Substitution of the Human β-Spectrin Promoter for the Human ^Aγ-Globin Promoter Prevents Silencing of a Linked Human β-Globin Gene in Transgenic Mice

Abstract

During development, changes occur in both the sites of erythropoiesis and the globin genes expressed at each developmental stage. Previous work has shown that high-level expression of human β-like globin genes in transgenic mice requires the presence of the locus control region (LCR). Models of hemoglobin switching propose that the LCR and/or stage-specific elements interact with globin gene sequences to activate specific genes in erythroid cells. To test these models, we generated transgenic mice which contain the human ^Aγ-globin gene linked to a 576-bp fragment containing the human β-spectrin promoter. In these mice, the β-spectrin ^Aγ-globin (βsp/^Aγ) transgene was expressed at high levels in erythroid cells throughout development. Transgenic mice containing a 40-kb cosmid construct with the micro-LCR, βsp/^Aγ-, ψβ-, δ-, and β-globin genes showed no developmental switching and expressed both human γ- and β-globin mRNAs in erythroid cells throughout development. Mice containing control cosmids with the ^Aγ-globin gene promoter showed developmental switching and expressed ^Aγ-globin mRNA in yolk sac and fetal liver erythroid cells and β-globin mRNA in fetal liver and adult erythroid cells. Our results suggest that replacement of the γ-globin promoter with the β-spectrin promoter allows the expression of the β-globin gene. We conclude that the γ-globin promoter is necessary and sufficient to suppress the expression of the β-globin gene in yolk sac erythroid cells.

Hemoglobin switching describes the changes that occur in the sites of erythropoiesis during development as well as the changes in the globin genes that are expressed (51). The human β-like globin genes (5′ɛ^Gγ^Aγδβ3′) are located in an ∼75-kb region on the short arm of chromosome 11 and are arranged 5′ to 3′ in the order in which they are expressed during development (9, 58). The earliest human erythroid cells express ɛ-globin and are derived from the embryonic yolk sac. At 16 weeks of gestation, the fetal liver is the site of erythropoiesis. Fetal liver-derived erythroid cells express the duplicated γ-globin genes, while the ɛ-globin gene is silenced (45). By birth, the major site of erythropoiesis is the bone marrow, which expresses the δ- and β-globin genes only (for reviews, see references 9, 23, 31, 51, and 58).

The locus control region (LCR), a group of DNase I-hypersensitive sites located upstream of the globin cluster, is the most important cis-regulatory element for the β-like globin cluster (14, 19). Many studies have shown that the LCR is essential for high-level expression of globin genes in transgenic mice (15, 19, 47). In the absence of the LCR, individual human β-like globin genes are expressed in transgenic mice at low levels but in a developmentally appropriate manner (7, 26, 29, 52, 53).

The effects of the LCR on the developmental stage-specific expression of the human β-like globin genes are complex. In transgenic mice with the LCR and a single human γ- or β-globin gene, no developmental stage-specific expression has been observed (23, 31). The globin genes are expressed in yolk sac, fetal liver, and adult erythroid cells. However, when the LCR was linked to both a γ- and a β-globin gene, the developmental changes in gene expression were restored; the γ-globin genes were expressed in yolk sac- and fetal liver-derived erythroid cells, and the β-globin genes were expressed in fetal liver and adult erythroid cells (4, 12). Other studies made use of marked β-globin genes whose mRNAs could be distinguished from those of unmarked β-globin genes. Marked β-globin genes were inserted at different positions within a β-globin locus cosmid. These studies showed that both the position of the marked β-globin gene and the distance between the LCR and the genes in the cosmid construct determined the levels of both the marked and unmarked β-globin mRNAs (10, 39).

