| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ARTICLES |
Thyroid Division, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts 02115
Address all correspondence and requests for reprints to: P. Reed Larsen, M.D., Thyroid Division, Brigham and Women’s Hospital, Harvard Institutes of Medicine, Room 560, 77 Avenue Louis Pasteur, Boston, Massachusetts 02115.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
50%) of plasma
T3 in the hyperthyroid human, but a much lower fraction in
the euthyroid state (3, 4, 5). In part, this is caused by the fact that
the expression of the hdio1 gene increases in response to
T3 (6). Recent studies have identified two thyroid hormone
response elements (TREs) in the first 2.5 kbp of the 5' flanking region
of hdio1 (Fig. 1
) (6, 7). A
proximal TRE (TRE1) is located approximately 90 bp upstream of the
transcription start site (TSS) on the lower strand and consists of a
direct repeat of idealized thyroid hormone receptor (TR) binding
octamers, with 10 bp separating the two TR binding sites. The response
of this TRE to T3 is independent of retinoid X
receptor (RXR) and, despite the wide separation between the two octamer
sites, both are required for a response to thyroid hormone. The
upstream TRE, TRE2, is located about 660 bp 5' to the TSS. It is a
classical direct repeat of RXR/TR binding half-sites, with a 4-bp
separation.
|
A second puzzling result was found in our earlier studies of TRE1. Using methylation interference footprinting, we found that the 3' of the two octamers was the dominant TR-binding octamer and that TR remained bound to this site in the presence of T3. However, mutations in critical nucleotides of either the 5' or the 3' octamer site eliminated the T3 response. This suggested that both octamer binding sites were required for TR binding and that there was functional cooperativity between the two. Such an effect could amplify the modest response to T3 conferred by a monomeric TR binding site (12, 13). It was also not clear from our earlier studies whether the direct repeat orientation of the two sites or the presence of the higher affinity octamer in the 3' position (analogous to the 3' location of the TR binding half-site in DR+4 TRE) was required for function. The present studies were designed to provide more detailed information about the mechanism by which hdio1 gene expression is increased in the hyperthyroid patient.
Lastly, recent studies have shown that the hdio1 5' flanking region confers a response to retinoic acid (RA) that is not eliminated by mutations in TRE2 (14). Our present results confirm these data and show, furthermore, that the RA response is eliminated by specific inactivating mutations in the two half-sites of TRE1 in the context of the 716-bp wild-type hdio1 promoter/5' flanking region. Thus, TRE1 acts as both a TRE and a retinoic acid response element in this promoter.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
) were inserted into the
BamHI site 5' to HDCAT 649 and HDCAT 430 (2B), respectively.
In the same way, synthetic double-stranded oligonucleotides for
wild-type or mutant TRE1 were inserted into the BamHI site
3' to the TK promoter in pUTKAT3, as described previously
(16, 17). All final constructs were sequenced using the dideoxy method
to confirm their structure.
|
(20, 21). The
latter was a gift from Dr. Ron Evans. An empty CDM8 vector was used to
keep the amount of transfected DNA constant. The medium was
charcoal-stripped FBS with or without 5 x 10-8
M T3 or 1 x 10-7
M all trans-RA. CAT activity was determined by a
phase extraction procedure modified from Seed and Sheen, and activity
was calculated as the percent conversion/100 µl cell extract protein
in an overnight incubation (15 h), and hGH as counts per min/100 µl
medium (22). CAT activities varied from 0.4–1000% converted/100 µl
extract/15 h, but cell lysate protein was diluted to keep CAT activity
within the linear range of the assay (0.3–50% converted). Medium hGH
counts also varied with the transfection, but again, media was diluted
to generate counts in the linear portion of the hGH assay (1000–50,000
cpm). T3 or RA responsiveness (with ligand/without ligand)
was expressed as the ratio of CAT/hGH from cells cultured in the
presence of ligand divided by the CAT/hGH ratio from the respective
paired plate cultured in the absence of ligand (17, 23).
