During chicken embryogenesis, the ρ--globin gene is expressed only in the early developmental stages. We have examined the mechanisms that are responsible for this behavior. The transcription of the ρ--globin gene is strongly correlated with the presence during development of primitive erythroid lineage cells, consistent with the idea that the expression of the ρ--globin gene is restricted to that lineage. The “switching off” of ρ--globin during development thus reflects the change from primitive to definitive cell lineages which occurs during erythropoiesis in chicken. We use transient expression assays in primary erythroid and other cells to show that the information for lineage- and tissue-specific expression of the ρ--globin gene is contained in a 456 bp region upstream of the gene’s translational start site. DNA-binding studies, coupled with analysis of the effect on expression of deletions and binding site mutations, were used to identify important control elements within this 456 bp region. We find that binding sites for the ubiquitous transcription factor Sp1, and the specific hematopoietic factor GATA-1, are crucial for expression of the gene in primitive erythroid cells. Quantitative analysis shows that nuclei of the primitive erythroid lineage contain 10-fold more of these factors than do the nuclei of definitive cells. We show that in principle these differences in factor concentration are sufficient to explain the lineage-specific behavior that we observe in our assays. We suggest that this may be an important part of the mechanism for lineage-restricted ρ--globin expression during chicken erythroid development. Similar mechanisms may be involved in regulation of other (but not all) members of the globin family.

The chicken α- and β -globin loci constitute two families of highly conserved genes whose transcription is tightly regulated during embryonic development. While the αA- and αD-globin genes are constitutively expressed through-out development, the other members of the two globin gene families exhibit differential activity as embryogenesis proceeds. In the early stages of development, the α π-, ρ-- and ∈ -globin genes (the embryonic globin genes) are active, while the β H- and β A-globin genes (the adult globin genes) are silent. Later in chick embryogenesis, this pattern is inverted, with the β H- and β A-globin genes active and the embryonic genes silent (for a review, see Bruns and Ingram, 1973). The underlying cellular and molecular basis for this ‘globin switch’ is not well understood.

Considerable evidence supports a clonal model of switching in chickens (Ingram, 1972; for review, see Nikin-maa, 1990). According to this model, cells expressing embryonic or adult genes derive from distinct cell lineages, each programmed to express exclusively one or the other set of globin genes. Chickens have two erythroid cell types: the primitive lineage cells, which predominate in the early stages of development (3-6 days), and the definitive lineage cells which become the majority cell type at later stages (day 7 on) of embryogenesis.

Although erythroid stem cells are committed to one of these two cell lineages quite early in hematopoiesis (Beau-pain, 1985; Dieterlen-Lievre, 1988), some of the mechanisms for lineage-specific globin expression apparently remain active throughout the lifetime of the red cell. Transient expression studies in primary erythroid cells have shown that the promoters of the β A- and ∈ -globin genes carry sufficient information to confer stage-specific expression (Choi and Engel, 1988; Hesse et al., 1986; Lieber et al., 1987; Nickol and Felsenfeld, 1988). By implication, the mechanism necessary for this behavior must persist in these cells. The chicken β A-globin promoter, and the β/ ∈ enhancer lying between the two genes, have been particularly well characterized. They contain regulatory elements common to all globin genes, as well as some unique to the β A-globin gene. In particular, expression of the β A gene in definitive lineage cells seems to be dependent not only on the presence of the enhancer, but also upon a stage selector element (SSE; Choi and Engel, 1988) located within the gene’s promoter (−39 bp, AAGAGGAGGGG). The SSE appears to interact with either the ubiquitous transcription factor Sp1 or a definitive lineage specific factor, NF-E4 (Evans, et al., 1990; Gallarda et al., 1989), which may play a critical role in stage-specific expression of the βA-globin gene. Other factors that bind to A sequences common to all globin genes may also play roles in modulating transcriptional activities (Lewis et al., 1988; Clark et al., 1990; Jackson et al., 1989; Emerson et al., 1989).

Despite the presence of several common regulatory motifs, there is no reason to think that the mode of globin gene regulation in primitive lineage cells will necessarily resemble that found in the definitive lineage. In order to investigate the mechanism underlying the primitive lineage-specific expression of embryonic globins, we have focused on the ρ--globin gene, a member of the -cluster of genes. We find that constructs containing 456 bp of sequence 5′ of the ρ-gene translational start site mirror the lineage specificity of endogenous ρ--globin gene expression, and thus contain the necessary information to direct lineage, as well as tissue-specific expression. Deletion and site mutagenesis reveal that both Sp1 and GATA-1 binding sites are essential to this behavior. Finally, we show by direct binding studies that the amounts of Sp1 and GATA-1 factors vary substantially between primitive and definitive cell nuclei, with primitive lineage cell nuclei containing as much as 10-fold more of these factors than definitive nuclei. These data suggest that the lineage-specific expression of the ρ--globin gene may depend, at least in part, upon lineage-specific differences in transcription factor abundance.

Plasmids

The ρCAT vector (Fig. 1A) was built in two steps. First, the region between the NcoI site adjacent to the ρ--globin gene translational start site and the SmaI site just 5′ of the cap site was synthesized (66 bp), and inserted into the SmaI polylinker site of pUC18CAT(#7) (gift from J. Hesse; Hesse et al., 1986). The 390 bp upstream of that region was isolated as a SmaI subfragment from pB2H2 (Dolan et al., 1981), and subcloned into the first construct. The DNA sequence of this 456 bp region (see Fig. 3) was determined by standard Sequenase (USB) protocols using appropriate oligonucleotide primers derived from the known 5′ sequence of the p globin gene (Dodgson et al., 1983). The β ACAT, β ACATE (pACAT and pACATE respectively in Hesse et al., 1986) and RSVCAT constructs were provided by J. Hesse (Hesse et al., 1986), and the αACATI construct (A′:J in Knezetic and Felsenfeld, 1989) was obtained from J. Knezetic.

Fig. 1.

(A) Schematic of the pCAT expression vector. (B) Activities of pCAT, αACATI, and β ACATE following transient transfection into primitive (5-day) and definitive (12-day) cells. Values are normalized to the mean of RSVCAT (as a control for transfection efficiency) ± s.e.m. (n>10).

Fig. 1.

(A) Schematic of the pCAT expression vector. (B) Activities of pCAT, αACATI, and β ACATE following transient transfection into primitive (5-day) and definitive (12-day) cells. Values are normalized to the mean of RSVCAT (as a control for transfection efficiency) ± s.e.m. (n>10).

Deletion mutants of the ρ-CAT construct were made by first isolating the 390 bp SmaI fragment from the p insert. This was cut with the appropriate restriction enzymes (Fig. 2) and the desired subfragments were gel purified, filled in with Klenow enzyme, and recloned into the SmaI cleaved vector employing standard methods (Maniatis et al.,1983). Cluster mutants were generated with an Amersham in vitro mutagenesis kit, using the synthetic oligonucleotides described below and in the text. Each ρCAT mutant was given a numerical designation reflecting the oligomer used to generate the mutant site, and the integrity of each mutant was confirmed by sequencing. All plasmids were purified by either standard CsCl gradient centrifugation (Maniatis et al.,1983) or with a Qiagen plasmid Maxi kit (Qiagen). Plasmids isolated by either method gave similar results.

Fig. 2.

Activities of pCAT deletion constructs transfected into primitive (5-day) and definitive (12-day) cells (normalized to the mean of wild-type pCAT) ± s.e.m. (n>10). BanI is located at −246 bp upstream of the cap site. Similarly, the BspMII site is at −181 bp, the HpaII site at −98 bp, and the SmaI site is at −23 bp.

Fig. 2.

Activities of pCAT deletion constructs transfected into primitive (5-day) and definitive (12-day) cells (normalized to the mean of wild-type pCAT) ± s.e.m. (n>10). BanI is located at −246 bp upstream of the cap site. Similarly, the BspMII site is at −181 bp, the HpaII site at −98 bp, and the SmaI site is at −23 bp.

Cells

Circulating red cells were isolated in bulk from appropriately aged White Leghorn chicken embryos (Truslow Farms, Chesterton, MD). Briefly, cells were isolated from early eggs (days 4-7, stages 24-31 at 37°C according to Hamburger and Hamilton, 1951) by collection of 150 embryos into PBS at room temperature. Periodically, these early embryos were gently lacerated with sharp scissors to promote bleeding. Once collection was completed, the embryo bodies were removed by filtration through gauze, and the cells were pelleted at 1000 revs/minute for 5 minutes at 5°C in a Sorvall RT6000B centrifuge. Cells were further purified on Lymphocyte Separation Medium (LSM, Organon Teknika) according to the manufacturer’s instructions. Cells were isolated from later eggs (days 8-12, stages 35-38 at 37°C according to Hamburger and Hamilton, 1951) by venous puncture of 48 embryos and col-lection of blood into PBS at room temperature, and purified as described above. Typical yields per preparation for 4.5-5-day embryos were ∼1.5 × 109 cells; 9-12-day preparations produced ∼3 × 109 cells. Smears were made of every preparation of cells and stained by the May-Grünwald Giemsa method (Lucas and Jamroz, 1961). It should be stressed that staging of embryos according to the photographic series of Hamburger and Hamilton (1951) was critical, and that simple chronological age was inad-equate for estimating developmental age.

