CACCC and GATA-1 sequences make the constitutively expressed
[alpha]-globin gene erythroid-responsive in mouse erythroleukemia cells
CACCC and GATA-1 sequences make the constitutively expressed [alpha]-globin gene erythroid-responsive in mouse erythroleukemia cells
Sicong
Ren
,
Jihong
Li
and
George F.
Atweh
*
Department of Medicine and Brookdale Center for Molecular Biology, Mount Sinai
School of Medicine,
New York
, NY 10029,
USA
Received August 31, 1995;
Revised and Accepted November 30, 1995
ABSTRACT
Although the human
[alpha]
-globin and
[beta]
-globin genes are co-regulated in adult life, they achieve the same end by very different
mechanisms. For example, a transfected
[beta]
-globin gene is expressed in an inducible manner in mouse erythroleukemia
(MEL) cells while a transfected
[alpha]
-globin gene is constitutively expressed at a high level in induced and
uninduced MEL cells. Interestingly, when the
[alpha]
-globin gene is transferred into MEL cells as part of human chromosome 16,
it is appropriately expressed in an inducible manner. We explored the basis for
the lack of erythroid-responsiveness of the proximal regulatory elements of the human
[alpha]
-globin gene. Since the
[alpha]
-globin gene is the only functional human globin gene that lacks CACCC and
GATA-1 motifs, we asked whether their addition to the
[alpha]
-globin promoter would make the gene erythroid-responsive in MEL cells. The addition of each of these binding sites
to the
[alpha]
-globin promoter separately did not result in inducibility in MEL cells.
However, when both sites were added together, the
[alpha]
-globin gene became inducible in MEL cells. This suggests that erythroid non-responsiveness of the
[alpha]
-globin gene results from the lack of erythroid binding sites and is not
necessarily a function of the constitutively active, GC rich promoter.
INTRODUCTION
The different genes of the human [alpha]- and [beta]-globin gene clusters are coordinately regulated to
produce embryonic, fetal and adult hemoglobins at specific developmental stages
(
1
,
2
). In the adult stage, the [alpha]-globin and [beta]-globin genes express equivalent levels of globin chains
to produce Hb A, the major adult hemoglobin. The molecular basis of the
coordinate regulation of these two genes has been the subject of intense
scrutiny. These studies have identified a number of important and unexpected
differences between the [alpha]- and [beta]-globin genes: (i) the [alpha]- and [beta]-globin genes are present on
different chromosomes. The [alpha]-globin gene is present on chromosome 16 (
3
) while the [beta]-globin gene is present on chromosome 11 (
4
); (ii) there are four copies of the [alpha]-globin gene in the diploid human genome and only two copies of the [beta]-globin gene (
5
,
6
); (iii) whereas the general intron-exon structure of the [alpha]-globin and [beta]-globin genes is similar, there is very little
homology between the proximal regulatory elements of the two genes; (iv) the [alpha]-globin gene is located in a GC rich domain while the [beta]-globin gene is present in an AT rich domain (
7
); (v) the [alpha]-globin promoter is constitutively hypomethylated in all cell types
(
8
,
9
) while the [beta]-globin promoter is heavily methylated in non-expressing cells (
10
-
12
); (vi) the [beta]-like globin gene cluster is in an open chromatin domain only in erythroid cells while the [alpha]-like globin gene cluster is located within a larger
chromatin domain that is open in all cell types (
13
). (vii) The [beta]-globin gene is not expressed efficiently in heterologous cells
unless it is linked to a viral enhancer (
14
-
16
). In contrast, the [alpha]-globin gene is expressed constitutively in all cell types
regardless of the presence of a linked viral enhancer (
14
,
15
,
17
); (viii) a stably integrated human [beta]-globin gene is expressed in an inducible manner in mouse
erythroleukemia (MEL) cells (
18
,
19
) while the [alpha]-globin gene is highly expressed in a constitutive and non-inducible manner (
18
); (ix) the human [beta]-globin gene with its immediate flanking sequences is expressed in a
tissue and developmental stage-specific manner when inserted in the genome of a transgenic mouse (
20
,
21
). In contrast, the [alpha]-globin gene with its immediate flanking sequences is not expressed in any
tissue in transgenic mice (
22
-
24
); (x) both the [alpha]-globin and [beta]-globin genes are under the control of distant major
regulatory elements known as locus control regions (LCRs). The [beta]LCR consists of four hypersensitive sites (HS1-HS4) (
25
-
27
) that have different activities at different developmental stages (
28
). In contrast, all the activity of the [alpha]LCR appears to be localized to a single discreet hypersensitive site
known as HS-40 (
29
,
30
).
