ABSTRACT
We have investigated the transcriptional regulation of the human embryonic
[zeta]
-globin gene promoter. First, we examined the effect that deletion of
sequences 5
'
to
[zeta]
-globin's CCAAT box have on
[zeta]
-promoter activity in erythroid cell lines. Deletions of sequences between -116 and -556 (cap = 0) had little effect while further deletion to -84 reduced
[zeta]
-promoter activity by only 2-3-fold in both transiently and stably transfected erythroid cells. Constructs containing 67, 84 and 556 bp of
[zeta]
-globin 5
'
flanking region linked to a
[beta]
-galactosidase reporter gene (lacZ) and hypersensitive site -40 (HS -40) of the human
[alpha]
-globin gene cluster were then employed for the generation of transgenic mice. LacZ
expression from all constructs, including a 67 bp
[zeta]
-globin promoter, was erythroid-specific and most active between 8.5 and 10.5 days post-fertilisation. By 16.5 days gestation, lacZ expression dropped
40-100-fold. These results suggest that embryonic-specific activation of the human
[zeta]
-globin promoter is conferred by a 67 bp
[zeta]
-promoter fragment containing only a CCAAT and TATA box.
The human haemoglobin molecule is encoded by genes within the [alpha]- and [beta]-globin gene clusters. The [alpha]-cluster consists of three functional genes arranged 5'-[zeta]2-[alpha]2-[alpha]1-3'
at the tip of chromosome 16p while the [beta]-globin gene cluster consists of five functional genes arranged 5'-[epsilon]-
G
[gamma]-
A
[gamma]-[delta]-[beta]-3' on chromosome 11p. The expression of genes within each
cluster is regulated in a tissue- and developmental stage-specific manner to produce embryonic ([zeta]
2
[epsilon]
2
, [alpha]
2
[epsilon]
2
, [zeta]
2
[gamma]
2
), foetal ([alpha]
2
[gamma]
2
) and adult ([alpha]
2
[beta]
2
and [alpha]
2
[delta]
2
) globins. While the mechanism(s) responsible for the co-ordinated expression of these genes is hypothesised to occur at the transcriptional level (
1
), little is known about the specific sequences within these genes that may play a role in the regulation of the switching
process.
High level expression of the [alpha]- and [beta]-like globin genes in stably transfected cell lines and
transgenic mice is dependent upon sequences located far upstream of each
respective gene cluster. The [beta] locus control region ([beta]LCR), is located 5-20 kb upstream of the [beta]-globin gene locus and consists of four erythroid-specific DNase I hypersensitive sites (
2
,
3
). The [beta]LCR can confer high level expression to a linked [beta]- or [alpha]-globin gene in a position- independent, copy number-dependent manner in transgenic mice (
2
-
6
). A region with some of these properties has been localised 40 kb 5' to the human [alpha]-globin gene cluster. Like the [beta]LCR, this region (HS -40) confers high levels of erythroid-specific expression to a linked [alpha]- or [zeta]-globin gene in
transgenic mice (
7
-
10
).
When a 70 kb fragment containing the whole of the [alpha]-globin cluster including HS -40 is expressed in transgenic mice, the human [zeta]-globin gene is developmentally regulated and
matches that of the endogenous mouse [zeta]-globin gene (
11
). This correct developmental regulation is also seen when the [zeta]-globin gene is attatched to an HS -40[alpha] gene fragment (
9
) as well as with [mu]LCR and [beta]LCR HS2 constructs (
12
,
13
). These results suggest that developmental regulation is controlled by
sequences in and around the [zeta]-globin gene itself.
Deletions of the [zeta]-globin gene promoter in HS2[zeta] constructs have suggested that as little as 128 bp of the promoter are
sufficient to confer embryonic expression on this gene (
14
). Furthermore, we and others have demonstrated that a 556 bp [zeta]-globin promoter fragment is sufficient to confer embryonic-specific expression to a linked lacZ reporter gene (
10
,
15
). The sensitivity of this assay, together with its ability to provide
quantitative data on expression levels and intercellular variability, led us to
use it to better define the promoter sequences that confer embryonic
expression.
In the present study, we first examined the effect that deletion of progressive
amounts of [zeta]-globin 5' flanking region had on [zeta]-globin gene expression in transiently transfected
versus stably integrated erythroid cells. [zeta]-globin promoter/lacZ deletion constructs were then employed for the generation of transgenic mice followed
by analysis of lacZ staining pattern and expression levels in transgenic
embryos and foetuses.