Chick embryo fibroblasts (CEFs) were prepared from 12-day embryos according to standard protocols (Groudine and Wein-traub, 1982), and grown in 8% fetal calf serum (FCS; GIBCO), 2% chick serum (CKS; GIBCO) in Dulbecco’s Modified Eagles Medium (DMEM; GIBCO).

RNA isolation and primer extension analysis

Total cellular RNA from circulating red cells of each day of embryonic development was prepared from frozen cell pellets using the guanidine thiocyanate method (Chirgwin et al., 1979). Primer extension (G. Felsenfeld and C. Trainor, unpublished; Reitman et al. 1990; Townes et al., 1985) was used to analyse the RNA. Each reaction used RNA from ∼5 × 106 cells. Synthetic oligonucleotide probes for the globin gene RNAs used as primers, and their respective extension products are described in Table 1. The 20 nucleotide primer used to detect GATA-1 mRNA has been described (Hannon et al., 1991). Products were quantitatively analyzed on a Molecular Dynamics PhosphorImager. Overall RNA integrity in each preparation was indicated by the presence of undegraded 18S and 28S RNA species in ethidium bromide stained 1% agarose/MOPS-formaldehyde gels.

Table 1.

Primer extension probes

Primer extension probes
Primer extension probes

Transient transfection of primitive and definitive red cells

Mature primitive (from 4.5-5-day embryos, stages 24-27) and definitive (from 9-or 12-day embryos, stages 35 and 38, respectively) erythroid cells were transfected with various plasmids as described previously (Hesse et al., 1986; Lieber et al., 1987), except that 13 A412 units of cells were used rather than 26. In both cell types, 13 A412 units of cells were equivalent to about 3 × 107 cells. Following transfection, cells were maintained in Liebovitz’s L15 media (GIBCO) + serum (see below) for 48 hours, harvested in 0.1 M NH4HCO3 (containing 0.1 mM phenylmethyl-sulfonylfluoride, PMSF) and frozen according to standard procedures. Transfected cell extracts were assayed for chloramphenicol acetyl transferase (CAT) activity as previously described (Hesse et al., 1986; Lieber et al., 1987). CEFs were transfected by the DEAE-dextran method (Hesse et al.,1986), harvested by scraping into 0.1 M NH4HCO3 (+ PMSF) and assayed according to standard methods (Hesse et al., 1986; Lieber et al., 1987). Acetylated products of [14C]chloramphenicol were assayed either by liquid scintillation counting or by PhosphorImager. All CAT activities were normalized to that of RSVCAT, which was transfected in parallel in every experiment to control for differences in transfection in the different cell types.

Initial results showed only low activity with standard post-trans-fection media (5% FCS, 2% CKS, and antibiotic-antimycotic mix), but it was found that increasing the amount of chick serum to 30% dramatically increased the ρ-CAT activity in primitive cells, while leaving RSVCAT controls virtually unchanged. The activity of the endogenous ρ--globin gene in mock-transfected primitive cells, as assayed by RNA dot blot (Blackman et al., 1986), showed a similar dependence upon serum concentration. In contrast, FCS did not have any stimulatory effect, indicating that factors specific to CKS were responsible. The CKS did not, however, have a significant effect upon definitive cells. All subsequent erythroid transfections were carried out using media containing 30% chick serum. This dependence on factors present in chick serum is not entirely surprising in the light of previous reports of erythropoiesis-inducing factors in chick serum (Coll and Ingram, 1978; Samarut, 1978, 1979; Samarut and Nigon, 1976), and a more recent report of in vitro erythropoiesis induction with purified TGF α, erythropoietin and insulin (Pain et al., 1991).

Nuclear extracts, DNAase I footprints and electrophoretic mobility shift assays

Crude nuclear extracts were prepared essentially as described previously (Jackson et al. 1989; Evans et al. 1988), with the following modifications: (1) pellets of either fresh or frozen cells were used, with no apparent differences between the two and (2) no dialysis step was used, and samples were instead frozen (−70°C) at this point. Buffer concentrations were adjusted prior to use by addition of 1 × binding buffer (Lewis et al., 1988). These two changes resulted in lower levels of degradation and a 5-fold increase in yield compared to previous methods. The integrity of extracts was tested on 15-20% SDS-polyacrylamide gels (run on a Phamacia PhastSystem apparatus), which revealed no substan-tial general degradation upon staining with Coomassie Blue, as judged by the presence of intact globin and other bands (all nuclear extracts unavoidably contained some contaminating hemoglobins, attributable to their tremendous abundance in the red cells). DNAase I footprinting on a 298 bp BanI/BamHI subfragment of ρCAT endlabeled at the Bam H1 site was performed according to methods described by Kadonaga (1990).

Electrophoretic mobility shift assays followed methods previously described (Minie, 1986; Singh et al., 1986), except that 30 cm long gels were used and both poly d[I.C] and poly [d(A-T)] served as non-specific competitors. In assays for abundances of various factors, both Sp1 and BGP1 were measured using the dsRGP2.0 probe (see Table 2), GATA-1 using the dsRGP3.0 probe, and PAL using oligomers containing either an α π PAL site or β A PAL site (provided by G. Felsenfeld and J. Knezetic, unpublished results, and Jackson et al., 1989), which gave comparable results. Sp1 was also measured using an oligonucleotide containing a canonical SV40 Sp1 binding site (Lewis et al., 1988), and gave essentially the same results as those for the dsRGP2.0 probe. Quantitation was done with a Molecular Dynamics PhosphorIm-ager.

Table 2.

Mobility shift probes, upper strands only

Mobility shift probes, upper strands only
Mobility shift probes, upper strands only

Measuring relative factor concentration

The relative amounts of a DNA-binding factor P in different extracts can be obtained from gel mobility shift titrations, even when a DNA competitor such as poly d[I.C] is used to suppress binding by non-specific factors present in the extract. In typical measurements, the specific DNA fragment is in large excess, so that the free DNA concentration is nearly equal to the total DNA concentration, i.e. (D)D0. Under such circumstances, we observe that the fraction of DNA in complex is directly proportional to the amount of specific protein factor added. In one set of experiments (see insert, Fig. 7C), we added increasing amounts of nuclear extract to fixed amounts of DNA probe, and showed that the fraction of the probe migrating as complex in the gel shift assay was linearly dependent on the amount of extract added. In other experiments, we used a pure synthetic peptide containing a single GATA-1 finger as a test molecule (G. Felsenfeld, unpublished data). Again, the fraction of gel-shifted complex was proportional to the amount of peptide added.

Analysis of the equilibrium binding equations shows why this method works. Let P0 be the total concentration of protein factor P and N0 the concentration of non-specific competitor DNA (in sufficient amount so that its concentration is not depleted by binding). The constant for specific binding is K1= (DP)/(D)(P), and for non-specific binding KN= (NP)/(N)(P), where (DP) is the con-centration of specific complex and (NP) is the concentration of non-specific complex. It is easy to show that

formula

Under these conditions, the value of P0 is proportional to the amount of complex whether or not most of the factor P is bound non-specifically, and it is possible to compare relative amounts of a given factor in different extracts. Provided the total probe concentration D0 is the same in all experiments, the ratio of band intensity in the complex to unshifted band intensity is a measure of P0, consistent with the experimental results.

Oligonucleotides

All oligonucleotides were synthesized on an Applied Biosystems Synthesizer and gel purified following standard protocols. Sequences of oligomers used for primer extension experiments are given in Table 1 along with the expected extension products (derived from Dodgson and Engel, 1983; Dodgson et al., 1983). Sequences of oligomers used in mobility shift experiments are given in Table 2. All oligomers for these experiments were end-labeled as single strands with T4 kinase as described by Lewis et al. (1988). Those used as mobility shift probes were made double stranded by annealing with their complementary strands using conditions described in Lewis et al. (1988).

The 456 bp 5′ of the ρ--globin gene contain the sequences necessary for lineage and tissue specific regulation

A segment of DNA containing the 456 bp 5′ of the ρ-globin gene translational start site was subcloned into a pUC18CAT reporter vector (Hesse et al., 1986, and Mate-rials and methods). This region is associated with a DNAase I hypersensitive site specific to the primitive erythroid lineage (Stalder et al., 1980; Reitman and Felsenfeld, 1990). This construct, designated ρCAT (see Fig. 1A), was used to explore whether elements in this region of the sequence were involved in the regulation of ρ-globin gene transcription. The ρCAT construct was transiently transfected into primitive lineage red cells (Hesse et al., 1986; Lieber et al., 1987), and CAT activity was assayed following 48 hours of culture in optimal media (30% v/v chick serum; see Materials and methods). ρCAT was active in the primitive lineage cells, and showed 25% of the activity seen with the RSVCAT control (Fig. 1B). Transient transfection of the ρCAT vector into definitive lineage cells showed it was nearly 10-fold less active in the definitive cells, with activity only slightly in excess of a pUC18CAT control. Thus, the ρCAT construct mimicked the lineage specificity shown by the endogenous ρ-globin gene (Figs 1B and 7B). Similarly, the pattern of expression from αACATI (A:J′ in Knezetic and Felsenfeld, 1990) and β ACATE (a β A-globin expression vector, pACATE in Hesse et al., 1986) paralleled the levels of their endogenous counterparts in primitive and definitive lineage cells (see Figs 1B and 7B). Furthermore, ρCAT failed to show activity over pUC18CAT levels in CEFs, even though an RSVCAT control showed high levels of CAT activity in those cells (data not shown). We conclude that the ρ-gene promoter elements contained in ρCAT must carry the necessary control elements for both the red cell lineage- and tissue-specific regulation of ρ-globin gene expression.