Mobility shift assay
Chicken TRa1 (cTRa) was overexpressed in Escherichia
coli and purified, and gel shift experiments were performed using
conditions previously described (24). The same wild-type and mutant
TRE1s used in the transient transfection constructions (Fig. 2B
) were
radiolabeled with [32P]deoxy-ribothymidine
5'-triphosphate (Dupont NEN, Boston, MA) by Klenow fill-in reaction and
gel purified. Labeled probe (15,000 cpm; 4.4 fmol) was incubated with
purified cTRa (5–50 fmol) in a 30-µl reaction containing 100 ng poly
(dI-dC), 88 mM KCl, 10% glycerol, 25 mM
Tris-Cl, 500 µM EDTA, 0.05% Triton X-100, 10
mM ß-mercaptoethanol, and 5 mg BSA.
Quantitative analysis of receptor binding
Quantitation of gel mobility shift autoradiographs was performed
on a Molecular Dynamics model 300 series computing densitometer
(Molecular Dynamics, Sunnyvale, CA) using Molecular Dynamics Image
Quant software, as previously described, and cooperativity plotted
using the approach of Tsai et al. (6, 25).
Statistical analysis and graphing
Statistical analysis was performed using StatView 4.0 (Abacus
Concepts, Berkeley, CA) to determine the mean and SE of all
samples indicated. Dunnett’s t test for multiple
comparisons or paired t tests were used to evaluate the
statistical significance of differences, as indicated in the Table
legends.
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). TRE2 is contained within this
Alu sequence, and there is an SP1 binding site immediately adjacent to
the 5' half-site of TRE2. A second SP1 binding site is located at
position -44. The presence of the SP1 binding site immediately
adjacent to TRE2, which was not included when this was transferred to
the TKCAT vector in our initial studies, suggested a potential
explanation for the absence of a stronger response of TRE2 to thyroid
hormone in its wild-type position (6). Binding of SP1 close to a
TR-binding site has been shown to influence the response to thyroid
hormone (26).
Effects of the SP1 binding site on the response of hdio1 to
T3
To analyze the effect of the SP1 binding site on TRE2 function, we
examined the T3 response of chimeric hdio1
promoter CAT constructs in the human embryonic kidney-293 cells. The
716-bp hdio1 5' flanking region increased CAT expression
5.9-fold in response to T3 (Table 1). Deletion of the SP1/TRE2 site in
HD649CAT reduced T3 responsiveness to 4.4-fold. Removal of
the entire Alu sequence by deletion to position -430 had no further
effect on T3 induction. The basal hdio1 promoter
construct containing 99 bp is not T3 responsive (6). The
T3 induction of the two shorter constructs (HD649 and
HD430) is thus accounted for by TRE1 (Fig. 1). The greater
T3 response of the TRE2-containing HD716CAT, as opposed to
the 649- and 430-bp constructs, was caused by both lower basal
expression, as reflected in the CAT/hGH ratio in the absence of
T3, and higher CAT expression in the presence of
T3 (Table 1), although neither of these latter differences
could be shown to be statistically different because of unequal
variances.
To test the effects of the SP1 binding site on the T3
response of TRE2, we inserted either TRE2 or the wild-type SP1/TRE2
5' to HD430CAT (Fig. 2A). TRE2 alone conferred a 9.7-fold response to
T3, which was significantly higher than that of
SP1/TRE2-HD430CAT (P < 0.02, Table 2A). The higher T3 response
is caused by significantly lower basal CAT expression with the TRE2
construct than with the SP1/TRE2 element (0.15 vs. 0.30,
P < 0.02). This 2-fold decrease in basal expression
more than balanced the higher CAT expression with SP1/TRE2-HD430CAT.
The lower basal CAT expression with the TRE2 construct suggested that
endogenous SP1 interfered with the repression of the TRE2 containing
plasmid by unoccupied TR.