To produce [zeta]-globin/CAT constructs containing varying amounts of [zeta]-globin 5' flanking region,
Bal
31 digestion of [zeta]-globin's 5' flanking region was employed as previously described (
16
). [zeta]-globin 5' flanking regions containing 84, 116, 195 or 556 bp of DNA
(cap = 0) were ligated into the
Sma
I site of the vector pCATO (
16
). A 4 kb
Hin
dIII fragment (
Bam
HI linkered) containing HS -40 was then cloned into the
Bam
HI site of each construct. To employ these constructs for the generation of
stable erythroid cell lines, a 2.7 kb fragment containing the SV40 early and
late promoter driving Neo resistance was cloned into the
Hin
dIII sites of the above CAT constructs (Fig.
1
B).
Transgenic mice were generated by micro injection of linear DNA fragments into
pronuclei of fertilised eggs from CBA * C57 crosses (
17
). Transgenic progeny were identified and copy number of transgene determined as previously described (
10
). Hemizygous lines were established by mating transgenic founders to CBA * C57 F1 mice.
To generate transgenic foetuses containing an [alpha]-globin promoter/ lacZ construct, [alpha]575/lacZ/HS -40 was cut with
Asp
718 and
Hin
dIII.
The released fragment was purified as previously described (
10
) and injected into fertilised F1 eggs followed by embryo transfer into pseudo-pregnant pathology outbred (PO) mice. After 16.5 days of gestation, the foetuses were removed and subjected to lacZ analysis
(see below).
Transgenic males from each line were mated to female wild-type F1 mice. The appearance of a vaginal copulation plug was considered day
0.5. At day 8.5-9.5 post-fertilisation, whole embryos were fixed and assayed for lacZ activity employing 5-bromo-4- chloro-3-indoyl-[beta]-d-galactopyranoside (X-gal) as
previously described (
10
). To assay for [zeta]-globin promoter activity at day 16.5, transgenic foetuses were identified, their livers removed, fixed and
subjected to lacZ analysis as above. Whole embryos and livers were photographed
on a dissection microscope. For more detailed histochemical analysis, embryos and livers were embedded in paraffin, sectioned (5 [mu]m) and counter stained with cresyl violet or eosin.
To analyse [alpha]575/lacZ/HS -40 expression, livers from 16.5 day old transgenic foetuses were
removed and fixed, followed by incubation in X-gal as previously described (
10
).
To determine the percentage of lacZ positive cells in the blood of transgenic
embryos and foetuses containing [alpha]- or [zeta]-globin/lacZ/HS -40 constructs, peripheral blood from 10.5 and
16.5 day old embryos and foetuses was isolated and stained with X-gal as previously described (
15
). The number of lacZ positive cells were counted employing a haematocytometer.
In order to quantitate [zeta]-promoter/lacZ/HS -40 expression levels during development, peripheral blood
from 10.5-16.5 day old transgenic embryos and foetuses was obtained as above. Blood
cells were pelleted by centrifugation and re-suspended in 250 mM Tris pH 7.5. The cells were subjected to freeze-thaw three times followed by a 10 min centrifugation in a microfuge. The supernatant was removed and assayed for protein concentration using a BioRad
protein assay kit. Analysis of lacZ activity in extracts was performed as
previously reported (
10
).
To quantitate lacZ activity in [alpha]575/lacZ/HS -40 transgenic foetuses, blood from 16.5 day old transgenic
foetuses was isolated and lacZ activity assayed as above.
K562 cells were maintained in DMEM supplemented with 10% foetal calf serum, 100 [mu]g/ml penicillin, 100 U/ml streptomycin and 2 mM glutamine. Putko cells were
maintained in RPMI 1640 supplemented with 10% foetal calf serum, 100 [mu]g/ml penicillin and 100 U/ml streptomycin.
Pools of K562 clones stably transfected with [zeta]-promoter/lacZ/HS -40/Neo constructs were generated as previously described (
18
). For transient transfections, Putko cells were electroporated with [zeta]-promoter/lacZ/HS -40 constructs as previously described (
19
). Plasmid pIRV (
20
) (5 [mu]g) was employed as a co-transfection control. Forty-eight hours after transfection, cells were harvested and
extracts produced as above. CAT and lacZ analysis was performed as outlined in
Pondel
et al.
(
19
).
We previously showed that deletion of sequences 5' to the human [alpha]-globin CCAAT box caused a significant decrease in [alpha]-globin promoter activity when linked to HS -40. Interestingly, this decrease in
expression occurred in stably but not transiently transfected cells (
18
). In order to determine how much sequence 5' to the [zeta]-globin gene was required to give readily detectable
expression, [zeta]-globin/CAT/HS -40 deletion constructs were employed for the generation of
stably transfected erythroid cell lines (Fig.