Deletions in ρCAT define regions containing regulatory elements

In order to analyze this promoter, a series of plasmids having various deletions in the p promoter but otherwise identical to ρCAT were constructed (see Materials and methods) and transiently transfected into both primitive and definitive red cells (Fig. 2). Truncation of the promoter from its 5′ side to the BanI site (−246 bp) resulted in only a 30% loss of activity in the primitive cell background compared to wild type. However, further truncation to the BspMII site (−181 bp) resulted in a 70% loss of activity relative to the wild type, and removal of sequence to the HpaII site (−98 bp) left merely 6%. These results suggested that sequences located within the region spanning the BanI to HpaII sites are critical to the regulation of the ρ-globin gene’s transcriptional activity. Site mutation experiments described below confirm this view. None of the deletion constructions gave activity greater than that of the pUC18CAT background control in definitive lineage cells.

Sequence analysis of the 5′ upstream region of the ρ-globin gene

Examination of the 456 bp of 5′ sequence of the ρ-globin gene indicated the presence of several regulatory elements common to other globin genes (see Fig. 3). First, there are two minimal BGP1 sites (GGGGGGG) (Lewis et al., 1988; Clark et al., 1990) in opposite orientation to one another, located at −279 and −125 bp, referred to here as BGP1distal and BGP1proximal respectively. The proximal BGP1 site is in opposite orientation to the distal one. Additionally, there is a single GATA-1 site (AGATAAG) (Kemper et al., 1987; Evans et al. 1988) located at −221 bp. There are also a single, non-canonical Sp1 site (GGGGTGGG) (Li et al., 1991) located at −50 bp and a TATA box located at −25 bp.

Fig. 3.

Sequence of 456 bp ρ-globin gene insert in ρCAT vector. Sites for defined factor binding are highlighted.

Fig. 3.

Sequence of 456 bp ρ-globin gene insert in ρCAT vector. Sites for defined factor binding are highlighted.

DNAase I footprinting of the region between −246 bp and the translational start site indicated a complex pattern of protection of these sequences by binding factors from extracts of both primitive and definitive nuclei (data not shown). Large regions of protection were seen between the BanI site and the cap site of the gene, and it was not possible clearly to correlate specific protected regions with discrete binding sites. The complexity of the binding necessitated further dissection of the region using specific double-stranded oligonucleotide probes in an eletrophoretic mobility shift assay.

The 456 bp of ρ-globin gene sequence was scanned for specific binding activities by an electrophoretic mobility shift assay using a series of double-stranded synthetic oligonucleotide probes. These double-stranded ρ-globin promoter (dsRGP) probes covered each region of the ρ-globin gene upstream sequence, and made it possible to dis-sect out and identify the various binding species (see Table 2 for list of probes). As shown in Fig. 4A-D, these oligomers produced discrete binding species whose conjugate DNA-binding sites could be readily identified by com-petition with both wild-type and mutant probes. In typical experiments, labeled probes containing putative binding sites were competed with unlabeled oligomers containing either mutated sites or binding sites for other factors (see Fig. 4 for details).

Fig. 4.

Electrophoretic mobility shift assays of various regions of the ρ-globin upstream region. Assays were as described in Materials and methods with duplex oligonucleotides described in Table 2 and in the text. The amounts of extracts used were optimized to allow easy resolution of the two Sp1 complexes. All cold competitions were done with 100× molar excess relative to labeled probe. Note that the open arrow indicates a degradation product of BGP1 which has been reported elswhere (Clark et al. 1990). (A) dsRGP4.0 (B) dsRGP3.0 (C) dsRGP2.0 (D) dsRGP1.0.

Fig. 4.

Electrophoretic mobility shift assays of various regions of the ρ-globin upstream region. Assays were as described in Materials and methods with duplex oligonucleotides described in Table 2 and in the text. The amounts of extracts used were optimized to allow easy resolution of the two Sp1 complexes. All cold competitions were done with 100× molar excess relative to labeled probe. Note that the open arrow indicates a degradation product of BGP1 which has been reported elswhere (Clark et al. 1990). (A) dsRGP4.0 (B) dsRGP3.0 (C) dsRGP2.0 (D) dsRGP1.0.

As shown in Fig. 4A, the probe dsRGP4.0 forms a complex, in agreement with the sequence analysis suggesting that this region contains a BGP1 site. The low levels of binding activity seen, however, indicate that this site is weak, consistent with previous observations of BGP1 binding to a minimal seven Gs (Clark et al., 1990), and consistent with the deletion of this region having little effect upon activity (Fig. 2). No other high affinity binding complexes were observed with either primitive or definitive red cell nuclear extracts, indicating that there are no other binding sites in this region. A mutant in which the BGP1 binding site was disrupted, dsRGP4.1, was unable to compete for binding in a cold competition assay, further identifying the distal G-string as a BGP1 binding site. A random duplex oligomer, dsRandom (Lewis et al., 1988; Clark et al., 1990), also failed to compete.

The single GATA site in the dsRGP3.0 (see Fig. 4B) probe clearly binds to GATA-1 factor. No other specific binding complexes are apparent, indicating no other binding sites exist in this region. This complex could be competed with a number of GATA probes (data not shown), but was not competed with a dsRGP3.0 analogue (dsRGP3.1) in which the GATA-1 site has been mutated to a non-binding form.

Two distinct sets of binding complexes (Fig. 4C) are associated with the region defined by the dsRGP2.0 probe: (1) an upper pair of complexes, which could be competed with oligomers containing Sp1 sites, as well as with an oligomer containing a minimal BGP1 site (rhoG1; Clark et al., 1990) and (2) a lower complex, which is competed only with the oligomer containing the minimal BGP1 site. This suggests that both Sp1 and BGP1 factors bind to sites within this region. No canonical Sp1 sites appear in this region, but it was possible to show with mutant oligomers that Sp1 binding indeed overlapped with the BGP1 binding site. As shown in Fig. 4C, mutant oligomers dsRGP2.1 and 2.2 are both incapable of competing out the upper bands, while only dsRGP2.1 affects the BGP1 complex. The dsRGP2.1 oligomer contains mutations at the 5′ border of the proximal BGP1 binding site, while dsRGP2.2 contains a disabled proximal BGP1 binding site. Conversely, dsRGP2.1 complexed with BGP1 but not Sp1, while dsRGP2.2 bound neither factor (data not shown).

The dsRGP1.0 oligomer (see Fig. 4D) also appeared to generate two complexes similar to the upper complex bands seen with dsRGP2.0, and the binding site of these factors was shown to be the GGGGTGGG sequence by the lack of competition with a cold mutant oligomer in which that site was destroyed (dsRGP1.1). Furthermore, both dsRGP1.0 and dsRGP2.0 oligomers mutually compete for these two complexes (data not shown), indicating that the same factor, Sp1, probably binds to both sites. This is the only observable factor binding within this region. The site within the dsRGP1.0 region is designated the Sp1proximal site, while the site in the dsRGP2.0 region is referred to as the Sp1distal. A TATA box is also present in this region, but no TATA site-dependent binding complexes are observed under the conditions used, consistent with the instability of higher vertebrate TFIID reported by others (Buratowski et al., 1988).

Both dsRGP1.0 and dsRGP2.0 Sp1 complexes were competed with a series of Sp1 sites derived from the A-globin gene promoter (a gift from T. Evans) as seen in Fig. 5A (data shown for dsRGP2.0 only). The identity of the factors causing the upper complexes seen with both probes was further verified by adding antisera specific for Sp1 (the kind gift of Drs R. Tjian and S. Jackson). The amount of the upper complexes was depleted by the anti-Sp1 antisera due to hypershifting and binding interference (Fig. 5B). These antisera had no effect upon the BGP1 complex. Preimmune control antisera had no effect upon the band shift patterns. Similar results were obtained for the complexes formed with the dsRGP1.0 probe (data not shown).

Fig. 5.

(A) Cold competition of complexes formed with dsRGP2.0 using various oligomers. All competitions used a 100× molar excess of cold oligonucleotides, which were added to reactions prior to extracts. The open arrow indicates a degradation product of BGP1. (B) Depletion of upper dsRGP2.0 complexes by Sp1 antisera. 2 μl of sera were added to each reaction prior to the addition of extracts. After extracts were added, reactions were incubated for 5 minutes at room temperature, and then loaded onto gels. For convenience of presentation, the top of the gel is not shown, but a new complex of much slower mobility was seen in the anti-Sp1 sera-treated samples, suggesting hypershifting.

