| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ARTICLES |
Thyroid Division, Department of Medicine, Brigham and Women’s Hospital Harvard Medical School (S.-W.K., B.G., H.M.T., J.W.H., P.R.L.), Boston, Massachusetts 02115; the Department of Physiology and Biochemistry, University of Veterinary Science (T.B., P.R.), H-1400 Budapest, Hungary; and the Dipartimento Di Biologia Cellulare E Molecolare L. Califano, Universtta Degli Studi Di Napoli (D.S.), 80131 Naples, Italy
Address all correspondence and requests for reprints to: P. Reed Larsen, M.D., Thyroid Division, Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts 02115. E-mail: larsen{at}rascal.med.harvard.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
24
nt 5' to the ATG), an alternatively spliced approximately 300-nt intron
in the 5'-UTR, and three short open reading frames 5' to the initiator
ATG. The previously reported approximately 7.5-kb D2 messenger RNA
(mRNA) is actually an approximately 7-kb doublet that is present in
thyroid, pituitary, cardiac and skeletal muscle, and possibly brain,
but with only the longer transcript in placenta. A canonical cAMP
response element-binding protein-binding site is present at about 90 bp
5' to the most 5'-TSS. It accounts for the robust response of the
6.8-kb hdio2 5'-FR to protein kinase A. Forskolin
increases D2 mRNA in human thyroid cells, which may explain the high D2
mRNA in Graves’ thyroid and thyroid adenomas. The hdio2
gene structure and Northern blot results suggest that D2 expression is
tightly controlled and tissue specific. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In rats, activation of the sympathetic nervous system by cold exposure
increases D2 mRNA and activity in brown adipose tissue (BAT), an effect
that can be blocked by 
1- or ß-adrenergic
blocking agents (1, 6). These agents also increase D2 activity in
isolated brown adipocytes by an actinomycin D-sensitive mechanism (7).
D2 mRNA is also increased by forskolin or 8-bromo-cAMP in rat
astroglial cells (8), explaining the rapid cAMP stimulation of D2
activity in these cells (9). Taken together, these results suggest the
presence of a cAMP response element (CRE) in the rat as well as in the
human dio2 gene.
The goal of the present studies was to analyze the promoter,
5'-flanking and 5'-UTR of hdio2. We found that two major
transcripts (6.8 and 7.5 kb) are expressed in thyroid and muscle and
probably brain, with only a single 7.5-kb transcript present in
placenta. We identified three transcriptional start sites (TSS), 708,
32, and 25 nucleotides (nt) 5' to the initiator methionine, and an
alternatively spliced intron in the 5'-UTR. A single functional CRE is
present in the 6.5-kb 5'-flanking region about 70 bp 5' to the upstream
TATA sequence, accounting for the increased D2 mRNA expression in
Graves’ or TSH- or forskolin-stimulated human thyroid cells.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-Fix II vector,
Stratagene, La Jolla, CA) was screened by filter
hybridization with a probe from the 5'-portion of the human cDNA Z44085
(nt -40 to +302) (2) using standard techniques as previously described
(10). This clone was provided by D. St. Germain and V. Galton (1). Two
positive clones were mapped by Southern hybridization after digestion
with various restriction enzymes. One of these, clone 2, was plaque
purified, and the NotI fragment was subcloned into a
pBluescript plasmid (pBS, Stratagene) using standard
techniques (10). It was partially sequenced and extends approximately 7
kb 5' to the initiator ATG codon.
RNA isolation and Northern blotting
One to 2 g of several human thyroid glands obtained at
surgery were homogenized in liquid N2 in a
mortar. The homogenized tissues were transferred to a 15-ml tube and
lysed in Trizol (Life Technologies, Inc., Grand Island,
NY). Total RNA was isolated according to the manufacturer’s protocol.