We had previously explored the molecular basis for the differences in the
behavior of the [alpha]- and [beta]-globin genes in transient expression systems. We showed
that [alpha]-globin promoter sequences upstream of the cap site are not
sufficient to account for the constitutive activity of the [alpha]-globin gene in erythroid and non-erythroid cells (
31
). We demonstrated that sequences immediately downstream from the cap site are essential for the full enhancer-independent activity of the [alpha]-globin gene promoter (
31
). We also showed that an extended [alpha]-globin promoter that includes sequences from the structural gene is
more responsive to the enhancing activity of the [alpha]LCR than a promoter element that is limited to 5' flanking sequences (
32
). James-Pederson
et al.
have recently made similar observations during their study of the regulation of
the rabbit [alpha]-globin gene (
33
). These studies of the human and rabbit [alpha]-globin genes highlight another feature that the [alpha]-globin promoter shares with promoters of many
housekeeping genes, i.e. extension of the promoter element into the structural sequences of the gene (
34
-
36
).
A number of erythroid-specific transcription factors play important roles in mediating the
activity of the proximal and distal regulatory elements of the different globin
genes. The binding sites for two of these erythroid transcription factors, GATA-1 and EKLF [an erythroid-specific transcription factor that binds CACCC sequences (
37
)] are present in the proximal regulatory element of every human globin gene
except the [alpha]-globin gene. It is not clear whether the lack of erythroid-responsiveness of the human [alpha]-globin gene in MEL cells is dictated by the
`housekeeping' properties of its promoter or by the lack of GATA-1 and CACCC sites in its promoter. We used inducibility in MEL cells as a surrogate
marker for erythroid-responsiveness. We asked whether the insertion of these binding sites in the [alpha]-globin promoter would allow the gene to respond to erythroid
differentiation of MEL cells by increasing its level of expression beyond the
constitutively high basal level of uninduced cells. The experiments described
in this report show that addition of GATA-1 and CACCC sequences, together but not separately, can make the [alpha]-globin promoter inducible in MEL cells. This suggests that the lack of
erythroid-responsiveness of the [alpha]-globin gene is not necessarily a result of its `housekeeping'
promoter which maintains it in an active state in homologous and heterologous
cells but more likely a result of the absence of binding sites for erythroid-specific transcription factors in its promoter. The implications of these
experimental observations will be discussed within the context of our current
understanding of the regulated expression of the [alpha]-globin gene.
MATERIALS AND METHODS
Plasmid constructions
A 1.5 kilobase (kb)
Pst
I fragment that carries the human [alpha]1-globin gene was cloned into the
Pst
I site of the Bluescript II KS
®
plasmid vector (Stratagene, La Jolla, CA). Upstream 5' flanking sequences were deleted from the cloned [alpha]-globin gene by exonuclease III digestion according to instructions provided with the Erase-a-Base
®
kit (Promega, Madison, WI). Initially, the plasmid carrying the [alpha]-globin gene was digested with
Xba
I and the 5' ends were protected by [alpha]-phosphorothioate incorporation. The plasmid DNA was then
digested with
Bam
HI followed by timed exonuclease III digestions. A plasmid library was generated
from these deletion mutants and several unique clones with different sized
deletions were identified. Expression from these truncated promoter constructs
was analyzed in a stable MEL cell expression system as previously described (
38
). Based on the results of this preliminary analysis, a construct which retained
142 bp of 5' flanking sequences ( [alpha]
-142
) was selected for use as a backbone for the generation of all constructs used
in this study (Fig.