To examine this issue directly, we compared CAT expression from these
constructs in the presence and absence of coexpressed TR. Expression
of TR caused more than a 2-fold repression of TRE2-HD430CAT
expression, relative to the hGH control (P < 0.05)
(Table 2B). However, inclusion of the SP1 binding site, though not
affecting expression in the absence of cotransfected TR,
significantly reduced the repression of basal expression by the
unliganded TR. There was no significant difference in SP1/TRE2-HD430CAT
directed CAT expression in the presence and absence of TR. Thus, the
SP1 site interferes with both the transcriptional response of TRE2 to
T3 and with the basal repression by APO-TR.
The effect of the SP1 binding site is mitigated in the wild-type
hdio1 gene
To determine whether this effect also occurred in the wild-type 5'
flanking region of hdio1, we prepared constructs in which
TRE2 or the SP1/TRE2 cassettes were placed 5' to position 649. This
results in virtually complete reconstitution of the wild-type 5'
flanking region, lacking only 6 bp (-649 to -655) (Fig. 1).
Surprisingly, inclusion of the SP1 binding site did not affect the
expression of the TRE2 HD649CAT construct. The response to
T3 was not significantly lower, nor was there a difference
in the basal CAT expression in the absence of T3, as was
the case with HD430CAT (Table 3). The
T3 induction of SP1/TRE2 HD649CAT (5.5 ± 0.64) was
not different from that of HD716CAT (5.9 ± 0.99, Table 1), also
indicating that sequences 5' to the SP1 binding site at position -680
in HD716CAT did not attenuate the T3 response of the longer
construct. These results suggest an important effect of position
within the hdio1 gene 5' flanking region on the effect of
SP1 on the T3 response of TRE2.
|
and (6)], we
also examined the effects of changing its structure. Our previous
studies had shown that the affinity of the 3' site (TCAGGTCA) for TR
was higher than that for the 5' site (CCGGGTCA) (6). Artificial TREs
were designed either to exchange the dominant 3' half-site with the
weaker 5' site (mutant 1) or to reverse the orientation of the 3'
octamer, making this a near palindromic sequence separated by 10 bp
(Fig. 2B). These changes in structure had no effect on the potency of
TRE1 to confer a T3 response to the TK promoter, which was
3-fold in each case (Fig. 2B).
We also performed biochemical studies to analyze the binding of TR
to these sequences to determine whether these properties were altered
by these mutations. The TR binding to mutant 1 was essentially
identical to the wild-type TRE1, with double occupancy at higher
concentrations but occupancy of only a single TR at low concentrations
(Fig. 3A). Mutant 2 bound only one
molecule of TR, with no evidence that both octamers can be occupied
simultaneously. There was virtually no cooperativity of TR binding for
either TRE1 or mutant 1, as shown by the parallel increases in the
fraction of single- and double occupancy with increase in TR (Fig. 3B) .
|
). Addition of T3 virtually
eliminated the binding of a second TR, whether the half-sites were
arranged in the wild-type configuration or in that of mutant 1.
T3 also caused a more rapid migration of the monomeric
DNA-protein complex typical of that observed with ligand-occupied
receptor (Fig. 4A). Interestingly, T3 seemed to enhance the
binding of TR to mutants 1 and 2 but not to the wild-type element (Fig. 4B). Thus, the arrangement of identical half-site sequences within the
TRE influences the binding affinity of liganded TR but does not affect
its function (Fig. 2B).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-chain subunit of the high-affinity IGE
receptor (9) and, negative regulatory elements arising in Alu repeats
also have been identified (8). Thus, formation of TRE2 is an example of
a gain of function mutation in an Alu repeat, the first example of this
phenomenon for a TRE. This extends the scope of the potential functions
of Alu repeats to include enhancement of hormone responsiveness in a
gene which already contains a TRE. Given the large number of Alu
repeats present in the human genome and the conservation of the 3'
half-site of this particular repeat, it would not be surprising if
other examples were identified in the future. In this context, it is interesting that there is high similarity (70%) between the most proximal 200 bps of the human and mouse dio1 promoters (but without preservation of TRE1) but little or no similarity with hdio1 in the more 5' 1300 bps (29). In fact, no potent TREs have been identified in the 1.5-kb 5' flanking region of mdio1 (29). Thus, although both human and mouse dio1 gene expression is increased by thyroid hormone, the DNA sequences directing these responses are quite different.