1
B-F). Since HS -40 does not confer complete position independent expression to a
linked [zeta]-globin gene (
8
-
10
,
15
), CAT assays were performed on extracts from pools of G418 resistant clones
(>100 clones per pool). By analysing CAT activity from complex pools of clones,
the effect position of integration has on [zeta]-promoter activity should be averaged out amongst the clones. The results of these experiments are depicted in
Figure
2
. Deletion of sequences between -116 and -556 (cap = 0) caused no significant decrease in [zeta]-promoter/CAT/HS -40 activity. When an additional 32 bp were deleted ([zeta]84/lacZ/HS -40), a 2-3-fold drop in promoter
activity was observed.
In the developing mouse, erythropoiesis first occurs in yolk sac blood islands
between 8 and 14 days gestation. Mouse [zeta]-globin expression at this site reaches its peak at ~9.5 days of development and then gradually decreases to almost undetectable levels by
15-16 days of gestation (
22
). We previously showed that 556 bp of human [zeta]-globin 5' flanking region linked to lacZ and HS -40 was sufficient to direct a similar pattern of lacZ
activity in transgenic mice (
10
). In order to more clearly define sequences that direct embryonic specific
expression of the human [zeta]-globin gene, transgenic lines containing 556, 84 or 67 bp of human [zeta]-globin 5' flanking region linked to lacZ and HS -40 (556, 84 and 67/lacZ/HS -40) were generated (Fig.
3
). Southern blot analysis identified six transgenic lines containing 12-150 copies of the [zeta]84/lacZ/HS -40 construct and nine transgenic lines containing 1-300 copies of [zeta]67/lacZ/HS -40 construct (Table
1
A). Embryos from three lines containing [zeta]84/lacZ/HS -40 and two lines containing the [zeta]67/lacZ/HS -40 construct showed no lacZ activity (Table
1
A). Southern blot analysis did not reveal any obvious rearrangement or deletion
of the above constructs in these mice suggesting that the absence of [zeta]-promoter activity is due to position effects. Two transgenic lines
containing 12 copies of [zeta] 556/lacZ/HS -40 (
10
) were employed as a control for correct developmental regulation.
To determine the degree to which [zeta]-promoter activity was suppressed at foetal stages of development,
peripheral blood from transgenic embryos and foetuses was isolated. Half of
each sample was employed for histochemical staining and the other half for quantitative [beta]-galactosidase assays. We observed a marked variation (0.1-100%) in the percentage of lacZ expressing erythroid cells in 10.5 day old embryos from different
transgenic lines (Table
1
A). This variation was consistent even after overnight incubation of blood in X-gal. By day 16.5 post-fertilisation, the percentage of lacZ positive cells in peripheral blood from all
transgenic lines dropped significantly. In contrast, day 16.5 peripheral blood
from all transgenic foetuses containing the [alpha]575/lacZ/HS -40 construct showed high proportions of lacZ positive cells (Table
1
B).
Robertson
et al
. (
15
) reported that expression of [zeta]-promoter/lacZ/HS -40 constructs in erythroid cells of transgenic mice was bi-modal (on/off). In contrast, we see heterogenicity of
intercellular lacZ expression (data not shown) in all of our transgenic mice. This suggests
that [zeta]-promoter activity is variable in the erythroid cells of any given
transgenic mouse.
LacZ assays on peripheral blood lysates showed that [zeta]-globin promoter activity was highly variable between different transgenic lines (Table
1
A). Expression levels in each line did not appear to be correlated with copy
number. However, statistical analysis revealed a significant correlation (r
2
= 0.911) between the number of lacZ positive cells and lacZ expression levels in
10.5 day old transgenic embryos. By 16.5 days of development, lacZ expression
in peripheral blood of all transgenic lines dropped significantly. In contrast,
peripheral blood from transgenic foetuses containing [alpha]575/lacZ/HS -40 showed abundant levels of lacZ expression in cell lysates (Table
1
B).
To carry out a more detailed analysis of [zeta] promoter activity during development, additional lacZ analysis was performed on five high-level lacZ expressing transgenic lines at four developmental time points (Fig.
6
). Peripheral blood lysates from all transgenic lines analysed showed a 2-3-fold reduction in [zeta]-promoter activity by 12.5 days post-fertilisation. By 14.5 days of development, lacZ
expression dropped on average, 11-fold. An additional 7-fold drop in lacZ activity occurred between 14.5 and 16.5 days of
development. This pattern of suppression matches that of the endogenous [zeta]-globin gene as well as that of the intact human [zeta]-globin gene in transgenic mice (
8
,
9
).
