptor gene (probe MN2): at the last point of digestion, the amount of DNA detected as mononucleosomes is higher than the DNA, which migrates at ∼1.3 kb. In contrast, for the same level of digestion with probe P1, DNA detected as mononucleosomes is as abundant as 1.3 kbp DNA. The digestion profiles observed with probe MN1 also show that the DNA located downstream of the folate receptor gene is more resistant than that located inside and upstream of HSA (probes MN2 and MN3).
These results suggest that in cells arrested at the CFU‐E stage, even in the absence of activation of the globin genes, the active locus containing the folate receptor gene is separated from the inactive β‐globin domain by a region condensed into MNase‐resistant chromatin. The observation of condensed chromatin between the folate receptor locus and the β‐globin locus in these cells is another indication that the β‐globin LCR is not implicated in the expression of the folate receptor gene.
We also found that HS3 is already formed in 6C2 cells, and that HS1 and HS2 are not detectable in contrast with erythrocytes, where HS1, HS2 and HS3 which define the LCR are present (Figure 8). The DHS over the β/ϵ enhancer is also present in 6C2 cells, but it is weaker than in erythrocytes (Boyes and Felsenfeld, 1996). These data reinforce the hypothesis that the β‐globin LCR is not involved in the activation of the folate receptor gene in 6C2 cells.
Establishment of a fully unmethylated domain
Another important transition which is also found at the 3′ boundary of the condensed region (5′ boundary of the β‐globin locus) is a progressive increase in DNA methylation as one proceeds into the condensed region in the 5′ direction (Chung et al., 1997). We examined the methylation of sites at the 5′ end of this region, near the folate receptor. Methylation of the first HpaII site, located 250 bp downstream of HSA (site 1, small arrow numbered 1 in Figure 9A), was analyzed in different tissue and cell lines by HpaII–MspI restriction analysis. In all the erythroid tissue and cell lines tested this site is cleaved by both enzymes, which shows that it is unmethylated (Figure 9B, lanes 2, 3, 5, 6, 8, 9, 11 and 12). In contrast to the disappearence of HSA, this unmethylated state is maintained in terminally differentiated erythrocytes (Figure 9B, Adult RBC). The site is unmethylated in all erythroid cells; in 11‐day brain and in a lymphoid cell line (DT40) it is methylated (Figure 9B, lanes 14 and 23). Methylation of the next HpaII site (site 2, small arrow numbered 2 in Figure 9A, located 1.2 kbp downstream of HSA, was also analyzed. This site is strongly methylated in HD24 (Figure 9C, lane 2), unmethylated in 6C2 (Figure 9C, lane 5) and methylated in 11‐day and adult erythrocytes (Figure 9C, lanes 8 and 11). These results indicate that absence of methylation at this site is precisely correlated with folate receptor expression. In contrast, the HpaII site 1 located proximally to HSA is unmethylated in all erythroid cells tested, but not in non‐erythroid cell lines [even in DT40, where HSA is detectable (data not shown)]. Since the DNA is unmethylated at this site in erythrocytes (adult RBCs), where HSA does not exist, it is clear that there is no direct correlation between DNA methylation and DHS formation at this locus.
It is known that high levels of methyl‐CpG correlate with transcriptional inactivity and nuclease resistance in chromosomes (Antequera et al., 1989). Our previous results indicated that at the CFU‐E stage (6C2 cells), the domain between HS4 and HSA is condensed. DNA methylation around the folate receptor has a tissue‐ and stage‐specific pattern, suggesting that there is precise control of methylation at this locus during erythropoiesis. As DNA methylation could be a mechanism for establishment or maintenance of condensed chromatin, DNA methylation was also analyzed along the domain between HS4 and HSA. HpaII sensitivity was measured in both directions from the center of the 16 kbp region. Over this region, a distinct characteristic pattern of DNA methylation is displayed in 6C2 cells. The DNA is completely methylated over this entire region, yet becomes unmethylated close to HS4 and HSA (Figure 10, lanes 3 and 10). In contrast, in both immature erythrocytes (11‐day RBCs) and mature, adult erythrocytes (Figure 10, lanes 5 and 12), only a fraction of the sites is methylated along the domain. Partial HpaII digestion is observed at every site (Figure 10, lanes 4 and 11), but the boundaries are maintained; HpaII sites close to HSA and HS4 are still fully unmethylated. As a control, the DNA methylation of this region was also analyzed in the embryonic chick brain. Two main features were observed. As in erythrocytes, only a fraction of HpaII sites is methylated along the 16 kbp domain, and HpaII sites close to the insulator are fully unmethylated (Figure 10, lane 14). In contrast to the erythroid cells, however, the DNA close to HSA is methylated, as shown by the presence of fragments derived from HpaII cuts beyond HSA (Figure 10, lane 7). This result is in agreement with the results presented in Figure 9.
