YAC DNA Isolation/Transgenic Mice.

YAC DNA was isolated from pulse-field gels according to standard protocols (47, 48) with minor modifications (26). The DNA was injected into fertilized mouse eggs (CD1; Charles River Breeding Laboratory) before transfer to foster mothers (B6D2F1/J) as described (49). Tail DNAs from offspring were analyzed for the presence of the left and right YAC vector arms by PCR, and then for integrity of the transgenic β-globin locus by Southern blot hybridization of SfiI/SalI digests of DNA from transgenic organs (see below; ref. 22).

Transgene Copy Number and YAC Integrity.

To assess the integrated copy number of human β-globin YACs, genomic DNA recovered from transgenics was digested with PstI, which has two sites in the pYAC4-derived left vector arm. A 1.7-kbp EarI fragment spanning the 3′ PstI site was used for Southern blot hybridization, where a 1.6-kbp common band, as well as others that vary in size, would be detected, depending on the transgene copy number and YAC integration pattern(s).

To assess the integrity of the integrated YACs, hematopoietic cells from 0.5 g of transgenic spleen were suspended in 1 ml of cell suspension buffer (10 mM Tris·HCl, pH 7.5/20 mM NaCl/50 mM EDTA, pH 8.0). This solution was then diluted 1:10 in 0.4% agarose solution and poured into clamped homogeneous electric field (CHEF) plug molds. Solidified plugs were then treated at 50°C in Proteinase K buffer (100 mM EDTA, pH 8.0/0.2% sodium deoxycholate/1% Sarcosyl/10 mg/ml Proteinase K) for 48 h, followed by several 1-h washes with 10 mM Tris·HCl (pH 7.4) and 1 mM EDTA buffer. The plugs were digested with SfiI at 50°C for 12 h, and then with SalI for 12 h before pulse field gel electrophoresis and Southern blot hybridization. Two radiolabeled probes were employed (corresponding to HS4 or the Aγ-globin gene) that would be expected to hybridize to 10-kbp (SfiI/SfiI) or 52-kbp (SfiI/SalI) bands, respectively, in the event that the integrated human β-globin YAC transgenes were intact.

Transgenic Mice/Transgene Structure.

The mutated YAC DNAs were isolated from pulse-field gels and purified as described (see Materials and Methods and ref. 26). These DNAs were then microinjected into pronuclei of fertilized mouse ova. Founder tail DNAs were initially screened by PCR, using primers specific for both right and left YAC vector arms, and then verified by Southern blot hybridization. Southern blotting of each of the F₁ or F₂ lines, using probes that span the locus (22) showed that each contained restriction fragments of the expected size (e.g., Fig. ​Fig.11C), and that no other bands (indicative of breakpoints in the YACs) were present. Fragmented YAC transgenes detected through this analysis were excluded from further consideration.

The YAC transgene copy number in the lines chosen for final expression analysis was determined using a YAC end assay, employing a probe that would detect a common 1.6-kbp band as well as other characteristic fragments in each digest. For example, a head-to-head tandem integration was predicted from the map of pYAC4 to yield an additional fragment 4.52 kbp in size (Fig. ​(Fig.22A, lanes 3 and 6), whereas a head-to-tail tandem integration would result in the generation of a fragment of 5.36 kbp (Fig. ​(Fig.22A, lane 5); if the YAC end was embedded in mouse chromosomal DNA (as either a single copy transgene or as the flanking copy of a multi-YAC integration), the size of the fragment would be dependent on the next random PstI site encountered in the mouse genome. From these data, combined with the data obtained from standard Southern blots and comparison of internal fragment intensities, we estimate that the three -silencer deletion lines contain three, two, and two transgene copies (Fig. ​(Fig.22A, lanes 1–3), respectively, whereas the two β-globin enhancer deletion lines both harbor four tandemly integrated YAC copies (Fig. ​(Fig.22A, lanes 5 and 6). Furthermore, the analysis shows that, as previously stated, the wild-type YAC transgenic line HS4321b (26) is single copy (Fig. ​(Fig.22A, lane 4).

