The Roles of 5′-HS2, 5′-HS3, and the γ-Globin TATA, CACCC, and Stage Selector Elements in Suppression of β-Globin Expression in Early Development*

The roles of HS2 and HS3 from the human β-globin locus control region and of the TATA, CACCC, and stage selector elements of the γ-globin promoter, in competitive inhibition of β-globin gene expression in early development, were tested using stable transfections of HEL and K562 cells. Cells with an HS3γβ construct demonstrate that HS3 exhibits enhancing activity, but compared with HS2, this site participates less consistently in the inhibition of embryonic/fetal β-globin expression. In cells with HS3HS2γβ constructs, the two HS sites act in concert to more effectively enhance γ-globin gene expression and to drive stage-specific expression of the γ- and β-globin genes. A γ-globin gene with a –161 promoter can competitively inhibit β-globin gene expression. HS3HS2γβ constructs were used to determine the effects of γ-globin promoter mutations within this region on competition. The CACCC and TATA elements, but not the stage selector element, inhibit inappropriate embryonic/fetal stage expression of the β-globin gene. The mutation in the γ-globin TATA element results in the use of two major alternative transcription start sites. The data suggest that proteins binding to the γ-globin CACCC and TATA elements interact with those binding to HS2 and/or HS3 to preclude β-globin transcription in early development.

Human ␥to ␤-globin gene switching is coordinated by several regulatory sequences including the gene promoters and the locus control region (LCR). 1 The LCR contains 5 DNaseI super-hypersensitive sites (HS1-HS5) located 6 to 26 kb upstream of the ⑀-globin gene in the ␤-globin gene cluster (1,2). The ⑀-(embryonic), ␥-(fetal), and ␤-(adult) globin genes are positioned in the order of their temporal expression. The LCR has enhancer activity (3,4) and can initiate and maintain erythroid-specific open chromatin structure to allow for the activation of gene expression (1,5). The HS sites of the LCR contain binding sites for transcription factors that are critical for its activity (reviewed in Ref. 6). In situ hybridization of primary globin transcripts in transgenic mouse cells has provided in vivo evidence that the LCR is likely to act as a holocomplex to activate transcription by stably interacting with one ␤-like globin gene at a time (7). A current model states that the fetal ␥-globin gene and its promoter play important roles in interactions with the LCR that result in the competitive inhibition of adult ␤-globin gene expression in early development (8,9). This is supported by the observation that the ␤-globin gene is expressed inappropriately early in mice and tissue culture cells with LCR constructs that do not contain the ␥-globin gene (10 -13). The factors that bind to the gene promoters and to the LCR that are important for such interactions have not yet been identified, although some of the DNA elements involved are elucidated in this work.
The individual roles of the HS sites are complex and are not yet completely understood. Studies with transgenic mice that contain individual or combinations of HS sites, deletions of specific HS sites, or deletions of the highly conserved core elements of the HS sites as well as experiments with knock-out mice have provided evidence of the roles of each HS site. HS2 and HS3 have been shown to exhibit the majority of the enhancing activity of the LCR (14) and may participate in additive interactions (15) or synergistic interactions for long range enhancer activity (16).
Mice with constructs containing HS2 alone maintain correct stage-specific expression of ␥and ␤-globin mRNA throughout development (10,17) and express equivalent levels of ␥-globin mRNA in the embryo and ␤-globin mRNA in the fetus and adult (18). In transgenic mouse studies in which HS2 is deleted from the intact human ␤-globin locus, there is only a small decrease in expression of the ⑀-, ␥-, and ␤-globin genes, suggesting that at least some of the properties of HS2 may be functionally redundant (19). However, others have reported dramatic decreases in gene expression and loss of copy-number dependence when a similar transgene is integrated into heterochromatic regions in the mouse genome, suggesting a role for HS2 in overcoming position effects (20). Deletion of murine HS2 in knock-out mice resulted in normal expression of the mouse embryonic genes but a 30% reduction of adult ␤-globin mRNA (21,22). The greater effect on the adult gene may be because of the longer distance between the ␤-globin gene and the LCR or gene order. It is important to consider that although mice are a convenient model for the investigation of human gene expression, with respect to HS2, the mouse ␤-globin locus and its regulation may not be identical to that of the human locus.