This study in 6C2 cells defines a methylated segment of DNA delineated by two unmethylated boundaries. In order to look at the accessibility of these boundaries in the nucleus, we tried to excise this piece of chromatin from 6C2 nuclei with a simple HpaII restriction digest. The prediction was that the unmethylated HpaII site close to HSA should be accessible, because at this stage this DHS is already formed but we did not know whether the chromatin close to the insulator was accessible in these cells. A discrete DNA fragment is indeed excised from 6C2 nuclei by HpaII digestion (Figure 11, lane 1). The cutting of chromatin occurs at both ends very close to HS4 and HSA (Figure 11, lanes 2 and 5). This result suggests that a fully methylated domain is surrounded by an active locus at its 5′ boundary and that its 3′ boundary is also already established. We found that HS3 and HS4 are already formed; HS1 and HS2 are not detectable (Figure 8). The DHS over the β/ϵ enhancer is also present in 6C2 cells but it is weaker than in immature erythrocytes (Boyes and Felsenfeld, 1996). All these results suggest that the β‐globin domain may already be pre‐set at the CFU‐E stage, reinforcing the idea that the 5′ β‐globin insulator and the condensed region that we have identified might be crucial for the proper independent regulation of the folate receptor and the β‐globin locus.
Expression of the folate receptor at a proliferative stage
The purpose of this study was to define new chromatin regions upstream of the well characterized chicken β‐globin domain and its 5′ insulator. We hoped that this analysis would provide some insight into the role of the boundary element in gene regulation. In the course of this study, we found a 16 kbp silent region upstream of the globin locus, and beyond that a novel gene that codes for a folate receptor and is expressed during erythroid differentiation. We first explored the pattern of expression of this gene in 1‐ to 3‐day‐old developing chick embryos, and found that it was confined to the blood islands (Figure 3); precursor cells at this stage do not express β‐globin genes (Groudine and Weintraub, 1981). Folate receptor gene expression was not observed, however, in circulating red cells isolated from later stage embryos or adults, which do express β‐globin genes. This pattern of expression is consistent with the observation that a dietary deficiency in folate induces megaloblastic anemia, characterized by a prevalence of large, early stage hematopoietic precursor cells (Koury et al., 1997). Moreover, it is known that folate receptors are attached to the membrane of human RBC and it has been shown that addition of antibodies to placental folate receptor increases the proliferation of BFU‐E and CFU‐E in culture (Antony et al., 1987, 1991). Our observations provide the first evidence that the expression of a folate receptor gene is regulated precisely during erythroid differentiation. This observation also raises the question of how two proximal loci, folate receptor and β‐globin, activated at two different stages of erythroid differentiation, are regulated.
In order to study chromatin structure and DNA methylation patterns in cells that might express the folate receptor gene, we made use of two virally transformed avian erythroid precursor cell lines. These lines (6C2 and HD24) result specifically from transformation of BFU‐E stage chicken pre‐erythroid cells. Detailed analysis of surface markers has determined that 6C2 cells and other avian erythroblastosis virus transformed lines are arrested at the CFU‐E stage of development (Beug et al., 1982), whereas the HD24 line is blocked at an earlier multipotent erythroid‐myeloid precursor stage (Metz and Graf, 1991). We therefore analyzed methylation and chromatin structural patterns in these lines with the expectation that they would reflect the pattern of gene expression we had observed in embryonic erythroid development. For this reason, we concentrated our attention on 6C2 cells, and found that these CFU‐E stage cells alone expressed the folate receptor. Consistent with these results, expression was also detected in the rapidly proliferating cells of the yolk sac blood islands, but not in differentiated erythrocytes. Like the 6C2 cells, yolk sac CFU‐E cells do not express β‐globin genes (Groudine and Weintraub, 1981). The HD24 line (arrested at MEP) expresses neither of these genes, and experiments are underway to determine whether embryonic blood precursor cells follow this same pattern of folate receptor and globin gene expression in vivo.
