Effects of the deletion of the human β-globin promoter on the locus

In our previous studies on the mouse and human β-globin locus, we have shown that the β-globin gene only interacts with the ACH when it is transcribed in definitive cells (Palstra et al. 2003). In differentiated primitive cells, the early genes (ε- and γ-) interact with the ACH, whereas in precursor cells, the LCR and the genes do not interact with the ACH. This poses the question of whether the genes need to associate with the ACH to be transcribed or whether the association with the ACH is the result of transcription. This question was addressed by deletion of the β-globin promoter without affecting any other parts of the locus. Two independent single copy transgenic lines were analyzed for expression of the ε-, γ-, δ-, and β-genes (Fig. 3A). The results show that early gene expression in the primitive erythroid cells (embryonic day 10.5 [E10.5] yolk sac) is not affected by the deletion of the β-globin promoter when compared with mice containing the wild-type locus. However, the level of γ-globin gene expression is higher in the E12.5 fetal liver. Similarly, δ-globin gene expression, when measured in adult blood cells, although still very low (note the scale in Fig. 3A), is higher compared with the wild-type locus. This result is in agreement with the notion that a single gene is expressed at any given time (Wijgerde et al. 1995) and that there is transcriptional competition between the genes (Dillon et al. 1997). When the chromatin structure of the Δβp locus was compared with the wild type locus both by DNaseI sensitivity and histone H3 acetylation (see Figs. ​Figs.5,5, ​,6,6, respectively [below]), the chromatin structure of the locus was essentially the same. A similar result was obtained for histone H4 acetylation (data not shown). We next asked the question of whether the β-globin gene still interacts with the ACH in the definitive differentiated erythroid cells when compared with control fetal brain cells. For this purpose, Chromosome Conformation Capture (3C) technology (Dekker et al. 2002; Tolhuis et al. 2002; Palstra et al. 2003) was used, which gives a measure of proximity of two DNA restriction fragments in the nuclear space using quantitative PCR of the cross-linking frequencies between the fragments in the nucleus. The assay is not biased for particular fragments, because of preferential restriction enzyme digestion of one site over the other as a result of different chromatin structure (Tolhuis et al. 2002; Palstra et al. 2003; Splinter et al. 2004). In the wild-type locus, the HS1–5 of the LCR, the transcribed gene, and the HS in the 3′ flanking region (3′HS1) interact with each other to form the ACH of the human β-globin locus in transgenic mice (Palstra et al. 2003). The result of the 3C analysis (Fig. 3B,C) shows that the spatial structure of the locus is essentially unchanged in the Δβp mutant locus. The ACH, consisting of the LCR and 3′HS1 (Palstra et al. 2003), is maintained. However, the relative frequency of association of the β-globin gene with the fragment containing the HS4, HS3, and HS2 of the LCR, but not the other regions, is reduced by 30% in the Δβp line when compared with the wildtype control. This reduction is presumably due to the absence of the factor-binding sites normally present in the promoter. In nonexpressing brain, both transgenic lines show an essentially linear conformation of the locus (Fig. 3B,C; data not shown; Palstra et al. 2003). Whether the remaining association of the β-promoterless gene is (in part) the result of γ- and δ-gene expression is presently not clear. In any case, we suggest that the association of the β-globin gene with the ACH is not the result of transcription and that it would normally precede transcription.

Open in a separate window

Figure 3.

The effect of the human β-globin promoter deletion on the locus. (A) S1 nuclease protection assays showing the up-regulation of the δ- and γ-globin genes in the absence of the human β-globin promoter in the adult and fetal developmental stages, respectively. S1 assays were quantitated by PhosphorImager analysis and corrected for copy number and probe specific activity. The levels of expression shown in the histogram on the right are per copy of the human globin genes, normalized to the level seen for the endogenous mouse-α globin gene. (B) Combined agarose gels of the PCR fragments resulting from the 3C ligation experiments. (R) Random control. (C) Histogram of the relative cross-linking efficiencies after quantitation and normalization of the gels shown in B. The histogram is the average of at least three separate experiments, each in duplicate.

Open in a separate window

Figure 5.