Similar studies have concentrated on the role of HS3, because HS3 is known to demonstrate enhancing and chromatinopening activities similar to those of HS2 (14). In transgenic mice with constructs containing individual HS sites, it has been shown that HS3 is the most active site in enhancing ␥-globin expression during the embryonic period and also supports ␥-globin expression during fetal hematopoiesis (18). In transgenic mice containing an otherwise intact human ␤-globin locus, an HS3 deletion results in a significant decrease of ⑀-globin, an increase of ␥-globin, and no change in ␤-globin gene expression (19). The importance of HS3 in the human ␤-globin locus in overcoming position effects has also been observed in deletion studies (20). Deletion of murine HS3 produces alterations in the expression of the ␤-like globin genes similar to those caused by deletions of HS2 (21,23). Transgenic mice with deletions of a 225-bp or a 234-bp human HS3 core have low or no detectable globin expression at all stages of development (24) or severely reduced ⑀-globin and no ␥-globin mRNA (25). Although these results appear inconsistent with the studies that delete the entire HS site, they substantiate the importance of the flanking sequences of the HS sites in the activity of the LCR. If the sequences outside of the HS core interact with the gene promoters but lack the enhancing activity of the core element, then a dominant-negative effect may be observed (6). This illustrates the value of studying each HS site in the context of its core and flanking sequences to define its role.
In previous work using transgenic mice and transfection analyses, we have shown that the -136 to ϩ56 region of the ␥-globin promoter is necessary for competitive inhibition of ␤-globin gene expression in embryonic/fetal cells. A ␥-globin gene with a Ϫ161 promoter can compete successfully with the ␤-globin gene; therefore, an element(s) downstream of -161 has a role in embryonic ␤-globin suppression (13,26). This region of the ␥-globin promoter contains a CACCC element, two CCAAT boxes, the stage selector element (SSE) and a TATA box (see Fig. 1). Targeted gene knock-outs in mice indicate that erythroid Kruppel-like factor (EKLF), which binds preferentially to the CACCC element in the ␤-globin promoter, is responsible for positively regulating this gene in the adult and has little direct effect on ␥-globin gene expression (27)(28)(29). It is plausible that an as yet unidentified ␥-globin gene-specific CACCC binding factor exists and is important for developmental switching. Simultaneous mutations of the two CCAAT boxes in the ␥-globin promoter have a negligible effect on ␥and ␤-globin gene expression in transgenic mice, so it does not appear that these elements play an independent role in ␤-globin suppression (30). Stage selector protein (SSP) binds to the 19-bp SSE near base position Ϫ50 in the ␥-globin promoter. The SSE is required for competitive inhibition of ␤-globin expression by a Ϫ260 ␥-globin promoter in transient transfection assays in K562 cells, although a deletional mutation of the SSE element had little direct effect on ␥-globin gene transcription in the same system (31,32).
The current study aims to determine the roles of HS2 and HS3 from the LCR and of the CACCC, SSE, and TATA elements in the ␥-globin promoter in the suppression of ␤-globin gene expression early in development using stable transfections in human HEL and K562 cells. We find that HS3, unlike HS2, does not consistently participate in competitive inhibition of ␤-globin expression by the ␥-globin gene. The HS3HS2 configuration has the combined enhancing activity of HS2 and HS3 and also the developmental specificity of HS2, suggesting that HS2 and HS3 act most effectively in concert. The CACCC and TATA, but not the SSE, elements in the ␥-globin promoter are involved in the inhibition of early ␤-globin gene expression. This work begins to identify the elements in the LCR and in the ␥-globin promoter that bind factors to establish competitive inhibition.