Twenty micrograms of total RNA were denatured and electrophoresed on a
2.2 M formaldehyde-1% agarose gel in 1 x MOPS
buffer and transferred to nylon membranes (GeneScreen, DuPont,
Wilmington, DE) as previously described (4). We also purchased a
multiblot from CLONTECH Laboratories, Inc. (Palo Alto,
CA), containing 2 µg polyadenylated [poly(A)+)
mRNA from various human tissues.
Membranes were prehybridized in 5 x SSPE, 50% deionized
formamide, 5 x Denhardt’s solution, 1% SDS, 10% dextran
sulfate (mol wt, 500,000), and 100 µg/ml denatured sheared calf
thymus DNA at 42 C for 3 h. cDNA probes were labeled with
[- 32P]deoxy (d)-CTP and random primers
using Prime-It II kit (Stratagene). For detection of the
coding region, an hD2 coding region fragment (708–1531 nt 3' to the
TSS) was used. For hybridizations to identify the 5'-portion of the
5'-UTR, a PCR fragment containing nt 18–321 was used, and to identify
a putative 5'-intron, a PCR-generated fragment containing nt 352–645
was used. Hybridizations were performed with 1 x
106 cpm/ml hybridization mix of denatured labeled
cDNA probe at 42 C for 16–24 h in a rotating bottle of the
hybridization oven. Membranes were washed with 2 x SSC (standard
saline citrate)-0.1% SDS at room temperature, 1 x SSC-0.1% SDS
at 42 C and finally with 0.25 x SSC-0.1% SDS at 42 C.
The membrane was exposed to x-ray film for the indicated periods.
Primer extension
The 5'-end of the hdio2 gene transcript was
determined using a primer extension system according to the
manufacturer’s protocol (Promega Corp., Madison, WI). To
determine the 5'-end of the larger transcript, a 29-mer primer (16R)
complementary to nt 225–253 (see Fig. 2
for primer locations) was
5'-end labeled with [
- 32P]ATP and
T4 polynucleotide kinase (New England Biolabs, Inc., Beverly, MA). Total RNA from human Graves’
thyroid tissue was prepared using Trizol (Life Technologies, Inc.) and subsequently treated with ribonuclease-free
deoxyribonuclease to remove residual genomic DNA. One hundred
femtomoles of 32P-labeled primer and 20 µg RNA
were annealed in 1 x primer extension buffer [50
mM Tris-HCl (pH 7.5), 50 mM
KCl, 10 mM MgCl2,10
mM dithiothreitol (DTT), 1
mM dNTPs, and 0.5 mM
spermidine] at 58 C for 20 min and cooled at room temperature for
10 min. Primer extension was performed by adding 2.8
mM sodium pyrophosphate and 0.4 U/µl AMV RT in
1 x primer extension buffer at 42 C for 30 min. The
resulting product was analyzed on a 5% polyacrylamide-8.3
M urea gel in parallel with a sequencing reaction
of hD2 genomic DNA.
|
). PCR was performed with two
different annealing temperature protocols. First, we denatured at
94 C for 2 min and then proceeded with two cycles consisting of
denaturation at 94 C for 1 min, annealing at 55 C for 1.5
min, with amplification at 72 C for 2 min. Subsequently, 35 cycles
of denaturation at 94 C for 1 min, annealing at 63 C for 50
sec, and amplification at 72 C for 1.5 min were performed. Lastly,
extension was carried out at 72 C for 10 min. The PCR products
were subcloned into pGem T-vector and sequenced using the Fidelity DNA
sequencing system (Oncor, Inc., France). To explore the possibility of an alternatively spliced intron in the 5'-UTR, two RT reactions were performed using two different protocols. In the first, 1 µg total RNA was mixed with 0.4 µg random hexamers, 0.5 mM dNTPs, 40 U RNasin, 5 mM DTT, 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl2, and 200 U Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.) in a 20-µl reaction volume. This was incubated at 37 C for 1.5 h. A second RT reaction was performed with Superscript RNase H--reverse transcriptase (Life Technologies, Inc.) at 50 C. Ten micrograms of total RNA were mixed with 10 pmol specific primer of 4R and 9R, 62.5 mM Tris-HCl (pH 8.3), 93.75 mM KCl, and 3.75 mM MgCl2 in a 16-µl volume, which was preincubated at 65 C for 10 min and 62 C for 20 min, then cooled to room temperature for 10 min. Subsequently, 0.5 mM dNTPs, 40 U RNasin, 10 mM DTT, and 200 U Superscript RNase H--reverse transcriptase (Life Technologies, Inc.) were added, the final volume was made to 20 µl, and this was incubated at 50 C for 1.5 h. Subsequent PCR of both RT products was performed with 2 µl RT product using Vent polymerase according to the manufacturer’s protocol (New England Biolabs, Inc.). PCR mixtures were denatured at 94 C for 4 min and then subjected to 30–33 cycles, consisting of denaturation at 94 C for 1 min, annealing at 58 C for 1 min with amplification at 76 C for 1.5 min, and extension at 76 C for 10 min.