DNaseI accessibility assessed by DNaseI–PCR analysis. PCR product series of the human γ-globin promoter (A) and HS5 (B), in the wild-type, Δβp, Δ2Δβp, Δ3, and Δ3Δβp PAC transgenic mouse lines. An increase in the relative amount of the PCR product, normalized to the signal of mouse Zfp37 PCR products, reflects a decreased DNaseI sensitivity. Quantitation of DNaseI sensitivity in the human γ-globin gene promoter or HS5, relative to the mouse Zfp37 promoter, is shown below each panel. (C) Southern blot analysis to show the loss of hypersensitivity in the LCR of the Δ3Δβp. Sizes of the PstI restriction fragments and the fragments resulting from DNaseI digestion of HS1 and HS2 are indicated. Probes were PCR fragments from the 5′ end of the HS1 fragment and the 3′ end of the HS2 fragment. (D) Summary of the data obtained during chromatin hypersensitivity mapping in various sites in the human β-globin locus and the LCR, as determined by DNaseI PCR assays in the transgenic lines analyzed. (A) Absence of HS due to the targeted deletion.

Open in a separate window

Figure 6.

ChIP assays. (A) The various regions within the human β-globin locus that have been analyzed for the acetylation status of histones H3 and H4. (B) Acetylation status of histone H3 throughout the human β-globin locus in the wild-type, Δβp, Δ2Δβp, and Δ3Δβp E12.5 transgenic fetal livers and brain. ChIP assays were carried out as described in Materials and Methods. The PCR products corresponding to the human β-globin locus were normalized against mouse Zpf37 promoter PCR products and inputs. The enrichment over Zfp37 is given below the panel. (I) Input; (N) no antibody control; (H3) acetylated histone H3 antibody.

Expression of the mutant Δ2Δβp and Δ3Δβp β-globin loci

The second aim of this work was to determine which of the most active sites of the LCR (HS2 and HS3) interact with the promoter of the β-globin gene, using double mutants. In addition, we wanted to know what effect these mutations would have on the 3-D structure of the locus. To this end, the HS2 or HS3 were deleted from the PAC, leading to the Δ2 and Δ3 lines (see earlier), and these deletions were combined with the deletion of the β-globin promoter, resulting in the Δ2Δβp and Δ3Δβp lines, respectively (see earlier). Like the Δ2 and Δ3 lines that have been previously analyzed by several laboratories as part of cosmid or YAC-based human β-globin loci (Bungert et al. 1995; Milot et al. 1996; Peterson et al. 1996), or in the endogenous mouse locus (Fiering et al. 1995; Hug et al. 1996), the deletions in our PACs result in a transcriptional loss of the β-globin gene of ~30% at the definitive stage (Fig. 4A), unless the constructs were integrated in heterochromatin, resulting in position effect variegation (data not shown; Milot et al. 1996). The double mutants Δ2Δβp and Δ3Δβp were first analyzed at the adult stage for δ- and β-globin gene expression (Fig. 4A). This shows that the level of δ-globin gene has increased in all four Δ2Δβp lines (single or multicopy) when compared with the levels observed in the wildtype locus and is very similar to what is observed with the single Δβp mutant locus. Thus, the additional ΔHS2 mutation has no effect, suggesting that HS2 normally interacts with the promoter of the β-globin gene at the adult stage. This would be in agreement with its role as a classical transcriptional enhancer. Surprisingly, the expression of the δ-globin gene in the Δ3Δβp double-mutant transgenic lines is completely extinguished. To ensure that the double Δ3Δβp mutant is not completely defective, we carried out timed mating for two lines of each construct.

Open in a separate window

Figure 4.

S1 nuclease protection assays. S1 nuclease protection assays are shown for one transgenic line for the Δβp deletion constructs in the context of a full or incomplete LCR at the adult (A), fetal (E12.5, B), and embryonic (E10.5, C) developmental stages (see also Materials and Methods). Two littermates for each developmental stage were analyzed, using ³²P-labeled probes for mouse-α and human ε-, γ-, δ-, and β-globin genes. Each of the protected fragment positions is indicated. S1 assays were quantitated by PhosphorImager analysis and corrected for copy number and probe specific activity. The values are per copy levels of expression of human globin genes, as a ratio of the endogenous mouse-α globin gene and shown as a histogram on the right. The histogram shows the analysis of all transgenic lines for each construct. Each experiment has been done in triplicate.

The results (Fig. 4B,C) show that ε- and γ-globin gene expression at the embryonic stage and the early fetal liver stage for the Δ2Δβp transgenic line is very similar to that observed for the Δβp line. At the early fetal liver stage, γ-globin gene expression has approximately doubled when compared with a wild-type locus because of the absence of competition of the β-globin gene. This was confirmed by primary transcript in situ hybridization analysis (Wijgerde et al. 1995; data not shown). In contrast, the ε- and γ-globin gene expression levels were the same as in the wild-type locus at the embryonic stage in primitive cells. At this stage, the β-globin gene is not expressed and hence the competitive effect is absent.