Preparation of the Constructs for the Stable Transfection Assays-
The constructs used in the stable transfection assays are shown in Fig.  2 and are HS3␥␤, HS3␤, HS3HS2␥␤, HS3HS2␤, HS3HS2␥TATA␤, HS3HS2␥CACCC␤, and HS3HS2␥SSE␤. Those with HS2 contain a 1.9-kb KpnI-PvuII fragment (GenBank™ HUMHBB coordinates 7764 -9653) normally located about 11 kb 5Ј of the human ⑀-globin gene. The constructs containing HS3 have a 1.9-kb HindIII fragment (HUMHBB 3266 -5172). The ␤-globin gene in these constructs is a 4.5-kb ApaI-EcoRV fragment (Ϫ1250 to ϩ3291) containing the 3Ј enhancer region. The LCR␥␤ constructs contain a HindIII fragment (Ϫ1350 to ϩ1951) encompassing a derivative of the human A ␥-globin gene, situated between the LCR sequences and the ␤-globin gene. The ␥and ␤-globin genes in all of the constructs are marked by a 4-bp insertion, made by filling in an NcoI restriction site at position ϩ50 using Klenow fragment to distinguish their mRNAs from the endogenous globin mRNAs.
The mutations in the ␥-globin promoter TATA, CACCC, and SSE elements were generated using the CLONTECH Transformer Site-Directed Mutagenesis Kit. An oligonucleotide was used to mutate the TATA, CACCC, or SSE elements and create a new restriction enzyme site. In addition to the mutagenic primer, a second primer was used to abolish an MluI site in the IBI-30 polylinker to facilitate the selection of mutated clones. The TATA box included in the sequence 5Ј-ATAAAA-3Ј near position -30 was mutated to a PstI restriction site (5Ј-CTGCAG-3Ј), the CACCC sequence included in the sequence 5Ј-CCACCC-3Ј near position Ϫ140 was mutated to a ScaI site (5Ј-AGTACT-3Ј), and the sequence 5Ј-GGCTGGCT-3Ј beginning at position -58 within the SSE element was also mutated to include a ScaI site (5Ј-AGTACTAG-3Ј) ( Fig. 1). Approximately 260 bp of DNA surrounding each of the mutations was sequenced on both strands, and it was possible to verify that the mutation was present and to establish that there were no other sequence changes.
Stable Transfection Assays and RNA and DNA Analyses-HEL and K562 cells were grown in RPMI media supplemented with 10% fetal calf serum. For each of the six transfections per construct, 2 ϫ 10 6 (HEL) or 1 ϫ 10 6 (K562) cells were electroporated with linearized plasmid DNA consisting of 35 g of the globin construct and 5 g of an SV40 early promoter neomycin resistance gene fusion construct. Two different plasmid preparations per construct were transfected. Pools of transfected cells were selected in 800 g/ml and maintained in 600 g/ml G418 sulfate (Mediatech, Inc.). Pools of cells were used to analyze the results of many different sites of insertion and thereby eliminate position effects. Erythroid differentiation of the transfected HEL and K562 cells was induced by treatment with 25 M hemin for 3 days. RNA was prepared from cells harvested after induction. mRNA expression levels represent the average of at least two primer extension assays quantitated using a Molecular Dynamics PhosphorImager. The oligonucleotide primers used are ␤-globin (5Ј-CAGGGCAGTAACGGCAGA-3Ј), which yields a 95-bp product for the endogenous and a 99-bp product for the marked ␤-globin mRNA, and ␥-globin (5Ј-TGCCCCACAGGCTTGT-GATA-3Ј), which yields a 105-bp product for the endogenous and a 109-bp product for the marked ␥-globin mRNA. The levels of transfected ␤and ␥-globin mRNA are expressed as a ratio of the endogenous ␥-globin mRNA as an internal control. Sequencing experiments to determine the two major alternative start sites for ␥-globin mRNA in cells with the HS3HS2␥TATA␤ constructs were performed using the fmol sequencing kit (Promega) and the above primer from the ␥-globin gene. The statistical analyses were performed using the nonparametric rank sum test, and all findings were judged to be significant at an ␣-level of Յ0.05.

HS3 Exhibits Enhancing Activity but Has a Variable Effect on Competitive Inhibition of ␤-Globin
Expression-Our previous results indicate that HS2 can independently participate in competitive interactions between the ␥and ␤-globin genes in transgenic mice and tissue culture cells (10,13). This work tests the roles of HS3 alone and in combination with HS2 in these competitive interactions and examines the possibility that these HS sites act in concert to enhance ␤and ␥-globin expression in a developmental stage-specific manner.
