Sequences within and flanking hypersensitive sites 3 and 2 of the
[beta]
-globin locus control region required for synergistic versus additive
interaction with the
[epsilon]-globin gene promoter
Sequences within and flanking hypersensitive sites 3 and 2 of the [beta] -globin locus control region required for synergistic versus additive interaction with the [epsilon]-globin gene promoter
John D.
Jackson
1,2,+
,
Webb
Miller
2,3
and
Ross C.
Hardison
1,2,
*
1
Department of Biochemistry and Molecular Biology,
2
The Center for Gene Regulation and
3
Department of Computer Science and Engineering, The Pennsylvania State
University,
University Park
, PA 16802,
USA
Received June 7, 1996;
Revised and Accepted September 20, 1996
ABSTRACT
The locus control region is required for high-level, position-independent expression of mammalian
[beta]
-globin genes. It is marked by five major DNase hypersensitive sites (HSs)
in a 16 kb region of chromatin, and the protein-DNA complexes that form these HSs may interact in a holocomplex that
carries out the full function of the locus control region. Previous studies
showed that a large rabbit DNA fragment containing both HS2 and HS3 in their
native sequence context and spacing produced a much larger increase in
expression of a linked reporter gene than the sum of the largest effects
observed with DNA fragments containing HS2 or HS3 individually. To test whether this reflected a synergistic interaction between the 200-400 bp cores of the HSs or if this effect required additional sequences outside the cores, combinations of different restriction fragments containing HS2 or HS3 were tested for their ability to increase the
expression of a hybrid
[epsilon]
-globin-luciferase reporter gene in transfected K562 cells. The results
show that the human HS2 and HS3 cores do not interact either additively or
synergistically with the reporter gene when juxtaposed, and separation by spacer DNA has little effect on their function. Fragments of human DNA containing cores plus flanking sequences for HS3 or HS2 show an additive effect in combination, whereas homologous fragments of rabbit DNA containing HS3 and HS2 interact synergistically. At least part of this
difference localizes to the rabbit DNA fragment containing HS3, which can
interact synergistically with the human DNA fragment containing HS2. The region 5
'
to the HS3 core plays a role both in the cooperative interaction observed with
the rabbit DNA fragment and the domain-opening observed with the human DNA. A minor DNase HS maps to this region,
and the pattern of sequence conservation is consistent with some difference in
function between species.
INTRODUCTION
The human [beta]-globin gene domain is located on chromosome 11 and includes five functional genes, [epsilon],
G
[gamma],
A
[gamma], [delta] and [beta] (referred to as [beta]-like globin genes). The entire domain of >= 100 kb is regulated by a locus control region (LCR) extending from
6 to 22 kb upstream from the [epsilon]-globin cap site (
1
). The LCR includes five strong DNase hypersensitive sites, HS1-HS5 (
2
-
4
). Deletion of a region encompassing HS2 through HS5 in Hispanic [gamma][delta][beta]-thalassemia abolishes expression of the [beta]-like globin genes and leaves them in a
closed chromatin conformation (
5
). The LCR confers high level, position-independent expression on a linked globin gene in transgenic mice (
6
). Thus, the LCR is required for domain-opening, enhancement of [beta]-like globin gene expression, and insulation from negative
position effects.
Much effort has gone into defining functional regions within the LCR. Deletional
analysis of DNA fragments containing individual HS1, 2, 3 or 4 has defined the
minimal fragments that confer position-independent expression on a linked [beta]-globin gene in transgenic mice. These `core' fragments are generally 200-400 bp (
7
-
9
). With the exception of HS1, which appears to be dispensable since a naturally
occurring deletion encompassing it does not affect [beta]-globin gene expression (
10
), other functions have been assigned to restriction fragments containing
individual HSs (Fig.
1
). For example, HS2 contains a strong enhancer (
7
,
11
,
12
), HS3 has a domain-opening activity (
13
,
14
), HS4 generates a DNase HS in transfected cells (
15
) and HS5, which is contained within a matrix attachment region (
16
), can insulate reporter gene expression from some position effects in stably
transfected cells (
17
,
18
).
MATERIALS AND METHODS
DNA constructs
Fragments from both the rabbit and human [beta]-globin LCRs were inserted 5' to the [epsilon]-globin gene promoter in pBS[epsilon]-luc.4, a hybrid [epsilon]-globin-luciferase
reporter gene previously described (
14
,
23
). Figure
2
shows a map of the LCR fragments with restriction endonuclease cleavage sites
and Table
1
lists the start and stop positions for each fragment.