These and other studies of hemoglobin switching have suggested several models for the developmental regulation of the β-like globin genes which are not mutually exclusive. One model suggests that the gene order and distance of a gene from the LCR (polarity) determine the correct developmental expression pattern (20, 39). A second model proposes that the LCR serves only as an enhancer element, allowing high-level expression of the globin genes which are activated by developmental stage-specific elements (31). The competition model proposes that erythroid cells of each developmental stage contain stage-specific elements which compete for and stabilize interactions between globin genes and the LCR (8, 23).

All of these models suggest that elements near or within the individual globin genes are required for developmental activation and/or suppression of the β-like globin locus. Physical evidence for interactions between the LCR and individual globin genes has been shown by in situ hybridization analysis of erythroid cells from transgenic mice with a complete β-globin locus. Analysis of individual nuclei labeled with LCR, γ-globin, and β-globin probes have shown that the LCR interacts with only “one gene at a time” (56).

Other studies have shown that transgenic mice containing an LCR, a γ-globin gene with a promoter deletion, and the β-globin gene do not express the γ-globin gene at any stage of development, while the β-globin gene is expressed throughout development (3). Similarly, transient transfection studies of K562 cells have demonstrated that deletion of the proximal γ-globin promoter allows expression from a downstream β-globin promoter (24, 25). Both groups concluded that the γ-globin promoter sequences were responsible for the silencing of the downstream β-globin gene, but they could not exclude the possibility that γ-globin gene expression prevented β-globin gene expression in embryonic erythroid cells.

One prediction of the competition model for hemoglobin switching is that a heterologous promoter attached to a γ-globin gene would not compete for stage-specific elements or the LCR and would therefore permit the expression of a downstream β-globin gene in embryonic erythroid cells. To test this hypothesis, we fused a 576-bp fragment of the human β-spectrin promoter characterized by Gallagher et al. (18) to a human γ-globin gene and generated transgenic mice containing the β-spectrin/^Aγ-globin gene. We chose the promoter from the β-spectrin gene because β-spectrin is expressed at relatively high levels in erythroid tissues and does not exhibit changes in expression during development (44, 57).

Analysis of transgenic mice containing only β-spectrin/^Aγ-globin genes demonstrated that the human β-spectrin promoter directed high levels of γ-globin mRNA in the absence of the LCR at all stages of development. In transgenic mice in which the human β-spectrin promoter was substituted for the human ^Aγ-globin promoter in a cosmid construct containing a micro-LCR (μLCR) and ^Aγ-, ψβ-, δ-, and β-globin genes, both the β-spectrin/^Aγ-globin and β-globin genes are expressed throughout development. We conclude from these data that sequences within the γ-globin promoter, and not γ-globin transcription or intragenic sequences, are necessary and sufficient to silence the downstream β-globin gene.

(This work was performed by Denise E. Sabatino in partial fulfillment of the doctoral degree requirements in the Graduate Genetics Program of the Columbian School of Arts and Sciences at George Washington University.)

MATERIALS AND METHODS

Plasmid and cosmid constructs.

A 576-bp fragment containing the β-spectrin promoter (βsp) was excised from pGL2B (18) as a KpnI/HindIII fragment and cloned into the KpnI/HindIII sites of pSP72. The HindIII site in plasmid pSP72 βsp was destroyed by digestion with HindIII, filling in the ends, and religation. A 1,909-bp BsaHI/HindIII fragment containing the coding region of the human ^Aγ-globin gene was cloned into the ClaI/HindIII sites of pSP72. A 2,614-bp AatII/PvuII fragment from the βsp plasmid was ligated to a 2,266-bp EcoRV/AatII fragment from the ^Aγ-globin plasmid to create pSP72 βsp/^Aγ. The construct was confirmed by sequencing across the βsp/^Aγ exon 1 region. The 2,483-bp βsp/^Aγ gene was excised from this plasmid with EcoRV and HindIII for microinjection.