To test the role of HS3 in the competitive inhibition of ␤-globin expression, pools of HEL and of K562 cells were stably transfected with each of two constructs containing HS3 and a marked ␤or ␥and ␤-globin gene (HS3␤ and HS3␥␤, Fig. 2A). The transfected cells were induced with hemin, and globin gene expression was measured by primer extension assays using human ␤and ␥-globin probes (Fig. 3). The amounts of mRNA from the transfected genes (␥mk and ␤mk) were normalized to that of the endogenous ␥-globin mRNA. The results depicted in Table I show that HS3 alone is an effective enhancer and can drive significant expression of the transfected genes. The transfected ␥-globin mRNA is expressed at means of 21 and 13% in HEL and K562 cells with the HS3␥␤ construct, respectively. When comparing cells with the HS3␤ to those with the HS3␥␤ constructs, however, the presence of the ␥-globin gene in cis does not significantly reduce ␤-globin expression in HEL cells. ␤-Globin expression in HEL cells with and without the ␥-globin gene is comparable at an average of 7 and 4%, respectively (Table I, Fig. 4). Although ␤-globin expression is significantly lower on average in K562 cells with the HS3␥␤ compared with the HS3␤ construct, there is variability between pools of cells. Some pools of cells with HS3␥␤ have less ␤-globin mRNA than HS3␤ pools, but others do not, indicating that HS3 alone does not consistently participate in competitive inhibition in either cell type. The amount of transfected ␤-globin mRNA was divided by the amount of transfected ␥-globin mRNA for each pool of cells with the HS3␥␤ construct to calculate the ␤to ␥-globin mRNA ratios (Table I). The ␤/␥ globin ratios are particularly useful for comparisons between transfections, because these values are not influenced by transfection efficiency. In HEL cells, the mean ␤/␥ ratio is 26% even though the ␥-globin gene is present in cis. The ␥-globin gene therefore does not always compete strongly with the ␤-globin gene in the presence of HS3 alone. These data reveal that HS3 possesses enhancing activity, but compared with HS2, it is less consistently able to participate in the competitive interactions that normally suppress ␤-globin expression in an early erythroid environment. The amount of the transfected ␤-globin gene expressed is generally less for all of the constructs in K562 cells compared with HEL cells, perhaps because of the absence of EKLF in K562 cells (33). For these and all of the transfections with other constructs discussed below, the mRNA expression levels in the tables are not divided by the average number of copies of the genes per cell, because in pools of cells with the same construct, a higher copy number correlated directly with lower expression per gene copy (data not shown).
HS2 and HS3 Act in Concert to Allow Stage-specific Expression of the ␥and ␤-Globin Genes-To investigate the ability of HS3 and HS2 to act together in ␤-globin suppression in early development, pools of HEL and of K562 cells were stably transfected with the HS3HS2␤ and HS3HS2␥␤ constructs ( Fig. 2A). Representative primer extension assays for the HEL cells are shown in Fig. 3. In contrast to the results with the HS3␥␤ and HS3␤ constructs, ␤-globin expression in both cell types is significantly suppressed in the HS3HS2␥␤ compared with the HS3HS2␤ transfected cells. HEL cells containing these constructs exhibit a 10-fold suppression in ␤-globin expression (36.6% for HS3HS2␤ and 3.6% for HS3HS2␥␤), whereas K562 cells show a 4-fold suppression of ␤-globin expression (4.7 and 1.2% ␤-globin mRNA) ( Table I). The developmental specificity of this LCR configuration is also demonstrated by the low ␤/␥ ratio in cells with HS3HS2␥␤ compared with those with HS3␥␤ (Table I, Fig. 4). The ␤/␥ ratios of 0.04 in HEL cells and 0.01 in K562 cells indicate that the adult ␤-globin gene is consistently less able to compete with the ␥-globin gene when both HS3 and HS2 of the LCR are present than with HS3 alone. These data demonstrate that HS2 and HS3 act together to effectively pro- vide developmental specificity.