Transfection of K562 cells and measurement of expression
Transient and stable expression experiments were performed and analyzed as
previously described (
14
), with the following minor modification to the stable expression experiments.
Genomic DNA from each stably expressing clone was digested with
Nco
I and
Eco
RV and probed with a 1.4 kb
Nco
I-
Eco
RV fragment isolated from [epsilon]-luc.4, which contains a portion of [epsilon]-globin exon 1 and most of the luciferase coding
region.
Clones with high copy numbers tended to have low expression per copy, indicating
that the concentration of endogenous transcription factors may be limiting in
these cases. Hence the mean expression per copy calculation for each construct
included only those clones with copy numbers of <= 10.
Sequence analysis
Sequence alignments and positions of both conserved sequence blocks and
differential phylogenetic footprints were computed as described in Slightom
et al
. (
24
).
RESULTS
Additive versus synergistic effects of human versus rabbit HS2 and HS3
We previously reported (
14
) that a large DNA fragment from the rabbit LCR containing both HS2 and HS3
synergistically increased expression of an [epsilon]-globin reporter gene in both transient and stable expression
assays, producing an effect much greater than the sum of the effects of the
individual HS cores or the sum of the individual HS units (summarized in Fig.
3
). None of the LCR fragments that we have tested in this assay in stably
transfected K562 cells give complete independence from position effects. The
site of integration could affect the level of stable expression, e.g., by
influencing the chromatin structure or by interactions between the integrated construct and other (unknown)
cis
-acting elements at a specific site. Therefore, it is necessary to examine
several clones for each construct. The overall profile of the expression per
copy in several clones, as well as the mean level of expression, allows one to
see effects beyond those exerted by the site of integration, e.g. as seen in
the absence of LCR fragments (
14
).
A single pair of HS2 and HS3 cores shows no effect in combination
Although the human DNA fragment containing both HS3 and HS2 showed a weaker
effect than the homologous rabbit DNA fragment, it was much stronger than the
effect of individual HS cores. Thus it is possible that combining the cores
could produce the 100-150-fold effect seen with the human DNA fragments containing both HS3
and HS2. However, as shown in Figure
4
, the eight clones stably expressing [epsilon]-luciferase regulated by the juxtaposed cores gave only an average 3-fold increase over the baseline (no added LCR), comparable
with that seen with either core alone and a much smaller effect than that seen
with the larger fragments in Figure
3
. Additionally, the transient expression results are about the same as with HS2
alone (Fig.
4
). Thus, the juxtaposed cores are not interacting to produce the strong positive effects observed with the native DNA fragments containing both HSs.
Proper spacing between HS2 and HS3 does not restore the full positive effect of
the native DNA fragments
Sequences surrounding human HS3 and HS2 contribute to their positive effects in
combination
To localize the sequences outside the cores that contribute to the positive
effect of the human DNA fragment containing both HS3 and HS2, a number of human
DNA fragments containing HS3 were combined with fragments containing HS2 and
tested for their effects on transient and stable [epsilon]-luciferase expression. The 1.9 kb
Hin
dIII-
Hin
dIII HS3 unit has the strongest activity in stable expression assays for any
fragment containing HS3, and the HS2 core has a strong enhancer activity. It is
possible that these could interact to produce the effect of the large native
fragment containing both HS3 and HS2. However, as shown in Figure
4
, this combination of LCR fragments generates effects comparable with that of the HS2 core alone in transient expression and that of the 1.9 kb fragment alone in stable expression, again indicating
independent function of human fragments containing HS3 and HS2. This effect is somewhat less than those of the 5.3 kb
Bam
HI-
Bgl
II native fragment in stable expression assays and considerably less in
transient expression assays, showing that no synergism occurs between the HS3
unit and the HS2 core. These data also indicate that the sequences outside the HS2 core are necessary, and
not providing, e.g., a redundant domain opening activity.