HS2 βsp/^Aγ was generated by a triple ligation using a 712-bp XhoI/HincII fragment containing hypersensitive site 2 (HS2) of the LCR (33, 50), a 2,483-bp EcoRV/HindIII fragment containing βsp/^Aγ, and the vector pBluescript II KS+ cut with XhoI and HindIII. The 3,182-bp HS2 βsp/^Aγ gene was excised from this plasmid with HindIII for microinjection.

The pBluescript II KS+ plasmid vector was modified by replacing the BstXI site in the polylinker with KpnI and Bsp120I sites to generate BS*. A 4,428-bp NotI/XhoI fragment of the μLCR^Aγψβδβ cosmid (12) containing the μLCR and a portion of the ^Aγ gene was cloned into the NotI/XhoI sites of BS* to generate μLCR ^Aγ 5′. This plasmid was digested with NotI and BamHI (polylinker) to generate the 2,972-bp fragment 1 of a four-part ligation. Fragment 2 was a 2,552-bp NotI/HindIII fragment containing the μLCR gene sequences from μLCR ^Aγ5′. Fragment 3 was a 964-bp HindIII/StuI fragment of μLCR ^Aγ 5′ containing the sequences from −1345 to −381 upstream of the ^Aγ promoter. The KpnI site in the 5′ polylinker of pSP72 βsp/^Aγ was destroyed by KpnI digestion, filling in the ends, and religation. Fragment 4 was a 1,042-bp EcoRV/BamHI fragment containing βsp/^Aγ gene sequences from this plasmid. The product of this ligation was designated μLCR βsp/^Aγ 5′. This plasmid was partially digested with Bsp120I and completely digested with HindIII to generate the 2,580-bp fragment 1 of a three-part ligation containing the μLCR. Plasmid μLCR βsp/^Aγ 5′ was digested partially with XhoI and completely with HindIII to generate the 2,059-bp fragment 2 containing βsp/^Aγ and upstream sequences. The third fragment was a 35.2-kb NotI/XhoI fragment of the μLCR^Aγψβδβ cosmid (12). The resulting ligation generated the μLCRβsp^Aγψβδβ cosmid. The 40-kb μLCRβsp^Aγψβδβ fragment was excised from this cosmid by digestion with KpnI for microinjection. The control μLCR^Aγψβδβ fragment was excised from the μLCR^Aγψβδβ cosmid by digestion with NotI and KpnI for microinjection as described previously (12).

Generation of transgenic mice.

Transgenic mice were generated as described by Hogan et al. (21). Fertilized eggs were collected from superovulating FVB/N female mice approximately 9 h after mating to CB6 F₁ male mice. Fragments for microinjection were separated on agarose gels and concentrated by using a DNA Geneclean isolation system according to the manufacturer’s instructions. The fragments were diluted to a concentration of 1 μg/ml in 7.5 mM Tris-HCl–0.25 mM EDTA (pH 7.5), and 1 to 3 pl was injected into the male pronuclei of fertilized eggs. The injected eggs were transferred to pseudopregnant CB6 F₁ foster mothers. Founder animals were identified by Southern blotting (30) of DNA extracted from tail biopsies. Southern blots were probed with human β-spectrin and/or human β-globin probes. Copy number was determined by comparing transgenic mouse DNA to K562 DNA on a Southern blot and analysis using a Molecular Dynamics PhosphorImager. Founder animals were crossed to FVB/N mice for propagation and developmental studies.

Cellulose acetate electrophoresis.

Blood samples were collected from 10.5-day embryos, 13.5-day embryos, and adult mice. Samples were lysed in cystamine as described by Whitney (55). Samples were run at 300 V for 20 min in a Titan electrophoresis chamber (Helena Laboratories), stained with Ponceau S (in 5% trichloroacetic acid) for 10 min, and destained in 7% acetic acid for 10 min (twice). The gels were fixed in methanol for 5 min (twice), cleared in a solution of 15 ml of glacial acetic acid, 35 ml of methanol, and 2 ml of Clear Aid for 7.5 min, and dried at 50 to 60°C for 15 min. The globin chain composing each hemoglobin tetramer was confirmed by acid-urea gel electrophoresis (1) of hemoglobins separated by cellulose acetate electrophoresis.