HS2 and HS3 Together More Effectively Enhance ␥-Globin Expression-The effect of HS2 and HS3 on ␥-globin expression may be elucidated by comparing the levels of ␥-globin mRNA in cells with HS2␥␤, HS3␥␤, and HS3HS2␥␤ constructs. We have previously shown that cells with HS2␥␤ express ␥-globin mRNA at an average of 46 and 53% endogenous ␥-globin mRNA in HEL and K562 cells, respectively (Ref. 13 and data not shown). HEL cells with the HS3␥␤ construct express an average of 21% ␥-globin mRNA compared with the endogenous ␥-globin mRNA, and those with HS3HS2␥␤ express an average of 84% ␥-globin mRNA (Table I). K562 cells with HS3␥␤ and HS3HS2␥␤ express 13 and 91% ␥-globin mRNA, respectively (Table II). In both cell types, ␥-globin expression with the HS3HS2␥␤ construct is significantly higher than with HS3␥␤. Therefore, HS2 and HS3 act in at least an additive manner to effectively enhance ␥-globin gene expression.
The Roles of the TATA, CACCC, and SSE Elements in the ␥-Globin Promoter in Competitive Inhibition-Our previous results are consistent with the hypothesis that sequences 3Ј of base position Ϫ161 in the ␥-globin promoter must be present in cis to inhibit embryonic expression of the ␤-globin gene (13). Mutations in elements downstream of Ϫ161 have been tested in the context of a construct containing HS3, HS2, the Ϫ1350 TABLE I ␤and ␥-globin gene expression in HEL and K562 cells with HS3 and HS3HS2 constructs The mRNAs are designated ␤mk for the transfected ␤-globin gene, ␥mk for the transfected ␥-globin gene, and ␥ for the product of the endogenous gene. The percentages are the mean of at least two primer extension assays. The means represent the average for all pools of cells transfected with the same construct. Data for the HS3HS2␥␤ construct in HEL cells are pooled with those from Table II (transfection numbers 4 -9 corresponding  to transfections 1-6 in Table II and Fig. 5). The ␤mk/␥mk ratio is the amount of ␤-globin divided by ␥-globin mRNA. The asterisks indicate that the mean %␥mk and the mean ␤mk/␥mk ratios in each column are significantly different from each other, that the HS3HS2␤ and HS3HS2␥␤ transfections have a significantly different %␤mk, and that the HS3␤ and HS3␥␤ transfections have a significantly different %␤mk in K562 cells. The statistical analyses were performed using the rank sum test, which is valid in cases where the data do not fit a normal distribution. ␥-globin gene and the ␤-globin gene. HS2 and HS3 were included in the constructs to provide both developmental specificity and strong enhancing activity. Six transfections of the HS3HS2␥␤, HS3HS2␥TATA␤, HS3-HS2␥CACCC␤, and HS3HS2␥SSE␤ constructs depicted in Fig.  2 were performed in HEL and in K562 cells. Representative primer extension analyses for the HEL cells are shown in Fig.  5. The most striking result from these experiments is that the mutation of the TATA box in the HS3HS2␥TATA␤ construct leads to a 10-fold increase in the transfected ␤to ␥-globin mRNA ratio compared with the wild-type HS3HS2␥␤ in HEL cells (0.39 and 0.04, Table II, Fig. 6). This reflects both a 5-fold increase in transfected ␤-globin mRNA and about a 2-fold decrease in ␥-globin mRNA on average (Table II). A similar though less pronounced effect was evident in K562 cells; about a 2-fold increase in transfected ␤-globin mRNA and a 3-fold decrease in ␥-globin mRNA on average was observed with HS3HS2␥TATA␤ compared with HS3HS2␥␤ (Table II). These data suggest that factors that bind to the TATA box are important for stabilizing interactions with HS3 and/or HS2, which allow the ␥-globin gene to compete favorably with the ␤-globin gene in the embryonic/fetal environment.