S1 analysis of hD2 mRNA
S1 analysis was performed using synthesized, single stranded DNA
probes as described previously (10). For determination of the second
and third transcription start sites, a cDNA probe complementary to the
RNA-coding strand, 716 5'-GGA TGC CCA TCT TCT CTG CCT CCT GAG TCA GTT
CCC TTG TGC GCT CTG GTT CCC TAG TG-3' 661 (TSS-2) was used. The 3' four
nucleotides were not complementary to the mRNA to permit S1 attack. For
the alternatively spliced intron in the 5'-UTR, a probe was designed to
detect the predicted 3'-intron boundary based on the RT-PCR results.
This primer was complementary to nt 695 5'-CCT GAG TCA GTT CCC TTG TGC
GCT CTG GTT CCC CTT CAC CCT CTT ATT TAA AAG GGC CCA-3' 640 and again
included four noncomplementary nucleotides (3'INT). Primers were 5'-end
labeled using T4 polynucleotide kinase. A mixture
of 2 pmol of each primer, 150 µCi
[-32P]ATP, 70 mM Tris-HCl (pH
7.6), 10 mM MgCl2, 5 mM
DTT, and 10 U T4 polynucleotide kinase in 25 µl
was incubated at 37 C for 1 h. The reactions were stopped by
heating at 75 C for 10 min and purified through a Sephadex G-25
column.
For the hybridization, 5 x 104 cpm probe were mixed with 50 µg total RNA, which was then precipitated using ethanol. After washing the pellet with 70% ethanol, the pellet was dried inverted in air for 30 min. Then, the pellet was dissolved in 20 µl S1 hybridization solution [80% deionized formamide, 40 mM PIPES (pH 6.4), 400 mM NaCl, and 1 mM EDTA (pH 8.0)], denatured at 65 C for 10 min, and subsequently hybridized at 30 C overnight. For the S1 reaction, the hybridized RNA was incubated in 300 µl 280 mM NaCl, 50 mM sodium acetate (pH 4.5), 4.5 mM ZnSO4, 20 µg/ml single stranded calf thymus DNA, and 300 U S1 nuclease at 30 C for 1 h. The reaction product was precipitated by adding 80 µl S1 stop buffer [4 M ammonium acetate, 20 mM EDTA (pH 8.0), and 40 µg/ml transfer RNA], and 1 ml ethanol. After washing and drying the pellet, it was dissolved in 3 µl TE buffer and mixed with 17 µl formamide loading dye. Ten microliters were loaded on a 6% polyacrylamide-urea gel to resolve the hybridized bands. This was subsequently exposed to BioMax x-ray film.
hD2 promoter-chloramphenicol acetyltransferase (CAT)
constructs
Genomic fragments were cloned into polylinker sites of the
pOCAT2 plasmid, a promoter insertion-CAT expression vector (11). Two
PstI sites were identified, 0.7 and 2.8 kb 5' to the
initiator ATG. The 3'-Pst site was at position +7 relative
to the 5'-TSS (see Results). BamHI sites were
identified about 0.4 and 4.3 kb 5' to the 5'-TSS (see Fig. 1). A 2.1-kb PstI fragment was
blunt ended and subcloned into the HincII site to make
hdio2CAT3. A partial HindIII (3') and
SacI (5') fragment was then inserted into the
SacI/HindIII sites of hdio2CAT3.