RNase protection assays.

Total cellular RNA was extracted from 10.5-day embryo blood cells, 13.5-day fetal livers, and adult reticulocytes, using TRIZOL reagent (Gibco BRL, Life Technologies, Inc., Grand Island, N.Y.). Linear DNA templates for the RNase protection assay were prepared by EcoRI (βsp/^Aγ), HindIII (mouse α), and HindIII (human β) digestion of cesium chloride-purified plasmid preparations. The templates were purified by agarose gel electrophoresis and purified by using a Geneclean II kit (Bio 101, Inc., Vista, Calif.). ³²P-labeled RNA probes were transcribed by using a MAXIscript in vitro transcription kit (Ambion, Inc., Austin, Tex.). Hybridization of the probe and the RNA (1 μg) was carried out overnight according to the standard procedure for the RPA II RNase protection assay kit (Ambion). RNase digestion was performed with an RNase A-RNase T₁ mixture in RNase digestion buffer (Ambion), and the protected fragments were separated on an 8% nondenaturing polyacrylamide gel (SEQUAGEL-8; National Diagnostics, Atlanta, Ga.).

Immunofluorescence analysis.

Fetal liver cells from transgenic mice and control littermates were stained with two monoclonal antibodies. One antibody, directed against human γ-globin, was directly conjugated to fluorescein isothiocyanate (FITC); the second monoclonal antibody, against human β-globin, was directly conjugated to rhodamine. The same field of cells was photographed under different exposure conditions: with an FITC filter (for detection of γ-globin), with a rhodamine filter (for detection of β-globin), and with first an FITC filter and then a rhodamine filter for a double exposure.

RESULTS

Transgenic mice with human γ-globin single-gene constructs.

We created two single-gene constructs containing the β-spectrin promoter fragment linked to the human ^Aγ-globin gene, βsp/^Aγ and HS2 βsp/^Aγ (6, 15, 27, 33, 42, 50, 54) (Fig. 1; see Materials and Methods). Three transgenic mouse lines and two 13.5-day fetal livers containing βsp/^Aγ were analyzed. RNase protection demonstrated that two of three βsp/^Aγ transgenic lines and one of two fetal livers expressed the βsp/^Aγ transgene in erythroid cells. No βsp/^Aγ mRNA was detected in erythroid cells of the third transgenic line and the other fetal liver. Analysis of RNA extracted from a variety of organs and tissues of adult βsp/^Aγ transgenic mice revealed high levels of βsp/^Aγ mRNA in reticulocytes, bone marrow, and spleen and lower levels in thymus and other tissues (Fig. 2) (18). Much of the globin mRNA in nonerythroid cells can be attributed to the high reticulocyte counts (∼15%) associated with a mild thalassemia caused by the excess of β-like globin chains in these and all other transgenic mouse lines described in this study (18).

FIG. 1.

β-Spectrin/γ-globin single-gene constructs used to generate transgenic mice. The β-spectrin promoter fragment (576 bp) is fused to the ^Aγ coding sequence at position −3 (A). A 712-bp fragment containing HS2 of the LCR is linked to the β-spectrin/γ-globin gene (B). Transgenic mice were identified by Southern blot analysis using a 576-bp β-spectrin promoter probe.

FIG. 2.