Open in a separate window

Figure 2

Copy number and integrity of YAC transgenes. Lanes: 1–3, -globin silencer deletion mutant transgenic lines 1–3, respectively; 4, wild type β-globin YAC transgenic line HS4321b (26); 5 and 6, β-globin enhancer deletion mutant lines 1 and 2, respectively (numbering at the top of each panel). (A) Tail DNA from F₁ or F₂ pups representing each of the transgenic lines was digested with PstI and electrophoresed and blotted normally. A 1.7-kbp EarI fragment isolated from plasmid vector pYAC4 was radiolabeled and then hybridized to the Southern blot, and finally exposed for autoradiography. Each lane displays a (common) internal fragment band from the pYAC4 vector (1.58 kbp) as well as other bands, depending on the transgene copy number and integration site. Head-to-head tandem integrations would be expected to yield a 4.5-kbp band, whereas head-to-tail tandem integrations are predicted to yield a 5.3-kbp band. (B) Spleen cells were isolated from each transgenic line and digested with SfiI plus SalI as described prior to electrophoresis on pulse field gels. The gel was then blotted to nylon and hybridized to probes corresponding to either LCR HS4 or the Gγ-globin gene.

The integrity of the YAC transgenes was assayed by examining spleen cell DNA that had been separated on pulse field gels for the presence of 10- and 52-kbp diagnostic bands on Southern blots after digestion with SfiI and SalI (Fig. ​(Fig.22B). The two β-globin locus fragments detected in this assay extend from beyond LCR HS5 at the 5′ end of the locus to a position near the start of the δ-globin gene (22). The data are shown in Fig. ​Fig.22B, where the two expected diagnostic bands are clearly the most prominent in each of the transgenic lines; in particular, it should be noted that there are no bands detected below 60 kbp other than the two that are expected, thus adding credence to the contention that the YACs examined here are unbroken within the 62-kbp interval surveyed in this assay. It should also be noted that other bands of higher molecular weight are clearly detected in the same blot, but other experiments show that these are primarily due to partial digestion of the spleen cell plugs by SalI, a vexing problem we have not yet fully surmounted. We assume that this partial digestion is due to methylation of some SalI CpG sites.

Sequences Within the -Globin Gene Silencer Are Important for Embryonic Stage-Specific Activation of Both the - and γ-Globin Genes.

In transgenic mice, the human -globin gene is expressed in the embryonic yolk sac beginning between day 8.5 and 9.5 days postcoitus of gestation and switches off by day 12.5 (23–25); this repression is postulated to result from the binding of negative regulatory proteins to the -globin gene silencer. Multiple regulatory elements have been described in the -globin gene upstream region (41), and the most prominent of these is a silencer located from −304 to −179 relative to the -globin gene transcription initiation site (27). This silencer element contains consensus binding sites both for GATA factors and for YY1. If the silencer harbors only the expected negative regulatory activity, one would predict that deletion of this element from an otherwise intact human β-globin gene locus would result in the continued expression of the -globin gene into the fetal stage of erythropoiesis. We therefore directly tested the hypothesis that the silencer acts to restrict expression of the -globin gene to the embryonic yolk sac stage.

To our surprise, deletion of the silencer did not lead to high level -globin gene transcription in the fetal liver period, as anticipated, but rather caused a significant decline in -globin gene transcription in the yolk sac, thus demonstrating that this narrowly defined silencer sequence element of 125 bp still harbors cryptic activity that is required for stimulation of -globin RNA synthesis (Fig. ​(Fig.3).3). An even more unexpected observation was that deletion of the -globin gene silencer not only diminished -globin gene transcription, but also resulted in a significant decline in γ-globin RNA synthesis in the yolk sac (Fig. ​(Fig.3). 3). Transcription of the γ- and β-globin genes during both the fetal liver and adult stages was not as significantly affected in these animals, thereby implying that these -globin 5′ sequences harbor a more general function in regulatory control over globin genes that are expressed in the embryonic yolk sac.

Open in a separate window

Figure 3

Expression of human β-globin YAC mutants during erythropoiesis. (A Upper). Typical reverse transcriptase–PCR expression data for the human -, γ-, and β-globin genes as compared with the murine α-globin used as the internal control for each RNA sample. The transgenic line examined is abbreviated at the top of each lane, while the developmental stage and sites are shown below each panel. A +1 refers to RNA isolated from the single copy wild-type YAC transgenic line, HS4321b (26). (B) Quantification of the expression data shown in A after PhosphorImager analysis and correction for gene copy number (Fig. ​(Fig.2). 2). Open rectangles represent -globin expression, shaded boxes represent γ-globin expression, and filled boxes represent β-globin expression levels.