Fig. 5.

(A) Cold competition of complexes formed with dsRGP2.0 using various oligomers. All competitions used a 100× molar excess of cold oligonucleotides, which were added to reactions prior to extracts. The open arrow indicates a degradation product of BGP1. (B) Depletion of upper dsRGP2.0 complexes by Sp1 antisera. 2 μl of sera were added to each reaction prior to the addition of extracts. After extracts were added, reactions were incubated for 5 minutes at room temperature, and then loaded onto gels. For convenience of presentation, the top of the gel is not shown, but a new complex of much slower mobility was seen in the anti-Sp1 sera-treated samples, suggesting hypershifting.

Site mutagenesis of the GATA-1, proximal Sp1, and TATA sites

Having identified factor binding sites, we proceeded to demonstrate that they are essential for ρ-gene expression. Cluster mutants were introduced into the GATA, Sp1proximal and TATA sites by in vitro mutagenesis (Materials and methods), rendering each site unable to bind to its cognate factor. A series of ρCAT derivatives, each containing one of the sites in its altered form, were tested for CAT expression in primitive and definitive cells (Fig. 6). A defective binding site for any one of these three factors results in at least a 10-fold reduction in the activity of ρCAT. These observations are consistent with data obtained from deletion mutants and strongly suggest a critical role for each of these factors in the regulation of ρ-globin gene activity.

Fig. 6.

Activities of ρCAT cluster mutants (normalized to the mean of wild-type ρCAT) ± s.e.m. (n>10). Transfections were into primitive (5-day) and definitive (9-day) cells.

Fig. 6.

Activities of ρCAT cluster mutants (normalized to the mean of wild-type ρCAT) ± s.e.m. (n>10). Transfections were into primitive (5-day) and definitive (9-day) cells.

Primitive red cell nuclei contain larger amounts of Sp1 and GATA-1 than do definitive red cell nuclei

We wished to determine the relative nuclear abundance of various erythroid cell components, and especially regulatory factors, at various developmental stages. To identify erythroid cell types, smears of blood from each day of chick development were made, stained by the May-Grünwald Giemsa method (Lucas and Jamroz, 1961), and scored for the presence of primitive and definitive cells. There are two distinct types of red cell lineages within chick embryonic development; their abundance in the embryonic circulation as a function of the number of days of development is shown in Fig. 7A.

Fig. 7.

(A) Smears of blood samples used for the RNA expression course were stained by the May-Grünwald Giemsa method and then scored (out of several random fields of ∼300 cells) for the presence of primitive and definitive erythroid cells. (B) Accumulation of ρ-, βA-, αA-globin RNA and GATA-1 RNA per cell (5 × 106 cells/primer extension reaction) as a function of chick development. Each RNA was measured using a specific oligonucleotide probe in a primer extension assay. The products of the extension assay were quantitatively analyzed on a Molecular Dynamics PhosphorImager. The highest value for each RNA was arbitrarily set to 1. (C) Estimations of the amounts of Sp1, GATA-1 and PAL factors as a function of development measured as described in Materials and methods. Values for each factor were normalized to the highest value measured for that particular factor. Sp1 was measured using the dsRGP2.0 probe, GATA-1 with the dsRGP3.0 probe and PAL with the α π PAL probe (see Materials and methods). The amounts of factors were measured with extract from 1.4 × 106 erythroid cells per lane. Inset shows linear response of assay for Sp1 in crude nuclear extract.

Fig. 7.

(A) Smears of blood samples used for the RNA expression course were stained by the May-Grünwald Giemsa method and then scored (out of several random fields of ∼300 cells) for the presence of primitive and definitive erythroid cells. (B) Accumulation of ρ-, βA-, αA-globin RNA and GATA-1 RNA per cell (5 × 106 cells/primer extension reaction) as a function of chick development. Each RNA was measured using a specific oligonucleotide probe in a primer extension assay. The products of the extension assay were quantitatively analyzed on a Molecular Dynamics PhosphorImager. The highest value for each RNA was arbitrarily set to 1. (C) Estimations of the amounts of Sp1, GATA-1 and PAL factors as a function of development measured as described in Materials and methods. Values for each factor were normalized to the highest value measured for that particular factor. Sp1 was measured using the dsRGP2.0 probe, GATA-1 with the dsRGP3.0 probe and PAL with the α π PAL probe (see Materials and methods). The amounts of factors were measured with extract from 1.4 × 106 erythroid cells per lane. Inset shows linear response of assay for Sp1 in crude nuclear extract.

We next analyzed the expression of the endogenous ρ-globin gene, as well as β A- and αA-globin as a function of chick development. A primer extension assay was used to measure the accumulated mRNAs of a variety of globin genes. The globin genes are highly conserved, making probe cross reaction a serious problem. The use of primer extension allowed the development of probes which were highly specific for each globin transcript (Dodgson and Engel, 1983; Dodgson et al., 1983; see Materials and meth-ods), thus eliminating this problem.

The expression of ρ-globin,β A-globin and αA-globin mRNAs was assessed on a cellular basis at various stages of chick development. ρ-globin mRNA accumulation peaked at day 5 of chick embryo development, and then decreased 10-fold by day 9 (Fig. 7B). The pattern of expression of β A mRNA was nearly the inverse of the ρ-globin pattern, while the αA-globin pattern showed a significant level of expression for that gene at all stages of development (Fig. 7B). These patterns of globin gene expression agree well with earlier globin protein and RNA measurements (Bruns and Ingram, 1973; Dodgson and Engel, 1983; Dodgson et al., 1983; Landes et al., 1982; Lois and Martinson, 1989).

We compared the relative abundance of individual DNA-binding factors using the electrophoretic mobility shift assay (see Materials and methods). Nuclear extracts from the circulating red cells of 4-12 day (inclusive) embryos were prepared by methods that reduced the opportunity for differential losses of factors and the extracts were assayed for each of several binding species by gel mobility shift assays (see Materials and methods and Jackson et al., 1989). The amounts of Sp1 and GATA-1 per cell vary dramatically as the erythroid cell lineage switch proceeds in embryonic chick development (Fig. 7C). The amounts of Sp1 and GATA-1 BGP1 in primitive cells are 10-fold greater than the amounts found in definitive lineage cells. In contrast, the amount of PAL site binding factor remains relatively stable in embryonic development, consistent with previous observations (Jackson et al., 1989; see Discussion), suggesting that the lower amounts of the other factors are not due to differential extraction. Quantitation of the extracts used in Fig. 7C was repeated three times, with similar results (the ratio of 5-day to 11-day abundance was 9 ± 0.5 for Sp1 and 11.6 ± 0.9 for GATA-1). The extraction process itself was repeated on other cell preparations; the 5-day:11-day ratio was 12.7 ± 3 and 15.2 ± 4 for Sp1 and GATA-1, respectively. The abudance of factors was determined by mobility shift assays using extract from 1.4 × 106 erythroid cells for each day of development. As a further control, the mobility shift experiments were also normalized to the total nuclear protein concentration in the preparation, with results essentially identical to those obtained by using cell equivalents. It should be noted that this assay measures the nuclear concentration of factor molecules capable of binding to DNA. It obviously will not detect modified factors that cannot function in binding or transactivation.

We also carried out primer extension assays to measure the abundance of GATA-1 mRNA in these cells. As shown in Fig. 7B, this message undergoes a 4-fold decrease in cellular abundance as development proceeds, consistent with the decrease in GATA-1 mRNA seen in avian virus trans-formed erythroblasts upon globin induction reported by Yamamoto et al.(1990). The quite different abundance pro-files of the globin mRNAs (Fig. 7B) suggest that the decrease in GATA-1 mRNA in definitive lineage cells does not reflect general RNA loss or degradation. (Because chicken Sp1 has not been cloned, it is not yet possible to monitor the abundance of Sp1 mRNA in the same way.)

These results imply that the nuclear concentrations of Sp1 and GATA-1 factors, both critical to the activity of the ρ-globin gene, are very different in primitive and definitive erythroid cells. The observed decrease in GATA-1 mRNA parallels the decrease in factor concentration.

We ultimately wish to understand the ‘switching’ mecha-nism responsible for the difference between the primitive lineage cells present in the circulation early in development, which express α1r, p and E, and the cells of the definitive lineage present at later stages, which express H and A. ( αA and αD are expressed in both lineages). ρ-globin mRNA accumulation correlates with the presence of primitive lineage erythroid cells in chicken embryogenesis. Conversely, β A mRNA correlates with the appearance of definitive lineage cells. These data are consistent with the clonal model of hemoglobin switching (see Introduction). Direct analysis of the globin proteins (Fucci et al., 1983, 1987; Mahoney et al. 1977; Schalekamp and Van Goor, 1984), and similar analysis of the mRNA species (Lois and Martinson, 1989) present in purified populations of primitive and definitive cells support this view. Furthermore, immunofluorescence studies using antibodies specific for embryonic and adult hemoglobins demonstrate lineage-specific restriction at the level of the individual cell (Beaupain, 1985; Shimizu, 1976). Such studies also show that this restriction of globin expression occurs even at the earliest stages of primitive and definitive erythroid differentiation in the yolk sac (Beaupain, 1985). This would suggest that the program for selecting which globin genes are to be expressed is activated at early times in development, perhaps at the stem cell level. Indeed, experiments using chicken/quail chimeras indicate that primitive and definitive erythroid cells may originate from two distinct, separate, precursor cell populations (Dieterlen-Lievre, 1975; for review, see Dieterlen-Lievre, 1988). The basis for this selection program is not known, but the mechanism responsible for it is maintained throughout the lifetime of the cell and is thus amenable to study by the methods used in this paper.