RNase protection analysis of βsp/^Aγ mRNA expression in transgenic mice. (A) RNase protection of 10 μg of K562 RNA and 1 μg of adult reticulocyte mRNA from strain βsp/^Aγ A. ³²P-labeled 1,493-bp RNA containing βsp/^Aγ gene sequences from exon 2, intron 1, and the βsp/^Aγ exon 1 fusion region was used as a probe. The positions of ³²P-labeled, protected RNA fragments corresponding to ^Aγ-globin exon 2, βsp/^Aγ exon 1, and ^Aγ-globin exon 1 are indicated. (B) RNase protection of 1 μg of bone marrow (BM) and spleen RNA and 10 μg of thymus RNA from strain βsp/^Aγ A. Labeled RNA fragments corresponding to the βsp/^Aγ (exon 2 shown), mouse α-globin (exon 2), and mouse actin (control) mRNAs were used as probes.

The number of transgenes in each line was estimated by Southern blot analysis to be between three and six copies per expressing animal (Table 1). The level of βsp/^Aγ mRNA was compared to the mRNA output of the four mouse α-globin genes. After correction for copy number, the mean βsp/^Aγ globin gene mRNA level in adult reticulocytes was approximately 32% of the level of the α-globin mRNA (Table 1).

TABLE 1.

Expression of βsp/^Aγ mRNA in transgenic mouse lines

Construct	Line	Mean human γ-globin RNA/mouse α-globin RNA ± SD			Copy no.	Mean γ-globin RNA/transgene copy no. in adults
		Yolk sac	Fetal liver	Adult
βsp/Aγ	A	0.37 ± 0.35	0.12 ± 0.02	0.34 ± 0.26	6	0.23
	B	No expression	No expression	No expression	16
	C	0.14 ± 0.03	0.12 ± 0.07	0.30 ± 0.22	3	0.40
	FL1		No expression
	FL2		0.48
Mean ± SD		0.26 ± 0.26	0.17 ± 0.14	0.32 ± 0.21		0.32 ± 0.12
HS2 βsp/Aγ	A	0.07 ± 0.04	0.04 ± 0.01	0.19 ± 0.09	3	0.25
	B	0.20 ± 0.25	0.28 ± 0.31	0.34 ± 0.27	3	0.45
	C	0.05 ± 0.04	0.24 ± 0.26	0.05 ± 0.01	3	0.06
	D	ND	ND	0.22	0.9	0.97
	E	ND	ND	0.13	0.5	1.04
	F	0.10	0.04	0.05	3	0.06
	FL1		0.26
Mean ± SD		0.11 ± 0.14	0.18 ± 0.21	0.18 ± 0.19		0.47 ± 0.44

Six transgenic lines and one 13.5-day fetal liver containing HS2 βsp/^Aγ were analyzed. Expression of βsp/^Aγ was detected in erythroid cells of all six lines and the fetal liver. Southern blot analysis revealed a copy number of 0.5 to 3 copies per animal. Two animals with less than one copy per cell did not transmit transgenes to F₁ progeny and were determined to be mosaic for the transgene. After correction for copy number, the mean level of βsp/^Aγ mRNA in adult reticulocytes was 47% of the level of α-globin mRNA (Table 1). This difference was not statistically different from the levels observed in βsp/^Aγ transgenic mice.

RNA was extracted from 10.5-day-postcoitum (dpc) yolk sac derived peripheral blood cells, 13.5-dpc fetal livers, and adult reticulocytes for RNase protection analysis of βsp/^Aγ mRNA levels during development. In all transgenic lines expressing either βsp/^Aγ or HS2 βsp/^Aγ, the γ-globin gene was expressed at all developmental stages at levels ranging from 4 to 37% of the levels of mouse α-globin expression in the same cells (Fig. 3; Table 1).

FIG. 3.

RNase protection of 1 μg of 10.5-day embryonic blood, 13.5-day fetal liver, and adult reticulocyte RNAs from strains βsp/^Aγ A and C (A) and from strains HS2 βsp/^Aγ B and C (B). Labeled RNA fragments corresponding to the βsp/^Aγ (exon 2 shown) and mouse α-globin (exon 2) genes were used as probes.