Restriction mapping showed that this construct was missing the two
3'-HindIII fragments (Figs. 1 and 2). To produce an intact
5'-flanking region construct, the internal (incomplete) 3.3-kb
BamHI fragment was deleted and replaced by the intact 3.9-kb
BamHI fragment cut from the genomic clone to form 6.5
hdio2CAT1. Several derivative constructs were prepared from
hD2CAT1, including an internal BglII deletion (6.5 hdio2
(/
BglII CAT); see Fig. 2),
which removes nt -491 to -84.
|
Transient expression assays
Transfections of HEK-293 cells were performed with CaP-DNA
precipitates using an internal TK human (h) GH control for transfection
efficiency as previously described (12). CAT assays were performed by
the method of Seed and Sheen (13), and results are expressed as the
ratio of cellular CAT to medium hGH (14). cAMP responsiveness was
tested by cotransfecting a plasmid that transiently expresses the a
catalytic subunit of protein kinase A (PKA) under control of the Rous
sarcoma virus promoter (provided by Dr. Richard Maurer) (15).
Human thyroid cell cultures
Cell cultures were prepared as previously described (16).
Briefly, tissue from a thyroid nodule of a patient undergoing
thyroidectomy for a toxic adenoma was cut into small pieces and
digested by type IV collagenase (Sigma, St. Louis, MO; 125
mg/ml) in Ham’s F-12 medium and 0.5% BSA overnight at 4 C in a
rotating shaker. Cells were pelleted by centrifugation at 150 x
g for 5 min, washed twice in BSA-Ham’s F-12 medium, seeded
in petri dishes, and cultured in 5% CO2
atmosphere at 37 C in Ham’s F-12 medium supplemented with 10% FCS.
Medium was changed every 3 days, and subconfluent plates were used for
treatment with forskolin (10 µM)
(Sigma).
Miscellaneous
All chemicals were of reagent grade and were obtained from
commercial sources. All molecular biological manipulations were
performed using standard techniques (10). Human tissues were obtained
under an institutional review board-approved protocol.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and 2). The sequence of an exon/intron junction at that
position was 5'-349 GAG/GT (Fig. 2B). The sequence of the
hdio2 gene from nt -650 to +1050 is shown in Fig. 2, together with the locations of introns, relevant restriction sites, and
oligonucleotides.
The upstream TSS in human thyroid mRNA was mapped using primer
extension with oligonucleotide 16R to a G residue 708 nt 5' to the
initiator methionine and 27 nt 3' to a TATA sequence (Fig. 3). However, Northern blots of mRNA from
thyroid and other tissues showed two distinct hybridizing bands,
suggesting that there was a second major TSS or an alternatively
processed transcript (Figs. 4 and 5). For example, a coding region probe
extending from position 708-1531 hybridized to a doublet of about 6.8
and 7.5 kb in the mRNA obtained from human thyroids and
glycoprotein-producing human pituitary tumors 1 and 2, but to only a
single 7.5-kb band in placenta and rat BAT mRNA (Fig. 4A). Inspection
of the 5'-UTR sequence showed several TATA or TATA-like sequences as
well as potential splice/donor and acceptor sites, suggesting the
possibility of an alternate TSS or an intron in the 5'-UTR (Fig. 2B).
We reprobed the blot with a PCR fragment derived from S2 + 12R (Fig. 2B), corresponding to nucleotides 18–321. This hybridized only to the
longest thyroidal mRNA, indicating the presence of an alternative TSS
3' to these sequences (Fig. 4B). This was confirmed by PCR analyses
using as template a 5'-RACE product that had been derived from a RT-PCR
reaction primed with oligonucleotide 9R (complementary to nt 1265–1288
in the coding region), poly(C)-tailed, and then amplified with the
abridged amplification primer and an antisense primer (oligo 4R in Fig. 2B), subcloned, and sequenced (see Materials and Methods).