We should note that the variation in expression levels is sporadically greater in all of the transgenic lines we have examined in this study than we had observed previously (26), where all the independent lines appeared to express the genes during different developmental stages with no greater than 10% variation. It may be that expression in our previous transgenic lines was fortuitously homogeneous for unknown reasons, or that the expression from single copy transgenes perturbs the erythroid cell transcriptional environment less than the presence of the two to four copies found in the animals under present analysis. Additionally, deletion of important regulatory DNA elements within the locus could (to some extent) affect the position-independent expression of the YAC (G. Li, J.B., and J.D.E., unpublished observations). For these reasons, we focused our interpretation and discussion on those results that display the most dramatic deviation(s) from the expression characteristics of the wild-type YAC transgenic lines and, more importantly, on those results that appeared to be consistent among the mutant lines.

It has been shown previously that binding sites for the ubiquitous transcription factor YY1 and for GATA factors contribute to the repressing activity of this element. However, Raich et al. (44) also demonstrated that two GATA sites immediately downstream of the YY1 site have a pronounced positive effect on -globin gene transcription. These results clearly demonstrate the presence of additional, functionally important positive regulatory elements even within this narrowly defined silencer region.

Because there are other positive regulatory elements both upstream and downstream of the silencer, it could be argued that the mutant is defective in activation because the missing sequence elements must cooperate with these flanking positive elements to confer efficient globin gene transcription (i.e., that the deleted sequences per se have no positive regulatory effect). Alternatively, changing the spacing between the more upstream and the more gene-proximal positive regulatory elements (after deleting the 125-bp silencer) might impair the formation of an active transcription complex on the -globin gene promoter. Finally, it is possible that the 3′ part of the deleted silencer is additionally the extreme 5′ end of an extended -globin gene promoter, thus affecting assembly of the basal apparatus (but clearly this cannot be the only effect; next paragraph). Further studies introducing specific mutations that do not alter the spacing within this region should distinguish between these several possibilities.

The more remarkable observation from analysis of the -silencer deletion mutant was that it also significantly reduced transcription of the γ-globin genes in erythroid cells of the embryonic yolk sac (Fig. ​(Fig.3).3). This effect was erythroid stage-specific, not significantly affecting expression of either the γ- or β-globin genes during either the fetal liver or adult stages. We interpret this observation to mean that deletion of sequences within the upstream region of the -globin gene generally affects embryonic stage-specific globin gene expression. This is the second such mutation in the locus that appears to quite specifically affect the activity of these same genes during the embryonic stage of murine erythropoiesis. We previously reported that duplication of HS4 of the LCR resulted in a decline in -, and a simultaneous, reciprocal increase in γ-, globin mRNA accumulation in the yolk sac (26). Taken together with the data presented here, both mutations point to an effect which may best be explained by assuming that the LCR as well as the - and γ-globin genes are in a different chromosomal domain in the yolk sac than in the fetal liver or adult spleen. Under the assumption that an LCR holocomplex must physically interact with other gene proximal protein/DNA complexes, these observations could be further interpreted to infer that deletion of the -globin silencer not only reduces the ability of the LCR to productively interact with the -globin gene promoter, but that these sequences may be involved in establishing this specific chromatin conformation that is required for proper γ-globin gene transcription in the murine yolk sac.

Previously, Wijgerde et al. (7) showed that two globin genes can be transcribed inside a single erythroid cell within only a brief time span. In terms of the competition model (10), these data were also interpreted to suggest that an LCR holocomplex (26) somehow distinguishes which gene to activate, presumably depending on the array and concentration of transcription factors present in an erythroid cell at different developmental stages. It has been shown experimentally that the human γ-globin genes are more abundantly expressed than the -globin gene in the murine embryonic yolk sac, thus implying that in a competitive regulatory environment, the human LCR interacts with the more distant human γ-globin gene promoters either more frequently or more stably than with the -globin gene promoter at this stage of murine erythropoiesis. However, the -globin upstream region might be important for recruiting or positioning the LCR, not only to enhance -globin gene transcription but also to allow efficient interaction with the γ-globin gene promoters by decreasing the distance between LCR and γ-globin genes. Deletion of such a hypothetical LCR “positioning” element would then result in a decline in expression of both the - and γ-globin genes.

We thank Jie Fan for outstanding technical assistance. We gratefully acknowledge fellowship support from the Cooley’s Anemia Foundation (Q.L.), the Deutsche Forschungsgemeinschaft (Grant A2:Bu 895/1-1), and the American Heart Association (J. B.), and research support from the National Institutes of Health (HL 24415; J.D.E.).