1
).
Generation of stable cell lines
Stable cell lines were generated by transfecting MEL cells with each of the five
plasmid expression constructs shown in Figure
1
. DNA (20 [mu]g) from each construct was co-transfected with 2 [mu]g of a plasmid carrying the neomycin resistance gene (
39
) by electroporation into MEL cells as previously described (
32
). The transfected cells were plated for two weeks in selective media which
contained G418 at an active concentration of 400 [mu]g/ml. Pools of transfected cells were established by expanding the G418-resistant cells that survived after 2 weeks in selective medium.
Single cell clones were isolated from these pools by limiting dilution.
Analysis of globin gene expression in transfected cells
Pools and clones of G418-resistant cells were induced to differentiate by exposure to 1.4% dimethyl
sulfoxide (DMSO) for 3 days as previously described (
32
). The success of each induction was confirmed by a marked increase in the level
of expression of the murine [beta]
maj
globin mRNA as previously described (
32
). DNA was isolated from these cells and used in Southern blot analyses to
quantify the number of integrated copies of the transfected [alpha]-globin genes. Ten micrograms of each DNA were digested with
Hin
dIII and probed with a 1.5 kb
Pst
I fragment that carries the human [alpha]-globin gene. Cytoplasmic RNAs were isolated from induced and
uninduced cells as described by Favaloro
et al.
(
40
) and analyzed for human [alpha]-globin gene expression by S1 nuclease assays as previously
described (
41
). The 3' end-labeled probe consisted of a 391 bp
Nco
I/
Hin
dIII fragment of the [alpha]-globin gene of which 97 nt are protected by [alpha]-globin mRNA (
42
). Since the basal level of expression of the human [alpha]-globin gene in MEL cells was very high, we only used 3-5 [mu]g of cytoplasmic RNA in each S1 assay to ensure that all
hybridizations will take place in DNA probe excess. We also included 5 [mu]g of RNA from human K562 cells as a control in every experiment to be able
to make comparisons across assays. Each S1 assay was repeated at least three
times with consistent findings. Autoradiographs of Southern blots and S1 assays
were analyzed by densitometric scanning to determine the number of integrated
copies and the level of expression of the [alpha]-globin gene respectively.
We also analyzed the effect of CACCC and GATA-1 sequences on the activity of the [alpha]-globin gene in fibroblasts. NIH 3T3 cells were transiently
transfected by calcium phosphate-DNA coprecipitation as previously described (
41
). Twenty micrograms of either the [alpha]
-142
or the CCGG[alpha]
-142
construct were co-transfected with a control construct consisting of a human [gamma]-globin gene linked to the SV40 enhancer (
42
). The levels of expression of the [alpha]- and [gamma]-globin genes were assessed by an S1 nuclease
protection assay as previously described (
42
). The relative activity of the two [alpha]-globin gene constructs in fibroblasts was expressed as the ratio of
the densitometric signals corresponding to [alpha]-globin mRNA over [gamma]-globin mRNA.
Graphics
Autoradiographs were imaged using a Silverscan flatbed digital scanner (La Cie, Ltd, Beaverton, OR). Image processing was performed on a
Macintosh Quadra 800 (Apple, Cupertino, CA) using Adobe Photoshop (Adobe,
Mountain, CA) and Aldus Persuasion (Aldus, Seattle, WA) software. Figures were
printed using a XL7700 Digital Continuous Tone Printer (Kodak, Rochester, NY).
RESULTS
We first analyzed the activity of different truncations of the [alpha]-globin promoter by expression assays in stably transfected MEL
cells to identify a minimal promoter that retains full constitutive activity. A
previous study had shown that an [alpha]-globin promoter truncated to position -87 retains full activity in a transient expression assay in
COS cells (
17
). The activity of the [alpha]
-142
globin construct was found to be essentially identical to the non-truncated [alpha]-globin promoter in induced and uninduced MEL cells (data not shown). Thus, we selected this minimal [alpha]
-142
construct as a backbone for the generation of the other expression constructs
shown in Figure
1
. By design, these expression constructs carried two copies of the binding sites
for CACCC and GATA-1 since these sites are commonly duplicated in globin regulatory elements
in vivo
.