The sequence indicated that the 5'-end of the shorter transcript was
nucleotide 676 (Fig. 2B). This TSS is 28 nt 3' to the TATA-equivalent
sequence, AATAA.
|
|
|
). The longest band (a) of 56 nucleotides represents a
completely hybridized transcript, and bands b and c (2 and 35 nt)
represent shorter transcripts. The 42-mer corresponds to the transcript
defined by the 5'-RACE at 676 nt, 29 nt 3' to a TATA box equivalent,
ATAA. The third TSS was approximately 7 nt 3' to this. These three
bands are found in three different samples of human thyroid RNA with a
ratio of about 8.8:1.8:1 by densitometry, suggesting that this is a
consistent finding in all thyroid tissues. By Northern analysis, the
two shorter transcripts would not be resolved at this position in the
gel, and this accounts for the appearance of only two bands. The
estimated ratio of the longer to the two shorter transcripts,
approximately 3:1, is consistent with the Northern results (Fig. 4A).
To examine the possibility of alternative splicing of the longer
transcript, we amplified the product of a RT-PCR reaction using the
sense oligo S2 and the antisense oligo 2R (Fig. 2B). The largest
amplicon was about 350 nt, and was subcloned and sequenced. The cDNA
was one in which the boxed 306-bp sequence in Fig. 2B was absent. This
was confirmed by amplifications using oligonucleotides S2 and 5R, which
showed the same 306-nt deletion, suggesting an alternatively spliced
intron in the 5'-UTR. Consensus splice donor and acceptor sites are
present in the appropriate positions in the sequence (Fig. 2B). Probing
the Northern blot with a PCR fragment corresponding to the putative
intron showed the same pattern of hybridizing bands as that in Fig. 4B
(data not shown), indicating that the putative intron sequences were
present only in the longer (7.5-kb) band. However, there was no clear
evidence of a transcript approximately 300 nt shorter than the
full-length 7.5-kb mRNA on blots probed with either the coding region
or nt 18–321 (Fig. 4A). This suggested the possibility that the RT
reaction had not faithfully transcribed the 306-bp sequence, perhaps
due to tertiary structure in this GC-rich region. Therefore, we
performed a repeat RT after preincubation at 65 C for 10 min (see
Materials and Methods). The product was
reamplified as described, but again only the 350-nt band was seen.
For independent confirmation of the presence of an alternatively
spliced intron at this position, S1 analysis with a primer
complementary to the 3'-boundary region of the putative intron was
performed. This confirmed the presence of two transcripts, one
containing the intron (band a in Fig. 5B) and another in which the
intron was spliced out (band b). The relative densities of bands a and
b indicate that most of the D2 transcripts contain the intron despite
its exclusion from the PCR-generated cDNAs.
To determine whether the pattern seen in thyroid and pituitary tumor
cells was typical of other human tissues, we probed a multi-tissue blot
with sequences containing the coding region (Fig. 6A) with nt 18–321 in the 5'-portion of the
coding region. The doublet at 7 kb was readily identified in heart and
weakly expressed in skeletal muscle (Fig. 6A), but only the shorter
band was seen in brain. With longer exposure, the same doublet bands
were also seen in kidney and pancreas, but not in liver or lung (Fig. 6A). However, only a single approximately 7.5-kb band was present in
placenta, in agreement with the result shown in Fig. 4A. With the
5'-UTR probe, the 7.5-kb band was seen in heart and placenta. In
addition, low intensity bands of about 5 kb were seen in all tissues
except liver and lung, and less distinct, smaller bands were seen in
heart and placenta (Fig. 6B). Probing with the intron fragment showed
the 7.5-kb band in heart and placenta, as well as a 2-kb band in
pancreas (data not shown). These results indicate that the hD2 coding
region is expressed in primary transcripts of 7.5 and 6.8 kb, and that
the 6.8-kb mRNA does not contain sequences 5' to nt 675. A third
transcript, the longer mRNA with the intron deleted, is not distinctly
identifiable in the Northern blots, except perhaps in brain (Fig. 6A).