Establishment of a condensed region between two erythroid loci
It has been demonstrated previously that for at least 4 kbp upstream of the chicken β‐globin locus, chromatin in 11‐day erythrocytes is DNase I‐resistant and packaged with hypoacetylated histones (Hebbes et al., 1994) (Figure 12). The establishment of a condensed region between the β‐globin locus and the folate receptor could be necessary for the proper repression of the folate receptor gene and/or the absence of β‐globin transcription at the CFU‐E stage. We find that in 6C2 cells, the region between the folate receptor locus and the β‐globin locus possesses properties of condensed chromatin (Figure 12). The DNA along this 16 kbp region is fully methylated, with boundaries that are unmethylated and accessible to nuclease digestion. In contrast, we find that this region is only partially methylated in erythrocytes, although it is known that at least 4 kbp of its 3′ part remains compacted (Figure 12). This suggests that when the folate receptor gene is active, the condensed DNA is maintained fully methylated, but that in differentiated cells in which it is inactive, DNA is randomly demethylated (note that adult RBCs are transcriptionally dead). It is not clear whether DNA methylation is a cause or effect of gene activation. It is possible, however, that DNA methylation along this domain is an important epigenetic mechanism for the maintenance of a condensed chromatin structure, especially given the known conection between DNA methylation and histone deacetylation in some systems (Jones et al., 1998; Nan et al., 1998). Once the condensed structure is established, perhaps the DNA can be randomly demethylated in quiescent cells without affecting the compaction of the DNA.
Recent results support this hypothesis. A mechanistic bridge between DNA methylation and histone deacetylation has been established in mammalian cells; it has been shown that at least one methyl‐CpG‐binding protein (the mouse protein MeCP2) associates with a corepressor complex that includes histone deacetylases (Jones et al., 1998; Nan et al., 1998). Such indirect recruitment of histone deacetylases by methyl‐CpG could be an efficient mechanism for re‐establishment of repressive chromatin structure after the passage of the replication fork. Our observations suggest that at the stage where the folate gene is expressed the entire region between the β‐globin domain and the folate receptor may contain hypoacetylated histones. We have already shown that in 6C2 cells, the boundaries of this region are both accessible to nucleases and unmethylated. Whether these boundaries also mark a transition in histone acetylation, from hypoacetylated to hyperacetylated histones, is currently being investigated. The presence of a condensed chromatin structure close to active loci raises the question of how compacted chromatin can be restricted to a precise region. We suggest that the insulator element previously identified at the 5′ end of the β‐globin locus may participate in establishing such a structural boundary at the 3′ end of the condensed region. This element, contained in a 1.2 kbp fragment which includes HS4, has been shown to function both to block activation of an enhancer by a promoter when placed between them, and in Drosophila and 6C2 cells to confer position independence of expression on a weakly expressed reporter (Chung et al., 1993; Pikaart et al., 1998).
The 5′ boundary of the condensed region is not yet as well defined as the 3′ boundary. However, we have identified a region upstream of the folate receptor gene sufficient to induce strong expression of a transgene in 6C2 cells, which shows a fairly linear dependence on copy number. This observation suggests that the region may contain regulatory elements sufficient to prevent chromosomal position effects. Further characterization of this region will tell us whether such a function is provided by an LCR‐like activity alone, or whether an insulating activity is also present. It is possible that repression of transcription of the folate receptor gene in erythrocytes involves the spreading of the condensed domain toward the folate receptor (Figure 12).