It was somewhat surprising that the TATA mutation, which obliterates the recognized consensus sequence, reduces ␥-globin gene expression by only 2-to 3-fold compared with the HS3HS2␥␤ construct in HEL and in K562 cells (see "Discussion"). The mutation leads to a great reduction in transcripts beginning at the normal start site and to the use of two major alternative transcription start sites in both HEL and K562 cells, at positions Ϫ8 and Ϫ33 in the ␥-globin promoter (Figs. 1  and 5). The percentages calculated in Table II represent the sum of all the mRNA products.
There is also a significant, greater than 2-fold increase in ␤-globin transcription in the HEL cells with the HS3HS2␥-CACCC␤ construct compared with those with the HS3HS2␥␤ construct (Table II). This is accompanied by a small but signif-icant decrease in ␥-globin gene expression. In K562 cells, there is also a significant increase in ␤-globin mRNA and a significant decrease in ␥-globin gene expression in cells with the HS3HS2␥CACCC␤ construct compared with cells with the HS3HS2␥␤ construct. The transfected ␤to ␥-globin mRNA ratio was significantly increased by 2-or 3-fold in K562 or HEL cells with HS3HS2␥CACCC␤ compared with those with HS3HS2␥␤, representing a shift in the ability of the ␥and ␤-globin genes to compete with each other. This indicates that the CACCC element is also important in promoter competition. Again, it is somewhat surprising that the mutation in the CACCC element has only a small effect on ␥-globin gene transcription, perhaps suggesting that its role is redundant.
As can be seen in Table II, no significant change in the amount of transfected ␥or ␤-globin mRNA expressed resulted from the SSE mutation in either HEL or K562 cells. The ␥and ␤-globin genes in cells with HS3HS2␥SSE␤ were expressed at much the same level as in cells with the HS3HS2␥␤ construct. The data suggest that the SSE plays no role in competitive inhibition of ␤-globin gene expression or at least that another element in the Ϫ1350 ␥-globin gene can compensate for the loss of the Ϫ50 SSE site. DISCUSSION We have demonstrated that HS2 and HS3 each contribute distinct properties to the ␤-globin LCR. Although HS2 was previously shown to participate in competitive interactions between the ␥and ␤-globin genes (10, 13), we find that HS3 does not consistently play a role in these interactions in HEL and K562 cells. Both high level and stage-specific expression of the ␥and ␤-globin genes is observed in cells with HS3HS2␥␤ constructs, because of the presence of both HS3 and HS2. This work also demonstrates that multiple elements downstream of -161 in the ␥-globin promoter, including the CACCC and TATA but not the SSE, are required for competitive inhibition of ␤-globin gene expression.
Our evidence suggests that HS3 alone plays a less consistent role in competitive inhibition than does HS2. This complements and expands upon studies in which deletions of HS3 have no effect on the temporal order of ␤-like globin gene expression (19). Although Fraser et al. (18) observed developmental specificity of ␥and ␤-globin gene expression in two lines of transgenic mice with HS3 constructs, this may not be a general phenomenon because the ␤-globin gene is suppressed more in some pools of cells than in others in this study. The greater ability of the HS3HS2 configuration, compared with HS2 (13) or HS3 alone, to participate in competitive interactions between the ␤and the ␥-globin genes strengthens the hypothesis that the HS sites of the LCR act as a complex to activate transcription by interacting with a single ␤-like globin gene at a time (7). The results from our study using chromosomally integrated genes confirm those of Jackson et al. (15), who found that HS2 and HS3 have additive enhancing activity in transient assays, and those of Bresnick and Tze (16), who have shown that these sites are synergistic in long range enhancer activity in cells with chromosomally integrated genes (16).