Cellulose acetate electrophoresis was performed to confirm the presence of both γ-globin chains in 13.5-day fetal liver and adult erythrocytes. The fetal and adult erythrocytes of βsp/^Aγ and HS2 βsp/^Aγ transgenic mice contained endogenous mouse hemoglobins as well as an additional hemoglobin band composed of two mouse α-globin chains and two human γ-globin chains (Fig. 4). These results were confirmed by acid-urea gel electrophoresis and high-pressure liquid chromatography (HPLC) analysis (data not shown).

FIG. 4.

Cellulose acetate electrophoresis of 14.5-day (A) and adult (B) peripheral blood lysates from strain βsp/^Aγ A. Positions of the endogenous mouse hemoglobins and the mouse α-2/human γ-2 hemoglobin tetramer are shown. +, blood lysate from transgenic animal; −, blood lysate from littermate control.

Transgenic mice with cosmid gene constructs.

A 37-kb μLCR^Aγψβδβ cosmid construct (Fig. 5) containing a 4-kb μLCR linked to the ^Aγ-, ψβ-, δ-, and β-globin genes was modified by replacing 381 bp of the ^Aγ-globin promoter with the 576-bp β-spectrin promoter (Fig. 5). Three transgenic founder animals were generated with the resulting μLCRβsp^Aγψβδβ cosmid. High levels of βsp/^Aγ mRNA were detected in RNA from reticulocytes, bone marrow, and spleen, and no βsp/^Aγ mRNA was detected in RNA from thymus (Fig. 6). Southern blot analysis showed that these lines contained between 8 and 40 copies of the cosmid. One transgenic founder animal (strain C) did not transmit the transgene to F₁ progeny and was concluded to be mosaic for the transgene. Transgene expression in this line was studied only in the founder animal. Three additional transgenic lines containing the control μLCR^Aγψβδβ cosmid construct were generated for comparison.

FIG. 5.

Control and βsp/^Aγ cosmid constructs used to generate transgenic mice. The control μLCR^Aγψβδβ cosmid originally described by Enver et al. (12) (A) was modified to replace the ^Aγ promoter region from −381 to −1 with a 576-bp fragment of the human β-spectrin promoter (B). Transgenic mice were identified by Southern blot analysis using a 576-bp fragment containing the human β-spectrin promoter region and a 960-bp fragment containing exon 2 of the human β-globin gene as probes.

FIG. 6.

Analysis of βsp/^Aγ mRNA expression in transgenic mice containing the μLCR^Aγψβδβ cosmid by RNase protection assay of 1 μg of bone marrow (BM) and spleen RNA and 10 μg of thymus RNA from strain μLCRβsp^Aγψβδβ A. Labeled RNA fragments corresponding to the βsp/^Aγ (exon 2 shown), mouse α-globin (exon 2), and mouse actin (control) genes were used as probes.

RNA was extracted from 10.5-dpc blood, 13.5-dpc fetal liver, and adult reticulocytes for RNase protection analysis of human γ- and β-globin mRNA levels during development. As described previously (12), we detected γ-globin mRNA in yolk sac and fetal liver and β-globin mRNA in fetal liver and adult erythroid cells of transgenic lines with the control μLCR^Aγψβδβ cosmid (Fig. 7; Table 2). In contrast, both βsp/^Aγ mRNA and human β-globin mRNA were detected at high levels in yolk sac, fetal liver, and adult erythroid cells of transgenic mice containing the μLCRβsp^Aγψβδβ cosmid (Fig. 7; Table 2).

FIG. 7.

RNase protection of 1 μg of 10.5-day embryonic blood, 13.5-day fetal liver, and adult reticulocyte RNA from control cosmid strain C (A) and from strain μLCRβsp^Aγψβδβ A (B). Labeled RNA fragments corresponding to the βsp/^Aγ (exon 2 shown), human β-globin (exon 3), and mouse α-globin (exon 2) genes were used as probes.

TABLE 2.