Two different DNA fragments containing HS3 were then linked to a larger HS2 unit
to test for their effects in the context of the HS2 core plus flanking
sequences. As shown in Figure
5
A, a 2.0 kb
Bam
HI-
Avr
II HS3 unit was combined with 1.5 kb
Kpn
I-
Bgl
II HS2 unit, which is essentially the human 5.3 kb
Bam
HI-
Bgl
II native fragment with a deletion between the internal
Avr
II and
Kpn
I sites. This construct caused a 73-fold increase on stable [epsilon]-luciferase expression, which is only slightly greater than
the 59-fold increase from the HS2 unit alone, but not as great as the 111-fold increase seen with the parental 5.3 kb
Bam
HI-
Bgl
II fragment (Fig.
5
A). Addition of the human HS3 unit had no effect on the enhancement by the 1.5
kb HS2 unit in transient expression. The results with both stable and transient
expression are consistent with the largely additive effects seen with the human
LCR fragments. They also suggest that sequences between
Avr
II and
Kpn
I (between HS3 and HS2) contribute to the positive effect of the human native
fragment containing both HS3 and HS2.
The 225 bp human HS3 core fragment had no positive effect on either transient or
stable expression (Fig.
5
A) when combined with the 1.5 kb HS2 unit. Indeed, the HS3 core caused a decrease in expression
in stably transfected clones, suggesting that sequences surrounding the HS3 core may be needed to prevent a negative interaction
between the HS3 core and the 1.5 kb HS2 unit.
Sequences surrounding rabbit HS3 contribute to its interaction with HS2
Two rabbit DNA fragments containing HS3 were also combined with the human 1.5 kb
HS2 unit to address whether they contributed to the synergistic effect observed
with the 5.6 kb rabbit DNA fragment. Addition of a 450 bp rabbit DNA fragment,
which contains a well-conserved AP1 binding site in addition to the HS3 core, to the HS2 unit
gave a large increase in both stable and transient expression (Fig.
5
B; note the difference in scale from Fig.
5
A). Addition of the larger 3.1 kb rabbit HS3 unit to the 1.5 kb human HS2 unit
caused a 258-fold increase in [epsilon]-luciferase expression, much greater than that of either the
3.1 kb (5-fold) or the 1.5 kb (59-fold) fragments alone. This construct combining a rabbit HS3 unit
and a human HS2 unit gave a much stronger effect on stable expression than the
combination of human HS3 and HS2 units (258-fold versus 73-fold). This argues that the rabbit DNA fragments containing HS3 can
interact with fragments containing HS2 for a stronger, synergistic function,
whereas the human HS3 unit operates independently of the HS2 unit. This is consistent with the strong synergism seen by the 5.6 kb rabbit fragment containing HS2 and HS3 (Fig.
3
).
DNA sequences between the HSs do not stimulate expression on their own
To address whether the functions associated with sequences outside the HS cores
were separable from those inside the HS cores, three fragments lacking the HS
cores were tested for effects on [epsilon]-luciferase expression in transfected K562 cells. As shown in Figure
6
, none of the fragments increased reporter gene expression, showing that the HS cores must be present to see the function of sequences
outside the cores.
DISCUSSION
Results in this paper show that the synergistic effects seen between HS3 and HS2
in a 5.6 kb rabbit DNA fragment containing both HSs (
14
) are not observed with a homologous human DNA fragment from the [beta]-globin LCR. The effect of HS3 and HS2 units from humans is additive
when combined in an [epsilon]-luciferase expression vector. However, this increase in expression is much greater than that observed with the individual HS cores, or
the combined cores in juxtaposition, or the two cores separated by spacer DNA
fragments. Further studies showed that both large units are needed for the
higher level of activity; a HS unit plus a HS core do not provide the full
activity of the native fragment with both units. Thus sequences outside the
cores are needed not only for the effects of individual HSs, but they are also
needed when HSs are combined. The regions flanking the cores presumably also
bind proteins to provide some additional function, perhaps proper orientation
of the cores within the LCR. Although the sequences between the cores are not
exclusively spacers, it is still likely that proper spacing between the HS
cores is maintained to achieve optimal LCR function. That spacing may be
generated by folding the cores and flanks, with associated proteins, into a
series of larger units that comprise the LCR.
The observations in this paper were made with an [epsilon]-globin reporter gene with the first intron and second exon replaced
by a luciferase coding block. An earlier study using an intact [beta]-globin reporter gene in transfected cells showed no additive effects
of HS3 and HS2 units, but increased expression was observed with combinations
of three HS units (
33
). Although both studies show the ability of the larger HS units to interact
functionally, the target reporter gene and the exact endpoints of the LCR
constructs appear to affect the results obtained. Only further studies will show whether this reflects a real difference in the LCR-requirements of different promoters, or is a result of partial function
from using only a subset of the LCR DNA.