Sequences in the ρ 5 ′ flanking region needed for lineage and tissue specific activity

How is ρ-globin expression confined to the primitive lineage? To address this question, we examined the properties of the 456 bp region immediately upstream of the p gene. Plasmids containing this region coupled to the cat gene were introduced into both primary chick embryo fibroblasts and primary erythroid cells. Expression of cat was observed only in the erythroid cells; furthermore, expression in erythroid cells was limited to the primitive lineage when sufficient chick serum was present to give optimal levels of expression. The 456 bp p promoter region thus contains sufficient information to specify both tissue- and lineage-specific expression. We therefore set out to determine the elements in the region that are responsible for this behavior.

As a preliminary step, we studied the effect of sequen-tial 5′ deletions of the promoter. As shown in Fig. 2, removal of the region between −246 bp and −181 bp in the promoter results in a 70% reduction in cat expression in primitive lineage cells; deletion of an additional 83 bp reduces expression nearly to background. Neither the full-length construction nor any of those with truncated promoters is active in definitive cells, arguing against a definitive cell-specific repressor.

Role of GATA and Sp1 sites in transcriptional activity

Our deletion studies identified promoter regions containing binding sites for trans-acting factors. Next, we showed with mobility shift experiments that actual binding occurs in these regions. These experiments demonstrate that Sp1, GATA-1, and BGP1 factors all bind to their defined cog-nate sites in the ρ-upstream region. No other factors were identified under the conditions used in these experiments. In one case, both Sp1 and BGP1 were shown to bind to the same site (BGP1proximal/Sp1distal). A similar dual BGP1/Sp1 binding site has been observed in the promoter of the α subunit of the chicken acetylcholine receptor gene (Piette et al., 1989).

Interestingly, either Sp1 site can interact with Sp1 to form a double band in gel shift experiments. In addition, mem-bers of the doublet vary in mobility depending on whether primitive or definitive cell extracts are used. The doublet bands of the primitive extract have a slightly but repro-ducibly higher mobility than those seen in the definitive extracts. Furthermore, mixing experiments show that the definitive extract pattern dominates the primitive extract pattern (data not shown). These observations suggest that the Sp1 complexes formed with definitive extracts are in some way different from those in the primitive extracts, possibly because of a covalent modification (such as phos-phorylation or glycosylation) or an association with an as yet unidentified co-factor (such as those reported by Dyn-lacht et al. (1991) to be associated with and necessary for the activity of TFIID), and will be the subject of further investigations.

Cluster mutations in the GATA-1, Sp1proximal or TFIID sites in the ρ-upstream element individually result in a pronounced loss of transcriptional activity, indicating that bind-ing of any one of these factors to its site is crucial to ρ-globin gene transcription. A similar requirement for a functional GATA site and a functional CACCC site (which can bind Sp1 as well as CON; Jackson et al., 1989) has been reported for the erythroid-specific promoter of the human porphobilinogen deaminase gene (Frampton et al., 1990).

Lineage-specific variation in the abundance of Sp1 and GATA-1

Genes regulated by ubiquitous transcription factors may nonetheless be affected by changes in the abundance of those factors. We find that the abundance of several duplex DNA-binding factors differs significantly in primitive and definitive lineage cell nuclei. In particular, the amounts of Sp1 and GATA-1 per nucleus decrease by a factor of about 10 as definitive red cells replace primitive cells in the circulation. Similar decreases in the amounts of GATA-1 in chicken red cells have been reported elsewhere (Nicolas et al., 1991; Perkins et al., 1989). Since the nuclear volumes are similar in the two lineages, their intranuclear factor concentrations must also differ 10-fold. This difference is not due to uniform losses of all binding activities during nuclear extraction, as the amount of PAL factor detected remains constant between days 4 and 12.

GATA-1 mRNA also undergoes a decrease in cellular abundance as development proceeds (Fig. 7B). The lower abundance of GATA-1 protein in primitive lineage cells presumably reflects these lower levels of mRNA. It is equally possible to argue that the lower levels of GATA-1 protein result in lower levels of mRNA, since the GATA-1 promoter contains active GATA-1 binding sites (Hannon et al., 1991). This autoregulation may be an important component of the switching mechanism.

We note that Whitelaw et al. (1990) report an increase in GATA-1 mRNA at later stages of erythroid development in the mouse. This may reflect a real difference between chicken and mouse erythropoiesis.

Earlier reports from this laboratory (Jackson et al., 1989) examined the change in the abundance of PAL protein during development; the relative amounts of protein present at different stages were calculated by normalizing to the abundance of Sp1. By this measure PAL abundance appeared to increase 10-fold between days 5 and 11 of embryogenesis. Our present results show that this change in the PAL/Sp1 ratio is attributable to a decrease in Sp1 concentration; in the 4-12 day period, PAL concentration in fact remains constant as noted above. Previous measurements of other factors, such as GATA-1 and BGP1, were also normalized to Sp1. These factors are now seen also to undergo a decrease in abundance per cell between days 4 and 12 of development. Nonetheless, there is clearly a major increase in PAL concentration that occurs subsequent to day 12, the developmental stages which were the focus of the earlier work.

Other ubiquitous transcription factors have been shown to vary in their abundance in a tissue-specific manner (Mitchell et al., 1991; Saffer et al., 1991; Snape et al., 1990, 1991), and these differences have been implicated in developmental regulation. For example, it has been proposed that the protamine-2 gene in mouse spermatocytes is regulated in part by changes in the concentration of Sp1 during spermatogenesis (Saffer et al. 1991; Bunick et al., 1990). It has also been suggested that high cellular concentrations of Sp1 play a role in early hematopoiesis in mouse embryos (Saffer et al., 1991). Similarly, it has been shown that AP-2 varies in abundance in a tissue-specific manner in both Xenopus and mouse development, and that high abundance of that factor in certain cells may drive the expression of several genes, including the keratin gene in Xenopus epidermal tissue (Mitchell et al., 1991; Snape et al. 1990, 1991). Thus, there is significant evidence that seemingly ubiquitous transcription factors vary in abundance in a developmentally specific manner, and that this variation may play a role in the stage-specific regulation of gene activity.

Our experiments do not rule out the possibility that other sites proximal or distal to the π promoter also contribute to expression of that gene. There may be other elements within the promoter region that we examined that serve to modulate expression in a lineage-specific manner. Our results show, however, that the Sp1 and GATA-1 binding sites are critical to the lineage-specific expression revealed by our transient expression studies. Furthermore, we show that the reduction in abundance of Sp1 and GATA-1 that accompanies progression from the primitive to the definitive linage would be sufficient to have a major effect on occupancy of these sites in vivo. Thus, sufficient information for stage-specific expression would be contained in principle in the interactions with Sp1 and GATA-1 sites of the upstream region that we have studied.

A model mechanism for the restriction of ρ-globin gene activity to the primitive lineage

While the cellular basis of the hemoglobin switch in chickens appears to be explained by the existence of two separate erythroid lineages, the molecular mechanisms under-lying the expression of specific globin genes in each cell type remains unclear. We propose that the differences in the abundances of Sp1 and GATA-1 between primitive and definitive cells play a role in the cell type restriction of globin gene expression in chicken and, in particular, may explain the stage-specific response of the p promoter described in this paper (Fig. 8A).

Fig. 8.

(A) Schematic of proposed mechanism for the restriction of ρ-globin gene activity to the primitive erythroid lineage by transcription factor abundance. Note that the association of Sp1 and GATA-1 depicted in the figure is not meant to imply an experimental observation of such an interaction, but instead represents simply one possible manner of interaction between the two factors. (B) Plot of fractional occupancy of ρ-globin Sp1 and GATA sites as a function of relative factor concentration, calculated from the equations in the appendix.

Fig. 8.

(A) Schematic of proposed mechanism for the restriction of ρ-globin gene activity to the primitive erythroid lineage by transcription factor abundance. Note that the association of Sp1 and GATA-1 depicted in the figure is not meant to imply an experimental observation of such an interaction, but instead represents simply one possible manner of interaction between the two factors. (B) Plot of fractional occupancy of ρ-globin Sp1 and GATA sites as a function of relative factor concentration, calculated from the equations in the appendix.