Interestingly, the expression of the γ-globin genes in the double mutant Δ3Δβp is completely absent in early fetal liver (Fig. 4B, confirmed by in situ primary transcript analysis, not shown). However, they are expressed at the embryonic stage (Fig. 4C), although the level has been decreased by ~40%. An even bigger decrease is observed for ε-globin gene expression. This result shows that the locus can be expressed in primitive, but not definitive, cells and suggests that HS3 interacts with sequences other than the basic promoter of the β-globin gene, for example, the more upstream part of the promoter, the enhancer just downstream of the gene, or sequences much farther away from the β-globin gene (for review, see Stamatoyannopoulos and Grosveld 2001). The result also suggests that the Δ3Δβp double mutation may completely destabilize the normal chromatin structure of the locus in definitive cells. We therefore carried out DNaseI sensitivity assays of the double- and single-mutant transgenic lines.

DNaseI hypersensitive site analysis

Globin gene transcription is normally correlated with the presence of DNaseI HSs at the promoter of the expressed genes and the LCR (Tuan et al. 1985; Grosveld et al. 1987). In order to determine the effect of the targeted deletions in the β-globin promoter alone and in combination with HS2 or HS3, we analyzed DNaseI hypersensitivity of the γ-globin promoter and of the individual HSs of the LCR in isolated nuclei from wild-type, Δβp, Δ2Δβp, Δ3, and Δ3Δβp E12.5 fetal liver cells, using DNaseI PCR (Fig. 5A,B; McMorrow et al. 2000) and Southern blots (Fig. 5C). The results show that the γ-globin gene remains DNaseI sensitive in all of the lines with the exception of the Δ3Δβp line. In this double mutant, sensitivity has become the same as observed in the control murine Zfp37 gene. The latter has a low DNaseI sensitivity (i.e., it is not insensitive, McMorrow et al. 2000) in the early fetal liver (Fig. 5A). The hypersensitivity of all of the other HSs has been maintained in all of the lines with the exception of the Δ3Δβp line (Fig. 5B–D). Surprisingly, in this line all hypersensitivity to DNaseI digestion has been lost. We conclude that this is due to the double mutation and not just the absence of HS3, because in the Δ3 single mutation line, HS5, HS4, HS2, and HS1 are still present (Fig. 5B [green line], D). The loss of all hypersensitivity suggests that the chromatin structure of the Δ3Δβp double mutant locus has been severely affected and that the entire ACH structure might have been lost in this double mutant.

Deletion of the β-globin gene promoter

A human β-globin promoter deletion was generated in the context of the full LCR to explore the interactions between the LCR and the β-globin gene and whether or not the interaction of the gene with the ACH is the result of transcription. We have recently shown for the mouse locus that committed precursor cells, which do not yet transcribe the β-globin genes, already show an interaction between the most 5′ part of the LCR and the HSs flanking the locus (Tolhuis et al. 2002; Palstra et al. 2003). Interestingly, all of these elements contain CTCF-binding sites (Farrell et al. 2002; Bulger et al. 2003). When the cells differentiate, the full LCR and the expressed genes participate in the interaction, forming the ACH and the active transcriptosome. However, those results do not show whether the interaction of the globin genes with the ACH is required for transcription or whether the association with the ACH is simply the result of transcription.

The deletion of the β-globin promoter does not lead to any appreciable change in the expression pattern of the ε- and γ-globin genes in primitive cells when the β-globin gene is normally not transcribed. However, in definitive cells, when the β-globin gene is normally transcribed, it results in an increase of transcription of the γ- and δ-globin genes, in agreement with the earlier results, showing that the genes are transcribed competitively with a single gene expressed at any particular moment (Wijgerde et al. 1995 and references therein), and the total output of the locus remains the same when another β-globin gene is added (Dillon et al. 1997; Wai et al. 2003). Such a mechanism suggests that the ACH would only be able to bind a single gene for transcription. Deletion of the β-globin promoter leads to a reduction in the frequency of interaction of the β-globin gene with the LCR in the ACH (~30%), but importantly, a substantial interaction remains. Unfortunately, our experiments do not show to what extent the remaining interaction is due to the interactions of the δ- and γ-genes or whether the remaining interaction of the β-globin gene still competes with those of the γ- or δ-globin genes with the ACH. In any case, the results of the deletion of the β-globin promoter strongly suggest that other (nonpromoter) interactions take place between the ACH and the β-globin gene (see following). In addition, the results suggest that the interactions with the ACH or the stability of the ACH is not dependent on transcription of the β-globin gene per se and that the association of the β-globin gene with the ACH is also not the result of its transcription. Instead, we suggest that this association may be a requisite for transcription and could provide an explanation for proposals like the LCR being involved in loading the polymerase on the promoter (Johnson et al. 2001).