These three hdio2 mRNA transcripts are depicted schematically in Fig 7.
|
|
).
|
). An hdio2
wild-type (633)CAT plasmid was not induced by a combination of phorbol
ester (3 x 10-7 M)
and A23187 (10-6-10-7
M), suggesting that this site may not be active
in the regulation of hdio2 gene expression. The
hdio2 gene constructs are cAMP responsive and discussed
subsequently.
The cAMP response element of the hdio2 gene
To confirm the function of the canonical CRE at -92 to -85 bp
(Fig. 2A) and to determine whether other CREs are present in the
5'-unsequenced portion of the hdio2 gene, the 6.5-kb
hdio2 5'-flanking region with and without a BglII
deletion and a 633-bp fragment with or without CRE mutations were
subcloned into the CAT expression vector pOCAT2 (Fig. 9). Transient expression of hdio2
wild-type (633)CAT was increased 3.6-fold by coexpression of the
catalytic subunit of PKA (Table 1).
Mutations in the CRE reduced the basal expression of the 633 5'-FR-CAT
construct and eliminated the response to PKA, indicating an important
role for this CRE in the regulation of expression of this gene. The
6.5-kb hdio2 FR-CAT construct was increased about 16-fold by
PKA. Deletion of the CRE on a BglII fragment lowered CAT
expression markedly and also eliminated the response to PKA, suggesting
that CRE at -90 is the only one present in the hdio2
5'-flanking region. To demonstrate that the endogenous hdio2
gene responds to cAMP, we prepared primary cultures of human thyroid
cells from a functioning human thyroid adenoma. Basal expression of D2
mRNA was barely detectable, but increased more than 10-fold within
3 h of exposure to 10 µM forskolin,
confirming the cAMP responsiveness of the endogenous hdio2
gene (Fig. 10).
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-glycoprotein-producing pituitary tumor cells, and is present in low
levels in kidney and pancreas. In this respect, the human D2 mRNA
resembles that identified in the rat central nervous system, in which a
doublet of about the same size was reported in various brain regions,
although not in BAT (1). This same doublet hybridizes to both the
coding region sequences and to a SECIS element-containing fragment from
the most 3' 300 nucleotides of the 3'-UTR region (3). In fact, primer
extension, PCR and S1 analyses indicate that there are four primary
transcripts in thyroid and possibly in several other tissues. These are
a full-length 7.5-kb mRNA, the predominant transcript; a 7.2-kb
transcript in which an approximately 300-nt intron is spliced out of
the full-length transcript; and two shorter transcripts using TSSs just
upstream of the ATG directed by a TATA equivalent (AATAA) sequence. All
four mRNAs can encode a full-length D2 selenoprotein. Expression of D2
mRNA in human kidney and pancreas has not been reported previously. We
also noted that 3'-fragments of the D2 cDNA containing the SECIS
element are found in expressed sequence tag sequences from
libraries prepared from prostate, breast, and uterus, which have also
not previously been shown to express D2 mRNA or activity (3).
Interestingly, placenta is the only human tissue in which only the
longer of the two major transcripts is expressed. However, in the
Northern blot shown in Fig. 6A, the longest brain mRNA is shorter than
that in placenta but longer than 6.7 kb, suggesting that it represents
the 7.5-kb transcript with the intron deleted. The presence of a
potential intron in this same general location was suggested by Celi
et al., although no sequence information was provided
(18).
Although the 7.5-kb transcript is the most highly expressed in human
thyroid, there is no systematic variation in the ratio of longer to
shorter transcript based on the type of thyroid abnormality. It might
be expected that thyroid from Graves’ patients or the TSH-stimulated
thyroid would have a higher ratio of the 7.5-kb to the 6.8-kb species
if the CRE has a greater effect on transcription from the 5'- than from
the 3'-TSS. This does not seem to be the case. The reason(s) for this
is not obvious, but could relate to differences in the half-lives of
the two species, recalling that the specimens were all obtained under
the constraints of surgical procedures and that the half-life of rD2
mRNA is short (2 h) (19). Furthermore, the levels of
thyroid-stimulating Ig were not measured in the Graves’ patients and
may vary considerably. There are marked variations in D2 activity
between Graves’ thyroids, suggesting considerable heterogeneity in
this population (see below) (4).