The competition model states that the protein-protein interactions that occur between factors binding to the LCR and to the ␥-globin promoter preclude LCR interactions with the ␤-globin promoter (8,9). The finding that HS2 is more important than HS3 in competitive inhibition directs further investigation toward the characterization of the roles of proteins that bind to HS2. The data also suggest that proteins that bind exclusively to HS2, and not to HS3, may be important for the observed properties of HS2. The 215-bp core element of HS2 has a cluster of binding sites for nuclear factors that appear to FIG. 4. The ␤to ␥-globin mRNA ratios for HEL and K562 cells transfected with the HS3␥␤ and HS3HS2␥␤ constructs. The ␤to ␥-globin mRNA ratio indicates the net effect of each LCR configuration on the competition between the ␥and ␤-globin genes. The greater S.D. in the HS3 data series, depicted by the larger relative size of the error bars, demonstrate the greater variability with which HS3 participates in ␤-globin gene suppression. be functionally required for enhancing activity (reviewed in Elnitski et al. (34)) and may therefore take part in interactions with the gene promoters. These proteins include NF-E2/AP-1, GATA-1, YY1, USF (upstream stimulatory factor), and HS2NF5. The binding of NF-E2/AP-1 provides much of the enhancing activity of HS2 (35,36). Mice lacking NF-E2 do not show impaired globin gene expression, suggesting that related family members can compensate for its loss (37,38). Binding sites for the erythroid factor GATA-1 occur in the HS2 core and in the ␥and ␤-globin gene promoters. The levels of GATA-1 vary during development, so it is possible that GATA-1 is involved in developmental regulation (39). Furthermore, mutations in a GATA-1 binding site reduce the enhancing activity of HS2 in transgenic mice (40). The protein YY1, which demonstrates positive and negative activities, interacts with the ␥-globin promoter (41) and with HS2. USF is a broadly distributed E-box-binding protein that interacts with HS2 and may facilitate preinitiation complex formation with other transcription factors on globin promoters (42). HS2NF5, another E-boxbinding protein, provides some of the enhancing activity of HS2 (43,44). USF and HS2NF5 have not been reported to bind to HS3, so they are good candidates to facilitate the role of HS2 in competitive inhibition. Of particular interest, HS2NF5 is the mammalian homolog of a Notch-regulated transcription factor, and this further implicates its potential for developmental control (44). None of these individual sites is essential for the position-independent expression of the ␤-like globin genes, suggesting that chromatin-opening activity may require the cooperation between these and other elements (45), some of which may be upstream of the classical LCR (46). It is unlikely that a TABLE II ␤and ␥-globin gene expression in HEL and K562 cells with ␥-globin promoter mutant constructs The mRNAs are designated ␤mk for the transfected ␤-globin gene, ␥mk for the transfected ␥-globin gene, and ␥ for the product of the endogenous gene. The percentages are the mean of at least two primer extension assays. The mean represents the average for all pools of cells transfected with the same construct. The ␤mk/␥mk ratio is the amount of ␤-globin divided by ␥-globin mRNA. The asterisks indicate that the % or ratio differs significantly (using the rank sum test) from that obtained with the wild-type construct, HS3HS2␥␤. It is not surprising that multiple ␥-globin promoter elements are required to suppress inappropriately early ␤-globin gene expression, because multiple promoter elements have previously been implicated in the response to the LCR, including TATA, CCAAT, and CACCC (47). It is possible that the CACCC element in the ␥-globin promoter plays a role similar to the CACCC in the ␤-globin promoter. EKLF binds preferentially to the ␤rather than the ␥-globin CACCC (29). In mice that have the human ␤-globin locus but lack EKLF, the expression of the ␤-globin gene is reduced, and that of the ␥-globin gene is increased (27,28). Therefore, it is likely that the CACCC in the ␤-globin promoter is influential in switching to ␤-globin expression, and we have now shown that the CACCC in the ␥-globin promoter is also important for switching. Paradoxically, our work indicates that a ␥-globin CACCC mutation can disallow ␤-globin suppression without greatly reducing expression of the ␥-globin gene. This is not the expected outcome from gene competition and suggests that the model will need to be tested further. It is possible that the CACCC and/or other as yet unidentified elements in the ␥-globin promoter is responsible for initiating stage-specific interactions between the ␥-globin gene and the LCR and that the TATA element is important for stabilizing these interactions. The lack of a role for the SSE in this system may reflect the fact that it has not previously been tested using a chromosomally integrated construct or with an intact globin gene rather than a reporter gene (31).