Deletion of the β-globin promoter combined with LCR HS2 and HS3 deletion

The β-globin promoter deletion was subsequently combined with the deletion of either HS2 or HS3 in cis, because previous data have implicated these HSs as functionally the most important in the LCR; that is, HS2 has the greatest enhancing activity and HS3 the most efficient chromatin opening activity (Ellis et al. 1996; Milot et al. 1996).

There was no significant difference in γ-globin gene expression between the wild-type and the Δβp and Δ2Δβp transgenic lines during primitive erythropoiesis. In contrast, in the absence of HS3, that is, in the Δ3 and Δ3Δβp lines, γ-globin expression is reduced, when compared with the wild-type PAC transgenic line. This decrease in γ- (and more severe decrease in ε-) globin gene expression in primitive cells of the Δ3Δβp lines is very similar to what is seen here and published previously for deletion of HS3 only (Milot et al. 1996; Navas et al. 1998). The fact that there is no difference between the Δ3 and Δ3Δβp lines confirms that HS3 is important in the primitive expression pattern.

The situation is very different in definitive erythroid cells. The absence of the β-globin promoter in conjunction with the HS2 deletion leads to very similar levels of expression of the γ- and δ-globin genes when compared with the β-globin promoter deletion alone. This suggests that HS2 normally interacts primarily with the β-globin globin gene promoter. If HS2 were to primarily interact with sequences other than the β-globin promoter (e.g., the remainder of the gene or its flanking sequences or the γ- or δ-globin gene promoter or yet other sequences), these “other” interactions would be lost with the HS2 deletion and would probably result in a decrease of transcription of the γ- and δ-globin genes. Although it has been suggested that HS2 and the β-globin promoter interact on the basis of the RNA TRAP analysis (Carter et al. 2002) or the 3C analysis (Palstra et al. 2003), these techniques currently do not have sufficient resolution to distinguish the different parts of the LCR and/or the globin genes to actually confirm an HS2/β-globin promoter interaction.

In contrast, the absence of the β-globin promoter in conjunction with the absence of HS3 has a dramatic effect on the expression of the γ- and δ-globin genes in definitive cells. The transcripts are no longer detectable and all DNaseI sensitivity is lost. This suggests that HS3 normally interacts with sequences other than the β-globin promoter and that this interaction is important for transcription of the globin genes and maintenance of the ACH. For the same resolution reasons as described for HS2, it is presently not clear what these sequences would be, although in the case of the β-globin gene, its 3′ enhancer would be a candidate for such an interaction (Liu et al. 1997).

Transgenic mice

Purified 185-kb NotI fragments were microinjected into the pronuclei of FVB fertilized mouse oocytes and transferred into the oviducts of pseudopregnant F1 (CABxC57B1) female mice. Transgenic founders were identified by Southern blotting of DNA isolated from tail biopsies and probed with cosmids (cos) LCRε and cosγγδβ, containing the 5′ and 3′ parts of the human globin locus, respectively (Strouboulis et al. 1992). In addition, PCR reactions were performed to verify the deletion junction points. Those founders that contained all of the characteristic EcoRI fragments for the entire locus were processed for breeding and further analysis. Copy numbers were determined by quantitation of Southern blots via PhosphorImager analysis. Internal probes were also used to verify the structure of the transgenes.

RNA isolation and S1 nuclease protection assay

RNA was isolated from frozen transgenic E10.5 embryonic yolk sacs, E12.5 fetal livers, and adult blood and subjected to S1 nuclease protection assays as previously described (Wijgerde et al. 1996). S1 probes were end-labeled with T4 polynucleotide kinase and equimolar amounts were used per reaction. Signal quantitation was performed using a PhosphorImager (Molecular Dynamics).