Boundaries and the role of the β‐globin insulator
The organization of the folate receptor and β‐globin gene cluster, and their programs of expression, suggests that there might be an important role for the β‐globin insulator in establishing and maintaining separate regulatory compartments. Recent studies in our laboratory (Pikaart et al., 1998) have shown that a stably integrated reporter gene can be protected from both positive and negative position effects if it is flanked on both sides by two copies of the 1.2 kbp β‐globin insulator element. This is consistent with a related role for the insulator element in its natural location at the 5′ end of the β‐globin locus, in protecting the folate receptor and β‐globin gene cluster from cross‐interaction. An analogous organization of two loci has been described in human T cells. T cell receptor α and δ gene segments are organized within a single genetic locus but are differentially regulated during T cell development. An enhancer‐blocking element has been identified between these segments within the locus and it has been proposed that this insulator functions as a boundary to separate the TCRα/δ locus into distinct regulatory domains (Zhong and Krangel, 1997).
Insulator elements also play a role in Drosophila development. Parasegment‐specific expression of the genes that comprise the Bithorax complex (BX‐C) is controlled by a cis‐regulatory region that spreads over 300 kbp of DNA. This large region is subdivided into nine parasegment‐specific cis‐regulatory subregions (for a review see Duncan, 1987). Mutations in the Mcp and Fab‐7 elements, which lie between particular pairs of parasegment cis‐regulatory subregions, affect control of expression in BX‐C, and consequently the identities of abdominal segments. This has been shown to arise from inactivation of an insulator function present in Mcp and Fab‐7 (Gyurkovics et al., 1990; Galloni et al., 1993; Karch et al., 1994; Mihaly et al., 1997). Mcp and Fab‐7 thus represent a distinct class of boundary elements; instead of separating adjacent domains that contain separate structural genes, they delimit cis‐regulatory domains.
The folate receptor and the β‐globin gene cluster are two loci that have different programs of expression during erythroid development. The folate receptor is expressed at an earlier proliferative stage than the globin genes, which are expressed during terminal differentiation. Throughout differentiation, both these loci display hypersensitive sites at their regulatory elements. In 6C2 cells arrested at the CFU‐E stage, we found both HSA at the folate receptor, and HS4 and HS3 in the globin LCR; it is also known that the hypersensitive site at the β/ϵ enhancer is present (Boyes and Felsenfeld, 1996). Furthermore, HSA is still detectable in immature erythrocytes (Figure 12). These two loci are separated by 16 kbp of condensed chromatin and by the insulator sequence. The distance between the loci is within the range over which LCR elements are capable of activating transcription. We propose that the insulator may be important for the independent regulation of the two flanking loci, serving to prevent inappropriate activation of genes on one side of the ‘barrier’ by transactivating elements on the other. This may be an activity separate from, but related to the boundary‐forming activity discussed above.
The proximity of two independently regulated erythroid‐specific loci separated by an insulator and a relatively short segment of condensed chromatin will be valuable for investigating the interplay of chromatin structure and gene expression.
Materials and methods
Cloning and sequencing of the region upstream of the β‐globin locus
Cloning of the upstream region was carried out by ‘chromosome walking’. Four partial libraries were made successively. For the construction of each library, Southern blot analysis was first used to determine an adequate restriction enzyme and the size of the genomic fragment after digestion. Genomic DNA was digested with appropriate restriction enzymes and then purified on an agarose gel. Fragments within ±500 bp of the size of the genomic fragment to be cloned were purified. After purification, partial libraries were generated by ligation in pBlueScript SK (Stratagene). For each library, 30 000 clones were screened. Isolation of the right clone was confirmed by DNA sequencing of the 3′ end of the clone and also by restriction digestion. Each clone was sent for DNA sequencing to Biotech Research laboratories (Rockville, MD). The sequence of the coding region was confirmed using the Thermo Sequenase system (Amersham).
DNase I, micrococcal and restriction endonuclease digestions in nuclei
For micrococcal nuclease digestion, 108 cells were washed twice in PBS, then resuspended in 1 ml of lysis buffer (10 mM Tris pH 7.5, 10 mM NaCl, 3 mM MgCl2, 0.4% NP‐40). Nuclei were pelleted and resuspended in the same buffer plus 1 mM CaCl2. Five microliters were mixed with 200 μl of NaCl 2 M/urea 5 M. The A260 was adjusted to 0.5. Six aliquots of 300 μl were digested with 0, 3, 6, 15, 30 and 60 U of MNase (Worthington) for 10 min at 37°C. The reaction was stopped with 60 μl of a 60 mM EDTA, 3% SDS. Samples were treated with 10 μg of RNase A at 37°C for 30 min, 100 μg of proteinase K was then added and samples were incubated at 55°C for 1 h. Samples were phenol–chloroform extracted three times, and chloroform extracted and then ethanol precipitated.