We have shown that both the Sp1proximal and GATA-1 sites of the ρ-globin gene promoter are necessary for the activation of transcription. It seems reasonable to speculate that these sites would be occupied in the factor-rich prim-itive cell. As shown in Fig. 8B (and in the Appendix), it is possible to develop a quantitative description of how such a system might work. Under suitable conditions of binding constants and factor abundance, a 10-fold decrease in Sp1 and GATA-1 concentrations could alter the fraction of pro-moter sites occupied by both factors by nearly two orders of magnitude. We suggest that the level of expression of p is determined by the level of occupancy of the Sp1 and GATA-1 sites by these factors; the calculated differences would be quite sufficient to explain our results, and to account for stage-specific regulation of p expression.

Implicit in this model is the assumption that the observed effect of the AGATAAG and GGGGTGGGG cis-acting elements in raising levels of π expression is due to the binding of GATA-1 and Sp1. The gel shift experiments show that these are the greatly predominant binding components in vitro, and transactivation studies in fibroblasts have shown that expression from promoters carrying GATA-1 binding sites are stimulated by GATA-1 expression (Evans and Felsenfeld, 1991). Nonetheless, we cannot rule out the possibility that the effective trans-acting species in vivo are in fact minor components unrelated to GATA-1 and Sp1, or that only suitably modified versions of these factors are capable of transactivation, or that the rate-limiting step in control of transcription is the binding of undetected cofactors. We believe, however, that the model we propose gives a simple explanation of the observations.

Further experiments could be designed to assess the effect of altered site occupancy by manipulating the levels of GATA-1 and Sp1 within erythroid cells, or by altering the binding site affinities for these factors. Interestingly, nature has already hinted at the result of such experiments. The human A γ and G γ globin genes, which also exhibit cellular restriction (Peschle et al., 1984, 1985), both exhibit abnormal expression in patients with non-deletional hereditary persistence of fetal hemoglobin (HPFH) diseases. In some of these patients, naturally occurring promoter mutations in Sp1 and GATA-1 binding sites have increased the affinities of these sites for their cognate factors (Mantovani et al., 1988; Ronchi et al., 1989; Sykes and Kaufman, 1990). Transient transfection experiments have shown that these mutants result in higher levels of promoter activity, and this higher activity may result in the observed loss in cellular restriction (Martin et al., 1989; Nicolis et al., 1989; Ronchi et al., 1989). Such mutations suggest that simple alterations in the affinities of either Sp1 or GATA sites for their cognate factors can result in the loss of lineage specificity.

It is instructive to compare ρ-globin gene expression with the definitive lineage-specific expression of the β A-globin gene. The β A-globin gene promoter has Sp1 sites, and the β/ ∈ enhancer downstream of the gene has a pair of GATA-1 sites. In the absence of other mechanisms, and assuming that our model is correct, the higher abundance of Sp1 and GATA-1 in primitive than in definitive lineage cells might be expected to result in a higher level of expression of β A-globin in primitive lineage cells, particularly given that the Sp1 and GATA-1 sites in the neighborhood of A have twice the affinity of those near p (data not shown and Letovsky and Dynan, 1989). The β A gene is presumably prevented from being expressed in primitive lineage cells by its absolute requirement for another factor, NF-E4, that is found only in definitive lineage cells (Choi and Engel, 1988; Gallarda et al., 1989) and may cross-bind the Sp1 sites in the A promoter (Evans et al., 1990). Stage-specific expression of the β A gene thus depends upon a stage-specific regulatory factor. In contrast, expression of the p promoter, as well as the α π globin gene (J. Knezetic and G. Felsenfeld, unpublished), may be controlled by erythroid lineage-specific differences in the concentration of ubiquitous factors. There is no reason a priori why individual members of the globin families should make use of identical regulatory mechanisms. As information becomes available about the other α- and β -globins, it will be interesting to see how many depend for stage specificity on the action of special factors, and how many of these genes make use of concentration-dependent mechanisms like those that we have proposed here for the ρ-globin gene.

Appendix

A. Effects of concentration on binding of multiple trans-acting factors

We wish to estimate the effects of concentration on the binding of Sp1 and GATA-1 to their sites in the p promoter. The data presented in this paper suggest that both sites must be occupied for effective promoter function, and that there is an approximately 10-fold difference in intranu-clear concentration of each factor between 4-day and 12-day embryonic cells. We therefore must calculate the fraction of sites occupied simultaneously by both factors as a function of concentration. We first analyze the equations for equilibrium binding to demonstrate the overall binding behavior, and then use our binding data to estimate the occupancy of promoter sites in vivo as a function of developmental stage.

B. Calculation of binding behavior

We assume that the promoter DNA can be doubly complexed to Sp1 and GATA-1 (DSG), singly complexed to either one (DS or DG) or protein-free (D). If the binding of S is independent of G,

formula

It is also essential to take into account the binding to non-specific sites, with concentration N, and binding constant KN, assumed for simplicity of illustration to be the same for both factors. The non-specific binding constant

formula

Because non-specific DNA-binding sites are in vast excess, the initial concentration N0 is unaffected by binding of S and G, and this DNA acts as a buffer for the free concentrations of these factors. The activity of the promoter depends on the concentration (DSG) of doubly occupied sites. If the ratio of open plus singly occupied sites to doubly occupied sites is f,

formula

where S0 and G0 are the initial (total) concentrations of S and G. For the sake of illustration and simplicity, we assume that the occupancy by S and G of their DNA sites is about the same, i.e. K1S0= K2G0. This simplification does not affect the general conclusions; in the next section, more accurate, individual estimates are made of K1, K2, S0 and G0. The equation for f can be recast to give the frac-tion Q of all sites that are doubly occupied, as a function of the total concentration of factor S0 (= G0):

formula

where r = K1S0/KNN0= LS0 and L is a constant. The equation allows us to calculate this fraction as a function of the change in concentration of factors. The result is shown in Fig. 8B, where both S0 and G0 are allowed to vary coordi-nately over a 20-fold range. For an approximately 10-fold decrease in concentration (between 4 and 12 days), the effective occupancy can change from 25% to 0.8%.

C. Occupancy of ρ promoter sites within the nucleus

The quantitative gel mobility shift data presented here, in conjunction with the above analysis, make it possible to estimate the fractional occupancy of the Sp1 and GATA sites within the nucleus. We have previously measured the specific affinity constant of chicken Sp1 (about 2.5 × 109, cited in Clark et al., 1990); it is similar to that measured for human Sp1 (Letovsky and Dynan, 1989). We have also estimated the affinity constant of GATA-1 for a strong binding site as ∼109 (G. Felsenfeld and C. Trainor, unpublished data). Typically, non-specific binding constants are about 104-fold smaller than specific constants. We can use this information to estimate the number of regulatory factor molecules present in nuclei: in 5-day cells, 15000 copies of Sp1 and 4500 copies of GATA-1; in 9-day cells, 1300 of Sp1 and 440 of GATA-1. To calculate fractional occupancy of sites in the nucleus (see calculation above), we assume that about 10% of the DNA in chromatin is accessible for non-specific binding within the nucleus. We then calculate that Q, the fraction of p promoter sites occupied simultaneously by Sp1 and GATA-1 is 0.079 in 5-day cells, and 0.0013 in 9-day cells, a 60-fold decrease, consistent with the general calculation presented above. The point of the general calculation above is to show that this behavior is to be expected over a considerable range of input parameters, and is not highly sensitive to the actual values of binding constants, factor abundances or nuclear DNA concentrations.

We thank Drs A. Wolffe, J. Grasso, E. Bresnick, J. Hesse, D. Clark, C. Trainor and all of our other collegues at the Laboratory of Molecular Biology for their insightful discussions and critical reviews of this work. Special thanks go to Drs J. Hesse and M. Lieber for their advice on transfection protocols and timely encouragement. Dr. Knezetic’s sharing of plasmids and unpublished observations are also gratefully acknowledged, as is the gift of anti-Sp1 sera from Drs S. J. Clark and R. Tjian.