The presence of the SP1 site just 5' of TRE2 in the Alu repeat led to
the interesting possibility that the binding of SP1 in this position
could influence the response of TRE2 to thyroid hormone. There are two
recently published examples of potential interactions between TR and
SP1 for other transcription factors. The first is in the human
epidermal growth factor receptor (EGFR) promoter, the expression of
which is repressed by thyroid hormone (26). In this case, the SP1
binding site included the first three G residues of a TR-binding
hexamer GGCGGGACT on the antisense strand in this promoter.
Because only a single hexameric TR-binding site was identified in this
region, the fact that unliganded TR inhibited EGFR expression in
transient assays through interference with SP1 binding was not
surprising. However, there was a modest enhancement of inhibition when
T3 was added. This result is somewhat unexpected unless, as
seems to occur with the TRE1 mutants (Fig. 4B), T3 enhances
monomeric TR binding. The authors concluded that TR reduced
expression of EGFR by steric interference with SP1 binding caused by
the significant overlap between the two binding sites (26). In the case
of TRE2, the situation is slightly different, in that the SP1 and the
RXR binding sites of the putative RXR-TR heterodimer do not overlap.
However, the results in Table 2 show that inclusion of the SP1 binding
site reduces the T3 induction by partially blocking
repression by unliganded TR because, in the absence of TR, SP1 does not
affect expression (Table 2B). The higher CAT expression in the presence
of T3-TR with the SP1/TRE suggests that both the
hormone-dependent and the hormone-independent (SP1) stimulatory factors
are operational (Table 2A). It is the blockade of the APO-TR-induced
repression of CAT expression that makes the calculated T3
induction of SP1/TRE2 HD430CAT lower (Table 2A).
|
B binding sites that contain a second TRE.
Interestingly, the two TR binding sequences in this LTR are arranged in
a DR+8 (or DR+10) orientation analogous to those of TRE1 of the
hdio1 promoter. In transient expression assays, the NF-B
TRE is dominant unless the viral tat factor is coexpressed,
in which case, the TRE embedded in the SP1 binding sites becomes active
(30).
Though the hdio1-SP1/TRE interactions shown in Table 2 are
less complex than those described by others, they provide another
variation on the theme of interaction between TRs and additional
transcription factors. Although such factors can interfere with the
magnitude of T3 induction (Table 2), they would also act
in vivo to stabilize expression of these genes when TR
occupancy is reduced.
Despite demonstration of SP1-induced interference with the APO-TR
effect in the 430-bp construct (Table 2A), a similar effect was not
seen with the 649-bp fragment (Table 3). Thus, the position of the SP1
site within the 5' flanking region influences its modulatory effect on
an adjacent TRE. Interactions between SP1 and TR in vivo
will be dependent on the relative concentrations of each of these
transcription factors in the nucleus. The high expression of TR in a
transfected cell could overcome the interference by endogenous SP1 when
the SP1-TRE complex is located in its wild-type position, but not in
the 430-bp construct. Thus, despite the negative results with the
649-bp fragment, SP1 might still influence basal D1 mRNA expression
in vivo.
The results in Tables 1 and 3 argue that the relatively weak
T3 effect contributed by TRE2 in the 716-bp wild-type
promoter is more likely caused by its position rather than by any lack
of responsiveness of the TATA-less hdio1 promoter. The
9-fold T3 response of the TRE2-HD430CAT construct (Table 2A) is quite similar to the T3 induction conferred by TRE2
on the TK promoter in earlier studies (6). Taken together, these
results suggest that the weak T3 induction conferred by
TRE2 is partly explained by its distance from the TSS, but in
vivo, it could also be influenced by SP1 binding adjacent to the
RXR binding site of the DR+4 TRE2 sequence.
With respect to TRE1, we found no functional effects of changing either
the position or the orientation of the two octamer TR binding sites.