For HpaII digestion, 108 cells were washed twice in PBS, then resuspended in 1 ml of lysis buffer 10 mM Tris pH 7.5, 10 mM NaCl, 3 mM MgCl2, 0.2% Triton X‐100, 0.5 mM EGTA, 1 mM DTT and PMSF 0.4 mM, aprotinin 10 μg/ml. The sample was washed once again in the lysis buffer and then washed twice in the same buffer without Triton. Nuclei were pelleted and resuspend in 500 μl of a buffer (10 mM Tris pH 7.9, 10 mM MgCl2, 50 mM NaCl, 1 mM DTT), and 10 000 U of concentrated HpaII (New England Biolabs) were added. The sample was incubated 1 h at 37°C. The reaction was stopped by addition of EDTA to a final concentration of 15 mM. DNA was extracted and precipitated as described above.
For DNase I digestion, ∼108 cells were washed twice in cold PBS. Cells were lysed in 5 ml of buffer A (10 mM Tris–HCl pH 7.5, 10 mM NaCl, 3 mM MgCl2, 0.2% Triton X‐100, 1 mM DTT and 0.5 mM EGTA) plus PMSF 0.4 mM, aprotinin 10 μg/ml. Nuclei were pelleted for 5 min at 1000 g and washed once in buffer A without Triton. The pellet was resuspended in 1 ml of buffer B (10 mM Tris–HCl pH 7.5, 10 mM NaCl, 3 mM MgCl2, 1 mM CaCl2) at room temperature, and 100 μl of the nuclei preparation were used for each DNase I (Worthington) digestion. The final DNase I concentrations were 0, 0.8, 1.6, 2.4, 3.2, 4, 4.8 and 5.6 U/ml. Digestions were performed at room temperature for 5 min and stopped by addition of 20 μl of buffer C (10 mM Tris pH 8, 60 mM EDTA and 3% SDS). Samples were treated with RNase A and proteinase K as described above. DNA was extracted and precipitated as described above.
Southern blotting and hybridization
Ten micrograms of each DNA sample was digested with the restriction endonucleases indicated in the particular experiment. Following electrophoresis through agarose gel, the gel was treated for 15 min in 0.25 M HCl and then incubated in denaturing buffer (3 M NaCl, 0.4 NaOH) twice, for 30 min each. After 15 min incubation in transfer buffer (3 M NaCl, 8 mM NaOH), the gel was transferred to a Genescreen plus membrane (DuPont) in 3 M NaCl, 8 mM NaOH for 90 min using a TurboBlotter system (Schleicher and Schuell).
Hybridizations were performed using 50 ng of probes labeled by random priming. Filters were prehybridized for 30 min and hybridized for 90 min at 68°C using Quick Hyb solution (Stratagene). Following hybridization, filters were washed as follows: once in 2× SSC, 0.1% SDS at room temperature, once in 2× SSC, 0.1% SDS at 68°C for 30 min and finally once in 0.1× SSC, 0.1% SDS at 68°C for 30 min.
In situ hybridization
Antisense riboprobes were prepared using digoxygenin‐UTP (Boehringer Mannheim) and T7 RNA polymerase (Stratagene) from linearized RNase I plasmid. Sense riboprobes were prepared with the insert cloned in the opposite direction from the same vector (pCR2.1; Invitrogen). Fertilized eggs were incubated until the desired developmental stage. The whole blastoderm was harvested and fixed flat in 2 cm tissue culture plates in 4% paraformaldehyde overnight at 4°C. Embryos were washed in PBS then dehydrated gradually on ice in increasing concentrations of methanol to 100%, then rehydrated into PBS–0.1% Triton X‐100 (PBT).