Curiously, the severe mutation we generated in the ␥-globin gene TATA box reduced ␥-globin mRNA by only about 2-or 3-fold. However, the two major transcription start sites utilized in cells with this construct are shifted to positions further upstream than that for the normal gene (-8 and -33, compared with ϩ1, Fig. 1). A similar mutation in the ␤-globin gene TATA box reduces transcription by 7-fold with no aberrant start sites in stable transfection assays in MEL cells with LCR constructs (48). Therefore, it seems likely that other elements in the ␥-globin promoter can compensate in the absence of the ATA-AAA at base position -30. The chicken ␤-globin gene, for example, contains a GATA-1 site at -30 rather than a TATA element. It can bind either GATA-1 or TFIID and is necessary for transcriptional activation. (49) Potential GATA-1 binding sites (50) are located approximately 30 bases upstream from each of the major aberrant start sites in cells with the ␥-globin promoter with the mutated TATA box (GAT at -38 and ATC at -72, Fig. 1), and these may be comparable with that in the chicken ␤-globin gene. Alternatively, or perhaps additionally, the ␥-globin promoter contains Sp1 binding sites in the -50 and -140 regions (51). In TATA-less genes, Sp1 sites upstream from the start site are often required for transcription (52). It is possible that the Sp1 sites in the ␥-globin promoter are utilized in a similar manner in the absence of the TATA box. Our data predict that human mutations in the ␥-globin TATA box may not be seriously deleterious.
In previous work, it was suggested that any transcribed gene intervening between the ␤-globin gene and the LCR might suppress ␤-globin expression. The ␤-globin gene was suppressed in embryonic transgenic mice containing either an ␣-globin (53) or a thymidine kinase-chloramphenicol acetyl transferase construct (26). However, Sabatino et al. (54) demonstrated that specific sequences in the ␥-globin promoter are FIG. 5. Representative primer extension analysis of RNA from HEL cells with the HS3HS2␥␤, HS3HS2␥TATA␤, HS3HS2␥CACCC␤, and HS3HS2␥SSE␤ constructs. The primer extension products are designated ␥mk for the mRNA from the transfected ␥-globin gene; ␥ for the product of the endogenous gene, and ␤mk for the mRNA from the transfected ␤-globin gene.
The arrows indicate major alternative transcription start sites at base positions -8 and -33 for the ␥-globin gene in the HS3HS2-␥TATA␤ construct.
FIG. 6. The ␤to ␥-globin mRNA ratios for HEL and K562 cells transfected with the HS3HS2␥␤, HS3HS2␥TATA␤, HS3HS2␥-CACCC␤, and HS3HS2␥SSE␤ constructs. The ␤to ␥-globin mRNA ratio indicates the net effect of the ␥-globin promoter mutations on the competition between the ␥and ␤-globin genes. The error bars represent the S.D. in each data series. required for competitive inhibition, because a ␥-globin gene with a ␤-spectrin promoter does not suppress ␤-globin expression in embryonic transgenic mice, even though it is transcribed (54). This evidence supports the hypothesis that competitive inhibition of the ␤-globin gene is mediated by specific ␥-globin promoter-LCR interactions, rather than by transcriptional interference by a transcribed gene situated between the LCR and the ␤-globin gene (55). It is possible that the thymidine kinase-chloramphenicol acetyl transferase construct and ␣-globin constructs, which can compete, contain sequences that allow these genes to interact with and sequester the LCR. It was shown that the expression of the ␤-spectrin/␥-globin fusion construct was not influenced by the LCR, suggesting that it does not contain such sequences (54). We have shown that the TATA and CACCC elements are necessary for efficient ␥-globin competition, but we suggest that they are probably not sufficient. Although the ␥-globin, ␣-globin, thymidine kinase, and ␤-spectrin promoters all contain TATA and CACCC/EKLF elements, the ␤-spectrin promoter cannot compete (56 -58). Therefore, it is likely that the precise positions of these elements, and/or that other elements, are also important for gene competition.
By identifying the specific sequences in the ␥-globin promoter and LCR that are required for ␤-globin gene suppression in early erythroid development, candidate proteins that are involved will be implicated for future investigation. This will further delineate the molecular mechanisms of developmental regulation of multiple genes by a distant enhancer.