Figure
2
shows the results of a quantitative S1 assay of [alpha]-globin expression in one pool and six clones of MEL cells
transfected with the [alpha]
-142
plasmid. Both the pool and the individual clones expressed high levels of human
[alpha]-globin mRNA that did not increase further with induction. In
contrast, the expression of the endogenous mouse [beta]
maj
globin genes increased from 20- to 50-fold with induction (data not shown). We analyzed a total of four
pools and 10 clones of MEL cells transfected with the [alpha]
-142
globin construct. None of the pools or clones analyzed showed inducible
expression of the [alpha]-globin gene in MEL cells.
We then investigated the erythroid-responsiveness of an [alpha]-globin gene with two GATA-1 sites inserted in its promoter (GG[alpha]
-142
). Figure
3
shows an autoradiograph of an S1 assay of [alpha]-globin expression in one pool and six clones of MEL cells
transfected with the GG [alpha]
-142
expression construct. Although the pool and all clones expressed the
transfected [alpha]-globin gene at a relatively high basal level, [alpha]-globin mRNA increased with induction in two of the six
clones analyzed (Fig.
3
, lanes 2 and 5). The level of induction was 1.7-fold in clone 2 and 3-fold in clone 5. We analyzed a total of three pools and 11 clones of
MEL cells transfected with the same construct. Inducible expression was seen in
only two of the 11 clones and in none of the three pools analyzed. No
correlation was noted between the number of integrated copies of the [alpha]-globin gene and inducibility in MEL cells.
An [alpha]-globin gene with two CACCC sites inserted in its promoter (CC[alpha]
-142
) was analyzed in a similar manner. Figure
4
shows an autoradiograph of an S1 assay of [alpha]-globin expression in one pool and six clones of MEL cells transfected with CC[alpha]
-142
. The expression of the human [alpha]-globin gene was not inducible in the pool neither in any of the individual clones. We
analyzed a total of two pools and 17 individual clones of MEL cells transfected
with CC[alpha]
-142
. None of the pools or individual clones showed inducible expression of the
human [alpha]-globin gene.
DISCUSSION
The first experiment described above confirmed the original observations of
Charnay
et al.
that the expression of a transfected human [alpha]-globin gene is not inducible in MEL cells (
18
). When we inserted GATA-1 binding sites in the promoter of the [alpha]-globin gene, inducibility was seen in a small minority (two out of
11) of the MEL clones that were analyzed. The significance of this observation
is not clear. Inducibility resulting from the presence of GATA-1 sites must be very uncommon since [alpha]-globin expression was not inducible in three different pools,
each consisting of 500-1000 individual clones. When we inserted CACCC sites in the promoter of
the [alpha]-globin gene, inducibility was not seen in any of the pools or
clones that were analyzed. The lack of inducibility of the CC[alpha]
-142
and the GG[alpha]
-142
constructs is unlikely to be a result of the high basal levels of [alpha]-globin expression in the uninduced cells since the average level of
expression in the different uninduced cell lines varied by <20%. In contrast, when the CACCC and GATA-1 sites were added together to the [alpha]-globin promoter, the gene became inducible in seven out of
ten clones and four out of six pools. Although these pools were generally of
high complexity consisting of a large number of independent clones, it is not
clear why the [alpha]-globin gene was inducible in some of the pools but not in others.
For this reason, it is always desirable to analyze several pools and an even
larger number of individual clones before making statements about inducibility.
ACKNOWLEDGEMENTS
We would like to thank Dr James Bieker for helpful discussions during the course
of this research. We are grateful for the excellent secretarial assistance of
Migdalia Torrres. This research was supported by a PHS grant to G.F.A. from NIH
(HL42919) and by a GCRC grant (MO1-RR00071) from NIH to the Mount Sinai School of Medicine.
*
To whom correspondence should be addressed at: Division of Hematology, Box 1079,
Mount Sinai School of Medicine, One Gustave L. Levy Place, New York, NY 10029,
USA