During normal mouse development, expression of the [zeta]-globin gene is essentially limited to the primitive erythroid cells
produced in the blood islands of the embryonic yolk-sac. The results presented here demonstrate that as little as 67 bp of the
human [zeta]-globin promoter, in the presence of HS -40, appears to be sufficient to confer embryonic stage
specificity on a linked lacZ reporter gene. Erythroid specific, high level
expression of the reporter was observed in embryonic erythroblasts in several
lines of transgenic mice. LacZ expression levels declined 40-100-fold by day 16.5 of gestation. These results extend those of Sabath
et al
. (
14
) who obtained similar results with a [zeta]-globin gene containing 128 bp of [zeta]-globin 5' flanking region, albeit under the control of the
[beta]LCR HS2.
Robertson
et al
. (
23
) have also used [zeta]-promoter/lacZ/HS -40 constructs and reported a much smaller decrease in
expression (only 1.5-15-fold) between days 12.5 and 17.5 with the [zeta]-globin promoter truncated to either -550 or -127 bp. However, in these cases, the HS -40 fragment was in the opposite
orientation relative to the [zeta]-promoter/lacZ fragment. When the HS -40 fragment was in the same 5'-3' orientation as the reporter gene (550[zeta]R in their nomenclature), the
decline in lacZ expression in their mice was similar to that in ours. As HS -40 has been shown to be orientation independent (
7
), it seems unlikely that this result is brought about by the orientation of HS -40 itself. However, reversing the orientation of that fragment within the
construct would bring the core enhancer sequences much closer (~1.0 versus ~3.5 kb) to the [zeta]-globin gene promoter of the next copy downstream in a
tandem head-to-tail array. We would suggest, therefore, that perhaps the close
proximity of the enhancer to the promoter has partially overridden the
developmental control mechanism in the studies of Robertson
et al
. (
23
).
Liebhaber
et al
. (
24
) have recently suggested that sequences in the [zeta]-globin promoter, the transcribed portion of the gene and 3' to the gene are necessary for complete silencing of the
human [zeta]-globin gene in post-embryonic transgenic mice. When the 557 bp [zeta]-globin promoter was attached to an [alpha]-globin gene there was only a 2.7-fold drop in expression between
9.5 and 16.5 days as opposed to a 50-fold drop with the intact [zeta]-globin gene. However, this result does not preclude the
possibility that the [alpha]-globin gene contributes sequences that oppose the silencing effects
of the [zeta]-promoter. Furthermore, these studies used the [beta]-globin [mu]LCR closely apposed to the [zeta]-globin promoter and again this could
affect the normal pattern of developmental regulation.
Watt
et al
. (
16
) showed that in the absence of HS -40, GATA-1 binding sites present in the 5' flanking region of the [zeta]-globin promoter direct its erythroid
specificity. Our data shows that GATA-1 binding sites in the 5' flanking region of the [zeta]-globin promoter are not required for erythroid specific
activity of the [zeta]-globin promoter when it is linked to HS -40. Since HS -40 enhancer capability is erythroid-specific (
7
,
19
,
25
), we hypothesise that erythroid-specific expression of [zeta]67/lacZ/HS -40 is mediated primarily by the HS -40 element.
A number of transcription factors bind to sequences within 67 bp of [zeta]-globin 5' flanking region. The factors CP1 or CP2 bind to the [zeta]-globin CCAAT box (
16
,
26
). The [zeta]-promoter also contains a TATA box, suggesting this region binds TATA
binding protein and TATA box associated factors (TBP and TAF). Although the human [alpha]-globin promoter also binds these proteins, it is transcriptionally active at all stages of development. Clearly, there are as of yet,
unidentified sequences within the [zeta]-globin proximal promoter that direct its embryonic-specific transcriptional activity. Alternatively, the [zeta]-globin CCAAT and/or TATA box may be interacting with developmental stage
specific transcription factors that regulate the switching process. Our delineation of a small region
capable of directing correct temporal transcriptional activity of the [zeta]-globin promoter should facilitate the identification of such sequences or core promoter elements. We will then be in a
position to study their interaction with nuclear regulatory factors that play
an important role in directing the [zeta]- to [alpha]-globin switch.
We thank Dr D. Higgs for useful discussions and encouragement throughout these
studies. This work was supported by a Wellcome project grant to N.J.P. and
M.D.P.
REFERENCES
Return