Expression of βsp/^Aγ mRNA in μLCRβsp^Aγψβδβ and μLCR^Aγψβδβ transgenic mouse lines

Construct	Line	Mean human γ-globin RNA/ mouse α-globin RNA ± SD			Mean human β-globin RNA/ mouse α-globin RNA ± SD			Copy no.	Mean γ-globin RNA/transgene copy no. in adults	Mean β-globin RNA/transgene copy no. in adults
		Yolk sac	Fetal liver	Adult	Yolk sac	Fetal liver	Adult
μLCRβspAγψβδβ	A	0.47 ± 0.37	0.42 ± 0.11	0.46 ± 0.01	1.04 ± 0.83	2.70 ± 2.79	0.66 ± 0.33	15	0.12	0.18
	B	0.23	0.54 ± 0.17	0.55 ± 0.15	0.96	0.40 ± 0.22	1.34 ± 0.58	40	0.06	0.13
	C	ND	ND	0.31	ND	ND	0.71	8	0.16	0.36
Mean ± SD		0.39 ± 0.29	0.48 ± 0.15	0.48 ± 0.14	1.01 ± 0.59	1.32 ± 1.88	1.01 ± 0.54		0.11 ± 0.05	0.22 ± 0.12
μLCRAγψβδβ	A	ND	0.16	0.06	0.00	0.52	0.64	34	0.01	0.08
	B	1.06 ± 1.10	0.05 ± 0.07	0.03 ± 0.01	0.02 ± 0.01	0.18 ± 0.25	0.53 ± 0.04	20	0.01	0.11
	C	1.54 ± 1.73	0.05 ± 0.01	0.04 ± 0.00	0.04 ± 0.04	0.20 ± 0.06	0.49 ± 0.41	4	0.04	0.49
Mean ± SD		1.30 ± 1.22	0.07 ± 0.06	0.04 ± 0.01	0.02 ± 0.03	0.26 ± 0.20	0.53 ± 0.021		0.02 ± 0.02	0.23 ± 0.23

Cellulose acetate electrophoresis was performed to confirm the presence of both γ- and β-globin chains in 13.5-day fetal liver and adult erythrocytes. Fetal erythrocytes from transgenic mice with either the control μLCR^Aγψβδβ cosmid or the μLCRβsp^Aγψβδβ cosmid contained endogenous mouse hemoglobins as well as two additional hemoglobin bands consisting of two mouse α-globin and two human γ-globin chains and two mouse α-globin and two human β-globin chains (Fig. 8). Adult peripheral blood of control μLCR^Aγψβδβ cosmid transgenic mice contained only the endogenous mouse hemoglobins and the mouse α-2/human β-2 band. In contrast, adult erythrocytes of μLCRβsp^Aγψβδβ transgenic mice contained the endogenous mouse hemoglobins and both the mouse α-2/human γ-2 and mouse α-2/human β-2 bands (Fig. 8). These results were confirmed by HPLC analysis (data not shown).

FIG. 8.

Cellulose acetate electrophoresis of 14.5-day (14.5d) and adult peripheral blood lysates from control μLCR^Aγψβδβ C (left) and μLCRβsp^Aγψβδβ A (right) transgenic mice. The positions of the endogenous mouse hemoglobins, the mouse α-2/human γ-2 hemoglobin tetramer, and the mouse α-2/human β-2 hemoglobin tetramer are shown. +, blood lysate from transgenic animal; −, blood lysate from littermate control.

Immunofluorescence analysis of 14.5-dpc fetal livers from μLCRβsp^Aγψβδβ embryos.

Immunofluorescence analysis of human γ- and β-globin gene expression was done on 14.5-dpc fetal livers from μLCRβsp^Aγψβδβ transgenic embryos and control μLCR^Aγψβδβ cosmid-containing embryos. As previously described, fetal liver cells from control μLCR^Aγψβδβ cosmid-containing mice expressed either human β-globin or human γ-globin (12). Fetal liver cells from μLCRβsp^Aγψβδβ cosmid-containing mice all expressed both human γ-globin and human β-globin (Fig. 9).