Despite this, there was a reduction in the capacity of TRE1 to bind two
TR molecules when the orientation of the 3' site was reversed (Figs. 3A and 4A). This effect of position suggests that there is a small degree
of cooperativity of TR binding to these two sites, at least in
vitro, despite the fact that these octamers are separated by 10
bp. This is consistent with studies showing that mutation of either of
the TRE1 TR binding sites decreased TRE potency (6). The fact that only
a single site is occupied, but function remains normal, suggests that
the requirements for function and the behavior in the gel shift assay
are not dependent on the same rate-limiting phenomena. Of further
interest is the fact that it is TRE1 that confers the response of the
716-bp hdio1 promoter to RA. This suggests that for RA, as
well as for T3, widely separated receptor-binding
half-sites can confer a transcriptional response.
The results of these studies illustrate several novel variations on the theme of ligand-responsive receptor-binding elements, which have specifc relevance for the expression of hdio1. The presence of a TRE in an Alu sequence has not previously been described, but may not be uncommon, given the fact that one AGGTCA motif is highly conserved in these sequences. The examples of transcription factor-TR interaction, shown here and by others, indicate that interference or synergism with such factors is likely to be a common theme for the T3 response of other genes (6, 26, 30). Lastly, the fact that two widely separated TR binding sites, such as those in TRE1 of hdio1 and in the HIV1 LTR, and recent studies showing a marked amplification of the T3 response by insertion of a single TR binding site at a remote location from a classical TRE (31), suggest that the requirements for a physiologically significant TRE are less restrictive than is often supposed.
|
Footnotes |
|---|
Received August 18, 1997.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. V. Capuco, E. E. Connor, and D. L. Wood Regulation of Mammary Gland Sensitivity to Thyroid Hormones During the Transition from Pregnancy to Lactation Experimental Biology and Medicine, October 1, 2008; 233(10): 1309 - 1314. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Anguiano, N. Aranda, G. Delgado, and C. Aceves Epididymis Expresses the Highest 5'-Deiodinase Activity in the Male Reproductive System: Kinetic Characterization, Distribution, and Hormonal Regulation Endocrinology, August 1, 2008; 149(8): 4209 - 4217. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Dong, C. L. Yauk, A. Williams, A. Lee, G. R. Douglas, and M. G. Wade Hepatic Gene Expression Changes in Hypothyroid Juvenile Mice: Characterization of a Novel Negative Thyroid-Responsive Element Endocrinology, August 1, 2007; 148(8): 3932 - 3940. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Sharma, W. R. Hays, W. M. Wood, U. Pugazhenthi, D. L. St. Germain, A. C. Bianco, W. Krezel, P. Chambon, and B. R. Haugen Effects of Rexinoids on Thyrotrope Function and the Hypothalamic-Pituitary-Thyroid Axis Endocrinology, March 1, 2006; 147(3): 1438 - 1451. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S.-W. Kim, S.-J. Hong, K. M. Kim, S.-C. Ho, E. C. So, J. W. Harney, and P. R. Larsen A Novel Cell Type-Specific Mechanism for Thyroid Hormone-Dependent Negative Regulation of the Human Type 1 Deiodinase Gene Mol. Endocrinol., December 1, 2004; 18(12): 2924 - 2936. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. C. Bianco, D. Salvatore, B. Gereben, M. J. Berry, and P. R. Larsen Biochemistry, Cellular and Molecular Biology, and Physiological Roles of the Iodothyronine Selenodeiodinases Endocr. Rev., February 1, 2002; 23(1): 38 - 89. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. L. Amma, A. Campos-Barros, Z. Wang, B. Vennström, and D. Forrest Distinct Tissue-Specific Roles for Thyroid Hormone Receptors {beta} and {{alpha}}1 in Regulation of Type 1 Deiodinase Expression Mol. Endocrinol., March 1, 2001; 15(3): 467 - 475. [Abstract] [Full Text]
|
|
|
|
 |
|
 H. Kobori, M. Hayashi, and T. Saruta Thyroid Hormone Stimulates Renin Gene Expression Through the Thyroid Hormone Response Element Hypertension, January 1, 2001; 37(1): 99 - 104. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