The small increase in activity observed when additional sequences located 5' to the HS3 core are included in the human DNA fragments containing both HS3 and HS2 (compare the 5.3 kb
Bam
HI-
Bgl
II fragment and the 6.1 kb
Nsi
I-
Bgl
II fragment, Fig.
3
) suggests that sequences 5' to the
Bam
HI site may also contribute to the increase in expression. This is supported by
the results of a 5' deletional analysis of the 1.9 kb
Hin
dIII-
Hin
dIII human DNA fragment containing HS3. Removal of ~600 bp from the 5' end caused a substantial reduction in the activity of the
individual HS3 unit (
24
). Thus sequences located between 700 and 1300 bp 5' to the HS3 core are needed for the function of the large HS3 unit. This
is close to the position of the minor HS mapped by Tuan
et al
. (
2
) and Stamatoyannopoulos
et al
. (
25
), shown in Figure
7
.
ACKNOWLEDGEMENTS
This work was supported by PHS grants RO1 DK27635 and RO1 LM05773 (to R.H.) and
RO1 LM05110 (to W.M.).
REFERENCES
1 Grosveld,F., Antoniou,M., Berry,M., de Boer,E., Dillon,N., Ellis,J., Fraser,P., Hanscombe,O., Hurst,J., Imam,A., Lindenbaum,M., Philipsen,S., Pruzina,A., Strouboulis,J., Raguz-Bolognesi,S. and Talbot,D. (1993) Philos. Trans. R. Soc. Lond.339, 183-191.
2 Tuan,D., Solomon,W., Li,Q. and London,I.M. (1985) Proc. Natl. Acad. Sci. USA82, 6384-6388.MEDLINE Abstract
3 Forrester,W.C., Thompson,C., Elder,J.T. and Groudine,M. (1986) Proc. Natl. Acad. Sci. USA83, 1359-1363.MEDLINE Abstract
10 Kulozik,A.E., Sail,S., Bellan-Koch,A., Bartram,C.R., Kohne,E. and Kleihauer,E. (1991) J. Clin. Invest.87, 2142-2146.MEDLINE Abstract
11 Tuan,D., Abelovich,A., Lee-Oldham,M. and Lee,D. (1987) in Stamatoyannopoulos,G. and Nienhuis,A. (eds), Developmental Control of Globin Gene Expression. A.R.Liss, Inc., New York, pp. 211-220.
12 Tuan,D., Solomon,W., London,I. and Lee,D. (1989) Proc. Natl. Acad. Sci. USA86, 2554-2558.MEDLINE Abstract
33 Collis,P., Antoniou,M. and Grosveld,F. (1990) EMBO J.9, 233-240.MEDLINE Abstract
34 Sadelain,M., Wang,C.H., Antoniou,M., Grosveld,F. and Mulligan,R.C. (1995) Proc. Natl. Acad. Sci. USA92, 6728-6732.MEDLINE Abstract
35 Akusjarvi,G. and Wadell,G. (1990) in O'Brien (ed.), Genetic Maps: Locus Maps of Complex Genomes, Book 1, Viruses. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 1.98-1.101.
36 Gumucio,D.L., Shelton,D.A., Blanchard-McQuate,K., Gray,T.A., Tarle,S.A., Heilstedt-Williamson,H., Slightom,J., Collins,F.S. and Goodman,M. (1994) J. Biol. Chem.269, 15371-15380.MEDLINE Abstract
37 Hardison,R., Chao,K.-M., Adamkiewicz,M., Price,D., Jackson,J., Zeigler,T., Stojanovic,N. and Miller,W. (1993) DNA Sequence4, 163-176.MEDLINE Abstract
38 Wijgerde,M., Grosveld,F. and Fraser,P. (1995) Nature377, 209-213.MEDLINE Abstract
*To whom correspondence should be addressed at: Department of Biochemistry and
Molecular Biology, The Pennsylvania State University, 206 Althouse Laboratory,
University Park, PA 16802, USA. Tel: +1 814 863 0113; Fax: +1 814 863 7024;
Email: rch8@psu.edu
+
Present address: Biology Department, University of Rochester, 425 Hutchison
Hall, Rochester, NY 14627, USA