FIG. 9.

Immunofluorescence analysis of βsp/^Aγ and human β-globin chains in fetal liver cells from μLCRβsp^Aγψβδβ transgenic mice. The fetal liver cells were fixed and stained with conjugated monoclonal antibodies against human γ-globin (FITC) and human β-globin (rhodamine). The identical field was photographed with an FITC filter (γ), a rhodamine filter (β), and both filters (γ+β).

DISCUSSION

Analysis of transgenic mice containing either cosmids or yeast artificial chromosomes (YACs) has shown that appropriate developmental regulation requires the presence of an active globin gene or genes upstream of the β-globin gene (4, 11, 12, 16, 17, 28, 37, 40, 43). However, these studies have not identified the specific sequences involved. Analysis of deletion constructs in transgenic mice or cultured cells demonstrated that deletion of the γ-globin promoter allowed expression of the downstream β-globin gene (3, 24). However, the same deletions which allow β-globin gene expression abolished γ-globin gene expression. Our data support the competition model for hemoglobin switching, which posits that the individual globin genes compete for stage-specific elements to stabilize transcription of individual globin genes. By replacing the γ-globin promoter with the β-spectrin promoter, the β-globin gene is expressed at all stages of development in the presence of γ-globin gene expression. We conclude that the sequences between −381 and +1 of the γ-globin promoter, and not γ-globin transcription or intragenic sequences, are necessary and sufficient to silence the downstream β-globin gene in the embryonic stage of development.

The observation that the β-spectrin promoter is an enhancer-dependent and erythroid cell-specific promoter makes it an ideal promoter for the studies described here. Previous analyses of transgenic mice carrying phosphoglycerate kinase-Neo gene constructs inserted into either the mouse β-globin locus (49) or the mouse (13, 22) and human (5, 38) LCRs had strikingly negative effects on globin gene expression. It is clear from these studies that the phosphoglycerate kinase gene promoter or Neo sequences are not neutral substitutions and that they may be strong competitors for stage-specific elements and/or the LCR (41).

Our data also support models in which specific sequences in the γ-globin promoter capture the LCR in the yolk sac erythroid cells and prevent β-globin gene expression (2, 8, 24, 25). The testing of this model, and the importance of gene order and polarity, would require substitution of the β-spectrin promoter for the ^Aγ-globin promoter in a YAC construct, which would place two globin promoters upstream of the β-spectrin promoter. If the LCR specifically interacts with these promoters, we hypothesize that expression of the β-globin gene will be suppressed in yolk sac erythroid cells.

Our data are not consistent with some aspects of the models in which stage-specific elements alone are responsible for hemoglobin switching (31). These models posit that the presence of an enhancer-dependent γ-globin gene might be sufficient to suppress β-globin expression in the yolk sac, in contrast to what we observed. The possibility that the β-spectrin promoter does not compete with the β-globin promoter for the same stage-specific factors which interact with the γ-globin promoter can be tested in the YAC experiments as well.

Since the βsp/^Aγ globin gene is expressed at relatively high levels in erythroid cells at all stages of development, βsp/^Aγ transgenic mice may be useful to test the effects of γ-globin on the amelioration of sickling in the sickle cell disease mouse model (35, 48). Previous in vitro studies have demonstrated that the amount of hemoglobin S (HbS) polymers decreases with increasing proportions of HbF, reaching a maximum inhibition of polymerization at 20% HbF (34). Other studies have found that as the HbF concentration increases, the severity of sickle cell disease decreases, with significantly less severe disease with >10% HbF (32, 36, 46). Generating sickle cell disease mice carrying the βsp/^Aγ gene would test whether the ^Aγ-globin expressed from the β-spectrin promoter is sufficient to reduce HbS polymerization in the mouse model.