Prehybrization was for 2 h at 65°C in hybridization buffer (50% formamide, 5× SSC pH 6.0, 100 μg/ml yeast tRNA, 100 μg/ml salmon sperm DNA, 50 μg/ml heparin) followed by hybridization overnight at 65°C with riboprobe. Embryos were washed extensively first in 2× SSC at 65°C then at 2× SSC at room temperature in PBT, then blocked in 5% sheep's sera (Gibco‐BRL) for 1 h, followed by a 2 h antibody incubation at RT (FITC‐anti‐Dig IgG; Boehringer Mannheim). PBT washes were monitored until background staining was negligible. Photographs were taken on an Olympus SZ12 microscope using Kodak 1600 ASA color film. After photography, embryos were counter stained with Giemsa (Sigma Accustain).
pGL3‐FR4 was constructed by coinserting two XhoI–BamHI fragments into pGL3‐basic linearized by XhoI (Promega). The XhoI–BamHI fragment linked to the luciferase reporter gene is located in the 3′ end of the 6.3 kbp BamHI clone. The other XhoI–BamHI fragment was excised from the 2.5 kbp BglII clone.
HD24 cells are chicken multipotent erythroid‐myeloid cells transformed by the E26 virus. They cannot differentiate as efficiently as primary transformants and express some markers for early erythroid progenitors. They were grown in blastoderm media (Metz and Graf, 1991). 6C2 are CFU‐E stage erythroid precursor cells, obtained by transformation of bone marrow with wild‐type avian erythroblastosis virus (AEV). 6C2 cells were grown in alpha modified Eagle's medium (MEM) supplemented with 10% fetal bovine serum (FBS), 2% chicken serum, 1 mM HEPES, 50 mM β‐mercaptoethanol and a standard complement of antibiotics. DT40 cells, purchased through American Type Culture Collection (ATCC, Rockville, MD), were grown in DMEM supplemented with 50 mM β‐mercaptoethanol, 2 mM glutamine, 10% FBS, 5% chicken serum, 10% tryptose phosphate broth and antibiotics. All the cells were maintained at 37°C in 5% CO2.
Cell culture and transfections
6C2 cells were grown, transfected and selected as described previously (Boyes and Felsenfeld, 1996). The test construct pGL3‐FR4 was linearized by either NheI or SalI. NheI cuts the vector just upstream of the 5′ end of the folate receptor upstream region; SalI cuts 3′ of the SV40 late poly(A) signal of the luciferase reporter gene. Both linearizations gave the same results. Individual colonies were picked after 2–3 weeks and expanded in medium containing 1000 U/ml hygromycin (Calbiochem), an amount that kills any non‐transfected cells. After cell growth for 4–6 days in 3 ml of fresh selective medium, half of the culture was spun down and resuspended in 500 μl fetal calf serum plus 10% DMSO and stored in liquid nitrogen. The rest of the culture was spun down and the pellet was treated for DNA extraction.
Determination of the copy number and transcriptional activity of 6C2‐transfected colonies
The cell pellet of each individual colony was resuspended in 250 μl of 400 mM NaCl, 200 mM Tris pH 8.5 and lysed by adding 250 μl of 0.4% SDS, 10 mM EDTA. The sample was incubated for 30 min at 37°C with 100 μg/ml RNase A and then with 100 μg/ml Proteinase K for 1 h at 55°C. Samples were extracted twice with phenol–chloroform and once with chloroform, and the DNA was precipitated with isopropanol. The DNA pellet was resuspended in 50 μl 10 mM Tris–HCl pH 7.5, 1 mM EDTA (TE). Ten microliters of the DNA sample was digested by HpaI and KpnI. Digested samples were analyzed by Southern blotting with probe RNase I. This probe detects a 4.9 kbp fragment for the transgene and a 7.5 kbp fragment for the endogenous folate receptor gene after HpaI and KpnI digestion. The ratio between the transgene copy number and the endogenous gene was quantified by Phosphorimage analysis. A ratio of 0.5 was identified as a single copy integration. This was confirmed by a restriction digestion which cuts once in the construct integrated. Clones stored in liquid nitrogen were cultured for 4–6 days in fresh medium containing hygromycin. For luciferase assays, cells from 1 ml of culture were harvested, washed twice in PBS and resuspended in 100 μl of Reporter Lysis buffer (Promega). The supernatants were assayed for luciferase activity using the Promega luciferase assay system. Luciferase activity was normalized to the protein concentration in each samples. Protein concentration was determined with Bio‐Rad protein assay kit.