The 5'-UTR of the hD2 mRNA
The previously reported partial hD2 cDNA clone Z44085 has its
5'-terminus at nucleotide +553 (within the alternatively spliced
intron) (1, 2). The rat D2 5'-UTR (accession no. AB011068) is 74%
identical to the hdio2 cDNA between nt 437 and the ATG (1).
Another 257-nt fragment of the rat D2 partial cDNA (nt 1–257) has 73%
identity to nt 152–406 of the human mRNA. This fragment includes an
alternative splice donor sequence GG/GT in the identical location as
the hD2 AG/GT (nt 347–350), raising the possibility of a similar,
alternatively spliced intron in the rat dio2 gene. This or
an alternative TSS, such as observed in the human gene, could explain
the doublet rD2 mRNA band in the Northern blots from the central
nervous system reported previously (1). Recently, the sequence of a
portion of the human chromosome 14q24.3 containing the complete
sequence of the human dio2 gene has appeared in GenBank
(AC007372) and agrees with that in Fig. 2. Aside from this, the only
sequence in the GenBank database with significant (59%) similarity to
sequences 5' to nt 152 is the recently reported chicken D2 cDNA (AF
125575) (20). Interestingly, the splice donor site is also conserved in
that cDNA at the same position. However, only a single D2 transcript is
found in chicken Northern analyses (20).
Two short (3- and 4-codon) and one long (12- or 15-codon) open reading frames (ORFs) are present in the 5'-UTR. Interestingly, the position and length of the first two ORFs are conserved in the rat dio2 5'-UTR, with the amino acid sequence of the larger putative peptide starting at nt 307 (MKVELPVFNGL-UGA-QR-UAA) virtually identical to one in the rat 5'-UTR starting at nt 129. A potential ORF of 33 codons starting with the AUG at nt 464–6 is presumably not initiated, as the Kozak sequence is quite poor. The first two ORFs have favorable Kozak sequences, a purine at -3 in the absence of a G at +4, whereas the putative initiator methionine of the major ORF of hD2 has both the -3 purine and a G residue at +4 (21). In about 90% of eukaryotic mRNAs, the most 5'-AUG is used as the transcription initiation site. However, the exceptions to this rule, i.e. mRNAs that contain one or more upstream ORFs, encode a group of proteins with interesting functions, such as protooncogenes, transcription factors, DNA-binding proteins, and receptors for various hormones, including thyroid, estrogen, and retinoic acid (reviewed in Refs. 21, 22). In general, such AUG-burdened 5'-UTR sequences impair translation, suggesting that these encode proteins for which excessive production has adverse consequences (21). How such mRNAs are translated has not been resolved despite considerable investigation. Two models are proposed. One is the yeast GCN4 mRNA model, in which ribosomes can reinitiate at downstream codons depending on the distance between the upstream and downstream ORFs and the concentration of the initiator ternary complex (reviewed in Refs. 23, 24). The second model is that of the internal ribosomal entry site, as exemplified in the picornaviruses (reviewed in Ref. 22). Given the conservation of position and sequence of the sORFs between human and rat, it seems likely that the D2 mRNAs will follow the GCN4 model, but there are no well studied mammalian examples of this.