DNA and primary transcript fluorescence in situ hybridization (FISH)

Peripheral blood cells were cultivated for 24 h in RPMI 1640 medium (GIBCO-BRL) and chromosome preparations were made according to standard procedures. DNA FISH was carried out as described in Mulder et al. (1995), using the human LCR probe labeled with digoxigenin and immunohistochemically detected with fluorescein. DNA was counterstained with DAPI.

Primary transcript in situ hybridization to detect transcription of the human and mouse globin genes in the yolk sac and fetal liver cells was performed as described previously (Wijgerde et al. 1995). Quantitation of transcriptional signals was done by counting at least 300 cells from each slide with a fluorescence microscope.

Isolation of nuclei and DNaseI hypersensitive site analysis

Nuclei suspension from six transgenic E12.5 fetal livers for each line was performed by 20 strokes of a Dounce pestle (type B) and 100-μL aliquots were digested with increasing amounts of DNaseI at 37°C for 3 min. Reactions were stopped and treated with proteinase K and extracted with phenol/chloroform. After ethanol precipitation, DNA was resuspended in water in a concentration of 10 ng/μL.

PCR: Radioactive PCR was performed to determine DNaseI sensitivity of human γ- and β-globin gene promoter and the individual LCR HSs, using 10 ng of template DNA as previously described (McMorrow et al. 2000). Primer sets were as follows: human γ-globin gene promoter (forward, 5′-GCCTTGACCAATAGCCTTGACA-3′; reverse, 5′-CTGGACTAGGAGCTTATTGATAAC-3′; fragment size, 171 bp), HS5 (forward, 5′-CTGACCTATATCTGGCAGGACTC-3′; reverse, 5′-GGAAACTGGAGAAACTGGGA-3′; fragment size, 205 bp), HS3 (forward, 5′-ACTTTGCGAGCTGGTGTGC-3′; reverse, 5′-ATCTCTCCTGGCTCCTGGGA-3′; fragment size, 180 bp), HS2 (forward, 5′-CCCTCTATCCCTTCCAGCATC-3′; reverse, 5′-TTGAGCCAGAAGGTTTGCTTA-3′; fragment size, 200 bp), HS1 (forward, 5′-TTCTGTGCTTTCCTGGCTG-3′; reverse, 5′-ATGACGGGTTGATGGGTG-3′; fragment size, 210 bp). Primer sets and product sizes for HS4, human β-globin promoter, mouse βmaj, and Zfp37 and PCR conditions have been published previously (McMorrow et al. 2000). Each reaction has been done in triplicate and band intensity has been quantitated using Phosphor-Imager analysis and normalized against either mouse Zfp37 or βmaj signal.

Southern blots: DNaseI-treated DNA was digested with HindIII or PstI, run on 0.7% agarose gels, blotted, and hybridized with PCR-generated probes from the ends of the restriction fragments corresponding to the different hypersensitive sites.

ChIP

Five transgenic E12.5 fetal livers and brains for each line were incubated in DMEM, containing 1% formaldehyde and 10% fetal calf serum at 37°C for 10 min. Following washing with PBS, the cells were pelleted and resuspended in 200 μL SDS lysis buffer (Upstate Biotechnology) for 10 min on ice. Cell lysate was then sonicated to reduce DNA length to between 200 bp and 1 kb and debris was removed by centrifugation. ChIP was performed using the acetyl-histone H3 immunoprecipitation assay kit (Upstate Biotechnology) and acetyl-histone H4 immunoprecipitation assay kit (Upstate Biotechnology), according to the manufacturer's instructions. Enrichment for specific immunoprecipitated sequences, including HS5, ε-, γ-, Ψβ-, and δ-globin genes promoters, were determined by radioactive PCR, using primer sets and conditions as described (McMorrow et al. 2000 and this study). Input and unbound DNA samples were included as negative controls. Each reaction has been done in triplicate and band intensity has been quantitated using PhosphorImager analysis and normalized against mouse Zfp37 signal.

We are indebted to Patrick Molenbeek, Vincent Maes, and Agnes Boer for animal care; to John Kong-a-San for microinjection; to Nynke Gillemans, Bas Tolhuis, Miyata Masato, and Petros Papadopoulos for technical assistance; and to Drs. and Sjaak Philipsen for critically reading of this manuscript. This work was supported by a Marie Curie long-term fellowship (IHPMFCT199900175) to G.P.P. and by the European Commission and the Nederlandse Organisatie voor Weteuschappelijh Orderzoek (The Netherlands) grants to F.G.G.

The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC section 1734 solely to indicate this fact.