Beaupain
,
D.
(
1985
).
Line-restricted hemoglobin synthesis in chick embryonic erythrocytes
.
Cell Diff
.
16
,
101
107
.
Beaupain
,
D.
,
Martin
,
C.
and
Dieterien-Lievre
,
F.
(
1979
).
Are developmental hemoglobin changes related to the origin of stem cells and site of erythropoiesis?
Blood
53
,
212
225
.
Blackman
,
M. A.
,
Tigges
,
M. A.
,
Minie
,
M. E.
and
Koshland
,
M. E.
(
1986
).
A model system for peptide hormone action in differentiation: interleukin-2 induces a B lymphoma to transcribe the J chain gene
.
Cell
47
,
609
617
.
Bruns
,
G. A. P.
and
Ingram
,
V. M.
, (
1973
).
The erythroid cells and haemoglobins of the chick embryo
.
Philos. Trans. R. Soc. Lond. Ser. B. Biol. Soc
.
266
,
225
305
.
Bunick
,
D.
,
Johnson
,
P. A.
,
Johnson
,
T. R.
and
Hecht
,
N. B.
(
1990
).
Transcription of the testis-specific mouse protamine 2 gene in a homologous in vitro transcription system
.
Proc. Natl. Acad. Sci. U.S.A
.
87
,
891
895
.
Buratowski
,
S.
,
Hahn
,
S.
,
Sharp
,
P. A.
and
Guarente
,
L.
(
1988
).
Function of a yeast TATA element-binding protein in a mammalian transcription system
.
Nature
334
,
37
42
.
Chirgwin
,
J. M.
,
Przybyla
,
A. E.
,
MacDonald
,
K. J.
and
Rutter
,
W. J.
(
1979
).
Isolation of biologically active RNA from sources enriched in ribonuclease
.
Biochemistry
18
,
5294
5299
.
Choi
,
O.-R. B.
and
Engel
,
J. D.
(
1988
).
Developmental regulation of - globin gene switching
.
Cell
55
,
17
26
Clark
,
S. P.
,
Lewis
,
C. D.
and
Felsenfeld
,
G.
(
1990
).
Properties of BGP1, a poly(dG)-binding protein from chicken erythrocytes
.
Nuc. Acids Res
.
18
,
5119
5126
.
Coll
,
J.
and
Ingram
,
V. M.
(
1978
).
Stimulation of heme accumulation and erythroid colony formation in cultures of chick bone marrow cells by chicken plasma
.
J. Cell Biol
.
76
,
184
190
.
Dieterlen-Lievre
,
F.
(
1975
).
On the origin of haemopoietic stem cells in the avian embryo: an experimental approach
.
J. Embryol. Exp. Morph
.
33
,
607
619
.
Dieterlen-Lievre
,
F.
(
1988
).
Birds
.
In Vertebrate Blood Cells
, (ed.
A. F.
Rowley
and
N. A.
Ratcliffe
) pp.
257
336
.
Cambridge University Press
.
Dodgson
,
J. B.
and
Engel
,
J. D.
(
1983
).
The nucleotide sequence of the adult chicken α-globin genes
.
J. Biol Chem
.
258
,
4623
4629
.
Dodgson
,
J. B.
,
Stadt
,
S. J.
,
Choi
,
O-R.
,
Dolan
,
M.
,
Fischer
,
H. D.
and
Engel
,
J. D.
(
1983
).
The nucleotide sequence of the embryonic chicken-type globin genes
.
J. Biol Chem
.
258
,
12685
12692
.
Dolan
,
M.
,
Sugarman
,
B. J.
,
Dodgson
,
J. B.
and
Engel
,
J. D.
(
1981
).
Chromosomal arrangement of the chicken -type globin genes
.
Cell
24
,
669
677
.
Dynlacht
,
B. D.
,
Hoey
,
T.
and
Tjian
,
R.
(
1991
).
Isolation of coactivators associated with the TATA-binding protein that mediate transcriptional activation
.
Cell
66
,
563
576
.
Emerson
,
B.
,
Nickol
,
J.
and
Fong
,
T. C.
(
1989
).
Erythroid-specific activation and derepression of the chick -globin promoter in vitro
.
Cell
57
,
1189
1200
.
Evans
,
T.
, and
Felsenfeld
,
G.
(
1991
).
trans-Activation of a globin promoter in nonerythroid cells
.
Mol. Cell Biol
.
11
,
843
853
.
Evans
,
T.
,
Felsenfeld
,
G.
and
Reitman
,
M.
(
1990
).
Control of globin gene transcription
.
Ann. Rev. Cell Biol
.
6
,
95
124
.
Evans
,
T.
,
Reitman
,
M.
and
Felsenfeld
,
G.
(
1988
).
An erythrocyte-specific DNA-binding factor recognizes a regulatory sequence common to all chicken globin genes
.
Proc. Natl. Acad. Sci. U.S.A
.
85
,
5976
5980
.
Frampton
,
J.
,
Walker
,
M.
,
Plumb
,
M.
and
Harrison
,
P. R.
(
1990
).
Synergy between the NF-E1 erythroid-specific transcription factor and the CACCC factor in the erythroid-specific promoter of the human porphobilinogen deaminase gene
.
Mol. Cell Biol
.
10
,
3838
3842
.
Fucci
,
L.
,
Cirotto
,
C.
,
Tomei
,
L.
and
Geraci
,
G.
(
1983
).
Synthesis of globin chains in the erythopoietic sites of the early chick embryo
.
J. Embryol. Exp. Morph
.
77
,
153
165
.
Fucci
,
L.
,
Vitale
,
E.
,
Cirotto
,
C.
and
Geraci
,
G.
(
1987
).
Evidences that hemoglobin switch in the chick embryo depends on erythroid cell line substitution
.
Cell Differentiation
.
20
,
55
63
.
Gallarda
,
J. L.
,
Foley
,
K. P.
,
Zhuoying
,
Y.
and
Engel
,
J. D.
(
1989
).
The - globin stage selector element factor is erythroid-specific promoter/enhancer binding protein NF-E4
.
Genes Dev
.
3
,
1845
1859
.
Groudine
,
M.
and
Weintraub
,
H.
(
1982
).
Propagation of globin DNase I-hypersensitive sites in absence of factors required for induction: a possible mechanism for determination
.
Cell
30
,
131
139
.
Hamburger
,
V.
and
Hamilton
,
H. L.
(
1951
).
A series of normal stages in the development of the chick embryo
.
J. Morph
.
88
,
49
92
.
Hannon
,
R.
,
Evans
,
T.
,
Felsenfeld
,
G.
and
Gould
,
H.
(
1991
).
Structure and promoter activity of the gene for the erythroid transcription factor GATA-1
.
Proc. Natl. Acad. Sci. U.S.A
.
88
,
3004
3008
.
Hesse
,
J. E.
,
Nickol
,
J. M.
,
Lieber
,
M. R.
and
Felsenfeld
,
G.
(
1986
).
Regulated gene expression in transfected primary chicken erythrocytes
.
Proc. Natl. Acad. Sci. U.S.A
.
83
,
4312
4316
.
Ingram
,
V. M.
(
1972
).
Embryonic red blood cell formation
.
Nature
235
,
338
339
.
Jackson
,
D. P.
,
Evans
,
T.
,
Nickol
,
J. M.
and
Felsenfeld
,
G.
(
1989
).
Developmental modulation of protein binding to -globin gene regulatory sites within chicken erythrocyte nuclei
.
Genes Dev
.
3
,
1860
1873
.
Kadonaga
,
J.
(
1990
).
JK Lab Techniques
. 3rd Edition.
Kemper
,
B.
,
Jackson
,
P. D.
and
Felsenfeld
,
G. F.
(
1987
).
Protein-binding sites within the 5′ DNAase I-hypersensitive region of the chicken αD-globin gene
.
Mol. Cell Biol
.
7
,
2059
2069
.
Knezetic
,
J. A.
and
Felsenfeld
,
G.
(
1989
).
Identification and characterization of a chicken α-globin enhancer
.
Mol. Cell Biol
.
9
,
893
901
.
Landes
,
G. M.
,
Villeponteau
,
B.
,
Pribyl
,
T. M.
and
Martinson
,
H. G.
(
1982
).
Hemoglobin switching in chickens: is the switch initiated post transcriptionally?
J. Biol. Chem
.
257
,
11008
11014
.
Lewis
,
C. D.
,
Clark
,
S. P.
,
Felsenfeld
,
G.
and
Gould
,
H.
(
1988
).
An erythrocyte-specific protein that binds to the poly (dG) region of the chicken -globin gene promoter
.
Genes Dev
.
2
,
863
873
.
Letovsky
,
J.
and
Dynan
,
W. S.
(
1989
).
Measurement of the binding of transcription factor Sp1 to a single GC box recognition sequence
.
Nuc. Acids Res
.
17
,
2639
2653
.
Li
,
R.
,
Knight
,
J. D.
,
Jackson
,
S. P.
,
Tjian
,
R.
and
Botchan
,
M.
(
1991
).
Direct interaction between Sp1 and the BPV enhancer E2 protein mediates synergistic activation of transcription
.
Cell
65
,
493
505
.
Lieber
,
M. R.
,
Hesse
,
J. E.
,
Nickol
,
J. M.
and
Felsenfeld
,
G.
(
1987
).
The mechanism of osmotic transfection of avian embryonic erythrocytes: analysis of a system for studying developmental gene expression
.
J. Cell Biol
.
105
,
1055
1065
.
Lois
,
R.
and
Martinson
,
H. G.
(
1989
).
Chicken globin gene transcription is cell lineage specific during the time of the switch
.
Biochemistry
28
,
2281
2287
.
Lucas
,
A. M.
and
Jamroz
,
C.
(
1961
).