RNA preparations and analysis
Total RNA was prepared using RNA STAT‐60 solution (TEL‐TEST) according to the manufacturer's instructions and was treated with RNase‐free DNase I (Boehringer Mannheim). RNA integrity of each preparation was checked on ethidium bromide stained 1% agarose/MOPS‐formaldehyde gels. RNA was prepared from HD24, 6C2 and DT40 chicken cell lines, circulating red blood cells of 11‐day embryos, chicken fibroblasts prepared by trypsin digestion of 11‐day embryo, brain (cerebral hemisphere) of 11‐day embryo. Fertilized White Leghorn chicken eggs were obtained from Truslow Farms, Chestertown, MD. The amount of RNA extracted per nucleus from 6C2 cells, HD24 and 11‐day RBCs was determined. Before extraction, cells were counted with a hemacytometer. RNA (4.7 pg) per nucleus, 7.2 pg of RNA per nucleus, and 0.6 pg per nucleus were extracted respectively from HD24 cells, 6C2 cells and 11‐day RBCs. For the RNase protection assay, 19.6, 30 and 2.5 μg of RNA from HD24, 6C2 and 11‐day RBCs, respectively, were hybridized.
RNA was analyzed by RNase protection assay using the RPA II kit (Ambion). Radiolabeled antisense transcripts were generated using T7 RNA polymerase (Stratagene) and [32P]CTP. RNA probes were acrylamide gel‐purified prior to hybridization. Hybridizations were performed overnight at 45°C. RNA was also analyzed by Northern blot. Thirty micrograms were loaded in each lane. The blot was hybridized with a 1.2 kbp XhoI–BglII fragment excised from 6.3 kbp BamHI clone.
P1: 404 bp, was generated by PCR amplification from the 9.9 kbp HindIII clone with oligonucleotides GATGGCTTCCCCAGCAGCAA and AATGGTCTTTGCCTGGGTCA.
P2: 566 bp, KpnI–XhoI restriction fragment excised from the 2.5 kbp BglII clone.
P3: 263 bp, BglII–BamHI restriction fragment excised from the 2.5 kbp BglII clone.
P4: 689 bp, EcoRV–KpnI restriction fragment excised from the 2.5 kbp BglII clone.
P5: 394 bp, NcoI–BamHI restriction fragment excised from pCBG D (Hebbes et al., 1994).
P6: 1062 bp, SacII–KpnI restriction fragment excised from pCBG C (Reitman and Felsenfeld, 1990).
MN1: 378 bp, was generated by PCR amplification from the 6.3 kbp BamHI clone with oligos CAGCTTCTGGAAATAACTGA and CTCTCCTGCACAGCTGTGGT.
MN2: 394 bp, was generated by PCR amplification from the 6.3 kbp BamHI clone with oligos CATGTCAGCATCCCACAAAG and CAGTGGTGTGGGTAATTGGA.
MN3: 377 bp, was generated by PCR amplification from the 2.5 kbp BglII clone with oligos CTGCTGATCACTGAACTGTG and TATGCTGGGAACAGCTATAC.
RNase I: 267 bp, was generated by PCR amplification from the 6.3 kbp BamHI clone with oligos ACCGTGCCTTCTCCTCCTCT and CTGTGTTGCTTTAATGCTGTCA. The PCR fragment was then cloned in pCR2.1 (Invitrogen) using the TA Cloning kit. Orientation of the insert was determined by restriction digestion.
RNase II: 295 bp, was generated by PCR amplification from the 6.3 kbp BamHI clone with oligos GCCTCCTGACACAGCAGAGC and GCAGGTTTGAGGCAAGTTAG. The PCR fragment was then cloned in pCR2.1 (Invitrogen) using the TA Cloning kit. Orientation of the insert was determined by restriction digestion.
We are grateful to Olivier Hyrien, Mike Litt, Felix Recillas‐Targa and Adam West for critical reading of this manuscript, and members of the Felsenfeld laboratory for helpful discussions. M.‐N.P. was supported by a long‐term EMBO fellowship and a postdoctoral fellowship from the Human Frontiers Science Project Organization.
- Copyright © 1999 European Molecular Biology Organization