Atlas of avian hemotology. Agricultural Monograph 25, USDA
.
Mahoney
,
K. A.
,
Hyer
,
B. J.
and
Chan
,
L.
(
1977
).
Separation of primitive and definitive erythroid cells of the chick embryo
.
Dev. Biol
.
56
,
412
416
.
Maniatis
,
T.
,
Fritsch
,
E. F.
and
Sambrook
,
J.
(
1983
).
Molecular Cloning: a Laboratory Manual
.
New York
:
Cold Spring Harbor Laboratory
.
Mantovani
,
R.
,
Malgaretti
,
N.
,
Nicolis
,
S.
,
Rochi
,
A.
,
Giglioni
,
B.
and
Ottolenghi
,
S.
(
1988
).
The effects of HPFH mutations in the human “f-globin promoter on binding of ubiquitous and erythroid specific nuclear factors
.
Nuc. Acids Res
.
16
,
7783
7797
.
Martin
,
D. I.
,
Tsai
,
S-F.
and
Orkin
,
S. H.
(
1989
).
Increased “f-globin expression in a non-deletion HPFH mediated by an erythroid-specific DNA-binding factor
.
Nature
338
,
435
438
.
Minie
,
M. E.
(
1986
).
The regulation of J chain gene transcription: Role of chromatin structural changes and protein-DNA interactions
.
Ph.D. Thesis, U. C. Berkeley
.
Mitchel
,
P. J.
,
Timmons
,
P. M.
,
Herbert
,
J. M.
,
Rigby
,
P. W. J.
and
Tjian
,
R.
(
1991
).
Transcription factor AP-2 is expressed in neural crest cell lineages during mouse embryogenesis
.
Genes Dev
.
5
,
105
119
.
Nickol
,
J. M.
and
Felsenfeld
,
G.
(
1988
).
Bidirectional control of the chicken - and E-globin genes by a shared enhancer
.
Proc. Natl. Acad. Sci. U.S.A
.
85
,
2548
2552
.
Nicolas
,
R. H.
,
Partington
,
G.
,
Major
,
G. N.
,
Smith
,
B.
,
Carne
,
A. F.
,
Huskisson
,
N.
and
Goodwin
,
G.
(
1991
).
Induction of differentiation of avian erythroblastosis virus-transformed erythroblasts by the protein kinase inhibitor H7: analysis of the transcription factor EF1
.
Cell Growth Differ
.
2
,
129
135
.
Nicolis
,
S.
,
Antonella
,
R.
,
Malgaretti
,
N.
,
Mantovani
,
R.
,
Gilglioni
,
B.
and
Ottolenghi
,
S.
(
1989
).
Increased erythroid-specific expression of a mutated HPFH “f-globin promoter requires the erythroid factor NFE-1
.
Nuc. Acids Res
.
17
,
5509
5516
.
Nikinmaa
,
M.
(
1990
).
Vertebrate Red Blood Cells: Adaptations of Function to Respiratory Requirements
.
New York
:
Springer-Verlag
.
Pain
,
B.
,
Woods
,
C. M.
,
Saez
,
J.
,
Flickinger
,
T.
,
Raines
,
M.
,
Peyrol
,
S.
,
Moscovici
,
C.
,
Mocovici
,
M. G.
,
Kung
,
H.-J.
,
Jurdic
,
P.
,
Lazarides
,
E.
and
Samarut
,
J.
(
1991
).
EGF-R as a hemopoietic growth factor receptor: the c-erbB product is present in chicken erythrocytic progenitors and controls their self-renewal
.
Cell
65
,
37
46
.
Perkins
,
N. D.
,
Nicolas
,
R. H.
,
Plumb
,
M. A.
and
Goodwin
,
G. H.
(
1989
).
The purification of an erythroid protein which binds to enhancer and promoter elements of haemoglobin genes
.
Nuc. Acids Res
.
17
,
1299
1314
.
Peschle
,
C.
,
Migliaccio
,
A. R.
,
Migliaccio
,
G.
,
Pentrini
,
M.
,
Calandrini
,
M.
,
Russo
,
G.
,
Mastroberardino
,
G.
,
Presta
,
M.
,
Gianni
,
A. M.
,
Comi
,
P.
,
Giglioni
,
B.
and
Ottolenghi
,
S.
(
1984
).
Embryonic>fetal Hb switch in humans: studies on erythroid bursts generated by embryonic progenitors from yolk sac and liver
.
Proc. Natl. Acad. Sci. U.S.A
.
81
,
2416
2420
.
Peschle
,
C.
,
Mavillio
,
F.
,
Care
,
A.
,
Migliaccio
,
G.
,
Migliaccio
,
A. R.
,
Salvo
,
G.
,
Samoggia
,
P.
,
Petti
,
S.
,
Guerriero
,
R.
,
Marinucci
,
M.
,
Lazzaro
,
D.
,
Russo
,
G.
and
Mastroberardino
,
G.
(
1985
).
Haemoglobin switching in human embryos: asychrony of ζ> α and ε>γ-globin switches in primitive and definitive erythropoietic lineage
.
Nature
313
,
235
238
.
Piette
,
J.
,
Klarsfeld
,
A.
and
Changeux
,
J.-P.
(
1989
).
Interaction of nuclear factors with the upstream region of the α-subunit gene of chicken muscle acetylcholine receptor: variations with muscle differentiation and denervation
.
EMBO J
.
8
,
687
694
.
Reitman
,
M.
and
Felsenfeld
,
G.
(
1988
).
Mutational analysis of the chicken β-globin enhancer reveals two positive-acting domains
.
Proc. Natl. Acad. Sci. U.S.A
.
85
,
6267
6271
.
Reitman
,
M.
and
Felsenfeld
,
G.
(
1990
).
Developmental regulation of topoisomerase II sites and DNase I-hypersensitive sites in the chicken - globin locus
.
Mol. Cell. Biol
.
10
,
2774
2786
.
Reitman
,
M.
,
Lee
,
E.
,
Westphal
,
H.
and
Felsenfeld
,
G.
(
1990
).
Site-independent expression of the chicken βA-globin gene in transgenic mice
.
Nature
348
,
749
752
.
Ronchi
,
A.
,
Nicolis
,
S.
,
Santoro
,
C.
and
Ottolenghi
,
S.
(
1989
).
Increased Sp1 binding mediates erythroid-specific overexpression of a mutated (HPFH) γ-globulin promoter
.
Nuc. Acids Res
.
17
,
10231
10241
.
Saffer
,
J. D.
,
Jackson
,
S. P.
and
Annarella
,
M. B.
(
1991
).
Developmental expression of Sp1 in the mouse
.
Mol. Cell Biol
.
11
,
2189
2199
.
Samarut
,
J.
(
1978
).
Isolation of an erthropoietic stimulating factor from the serum of anemic chicks
.
Exp. Cell Res
.
115
,
123
126
.
Samarut
,
J.
(
1979
).
Erythroietin-like factor production by young chick blastoderms
.
Dev Biol
.
70
,
278
282
.
Samarut
,
J.
and
Nigon
,
V.
(
1976
).
In vitro development of chicken erythropoietin-sensitive cells
.
Exp. Cell Res
.
100
,
245
248
.
Schalekamp
,
M.
and
van Goor
,
D.
(
1984
).
Haemoglobin and globin synthesis in the isolated primitive and definitive erythroid cells of chicken embryos. Evidence for a non-clonal mechanism at the haemoglobin switch
.
J. Embryol. Exp. Morph
.
84
,
125
148
.
Shimizu
,
K.
(
1976
).
Identification of hemoglobin types contained in single chicken erythrocytes by fluorescent antibody technique
.
Devl Biol
.
48
,
317
326
.
Singh
,
H.
,
Sen
,
R.
,
Baltimore
,
D.
and
Sharp
,
P. A.
(
1986
).
A nuclear factor that binds a conserved sequence motif in transcriptional control elements of immunoglobulin genes
.
Nature
319
,
154
158
.
Snape
,
A. M.
,
Jonas
,
E. A.
and
Sargent
,
T. D.
(
1990
).
KTF-1, a transcriptional activator of Xenopus embryonic keratin expression
.
Development
109
,
157
165
.
Snape
,
A. M.
,
Winning
,
R. S.
and
Sargent
,
T. D.
(
1991
).
Transcription factor AP-2 is tissue-specific in Xenopus and is closely related or identical to keratin transcription factor 1 (KTF-1)
.
Development
113
,
283
293
.
Stalder
,
W. E.
,
Larson
,
A.
,
Engle
,
J. D.
,
Dolan
,
M.
,
Groudine
,
M.
and
Weintraub
,
H.
(
1980
).
Tissue-specific DNA cleavage in the globin chomatin domain introduced by DNase I
.
Cell
20
,
451
460
.
Sykes
,
K.
and
Kaufman
,
R.
(
1990
).
A naturally occurring gamma globin mutation enhances Sp1 binding activity
.
Mol. Cell Biol
.
10
,
95
102
.
Townes
,
T. M.
,
Lingrel
,
J. B.
,
Chen
,
H. Y.
,
Brinster
,
R. L.
and
Palmiter
,
R. D.
(
1985
).
Erythroid-specific expression of human -globin genes in transgenic mice
.
EMBO J
.
4
,
1715
1723
.
Whitelaw
,
E.
,
Tsai
,
S-F.
,
Hogben
,
P.
and
Orkin
,
S.
(
1990
).
Regulated expression of globin chains and the erythroid transcription factor GATA-1 during erythropoiesis in the developing mouse
.
Mol. Cell Biol
.
10
,
6596
6606
.
Yamamoto
,
M.
,
Ko
,
L. J.
,
Leonard
,
M. W.
,
Bueg
,
H.
,
Orkin
,
S. H.
and
Engle
,
D. J.
(
1990
).
Activity and tissue-specific expression of the transcription factor NF-E1 multigene family
.
Genes Dev
.
4
,
1650
1662
.