Of special interest for the SECIS element-containing D2 mRNAs is the fact that there is an in-frame UGA in the second sORF (although followed by an in-frame UAA 3 codons 3' when the intron is present) and that the third sORF terminates with UGA. There is an in-frame UAA for this ORF at nt 564–566, but the deduced protein from this ORF would be 33 residues in length. A similar phenomenon would occur in the rat mRNA with a highly similar deduced peptide (73% identity). Whether the SECIS element would suppress the stop codon function in these 5'-UTRs at this distance from the SECIS element is not clear. It has been shown that the choice between a UGA stop and a UGA selenocysteine codon is strongly influenced by the nucleotide following the UGA (the fourth base) (25). A purine in this position favors termination, whereas pyrimidine residues, present as the fourth base for each of the UGA codons in the hD2 5'-UTR as well as the UGA in the highly conserved catalytic center, favor selenocysteine insertion (25). Interestingly, the 3'-UGA in the carboxyl-terminus of D2 is followed by a purine consistent with its frequent function as a stop codon, as we have seen in transient expression assays (2). Further studies will be required to unravel this issue, which is made more intriguing by the fact that if the 3'-TSSs are used, none of the sORFs is included. Nontheless, human placenta and rat BAT express only the longer transcript, and D2 activity is clearly expressed in these tissues (26, 27).
The CRE and other transcription factor-binding sites in the hD2
promoter and 5'-flanking region
Analysis of the proximal 5'-flanking region indicates consensus
binding sites for two ligand-dependent transcription factors previously
reported to increase D2 activity in BAT or in rat glial cells, a CRE
and an AP1 site (9, 28, 29). Despite the presence of a consensus
AP1-binding site, there is no response of the 633-bp hdio2
CAT construct to phorbol esters with or without calcium ionophore. On
the other hand, the CRE confers a potent response to the 5'-TSS and is
located in a typical position within 100 nt of the upstream TATA box.
Mutation or deletion of the CRE eliminates the cAMP response of the
hdio2 gene. The lack of a significant residual response of
the BglII-deleted construct suggests that it is the only CRE
in the 6.5-kb 5'-flanking region. A response to cAMP of the
hdio2 promoter is consistent with the high D2 mRNA
expression of some human Graves’ or TSH-stimulated thyroid relative to
that of normal glands (4). We found D2 activity levels 30–100 times
higher than normal in two such thyroids (4). Furthermore, the presence
of high levels of D2 mRNA in patients with toxic adenomas due to
presumptive activating mutations in the TSH receptor or
Gs subunit could also be explained by this CRE
(30). Forskolin increased the D2 mRNA over 10-fold in human thyroid
cells in culture, establishing that the high expression of D2 mRNA in
Graves’ and TSH-stimulated thyroid tissues can be cAMP mediated.
Recently, TSH and/or cAMP were shown to increase D2 activity in
cultured human thyroid cells as well (31, 32).
In summary, the present results demonstrate a complex pattern of expression of the hdio2 gene in human thyroid and other tissues. The presence of a relatively long 5'-UTR with several sORFs suggests that translation of this mRNA is carefully modulated through as yet unclear mechanisms common to a small fraction of mammalian genes. However, use of the downstream TSSs would bypass these ORFs, possibly enhancing translation. These sORFs together with the recent results showing that rat D2 has a short half-life and is degraded in proteasomes suggest that the level of this enzyme is tightly controlled at both transcriptional and posttranscriptional levels (33). The presence of a potent CRE is consistent with the brisk physiological response of this promoter to cAMP. The availability of the 5'-FR of this gene and the mapping of the 5'-UTR will allow study of the mechanisms involved in the tissue-specific expression of this protein as well as the determination of whether its transcription is negatively regulated by T3 as is the rat dio2 gene (19). The CRE-containing hdio2 promoter could be another example of a promoter for which competition for CRE-binding protein and other basal transcription factors between T3-thyroid hormone receptor complex and the CRE-CRE-binding protein complex would cause negative regulation by T3 (34).
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 T.B. and S.-W.K. are equal first authors of this paper.
3 The promoter and 5'-flanking sequences have been
submitted to GenBank (AF188709).
Received August 18, 1999.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  F. S. Celi, G. Coppotelli, A. Chidakel, M. Kelly, B. A. Brillante, T. Shawker, N. Cherman, P. P. Feuillan, and M. T. Collins The Role of Type 1 and Type 2 5'-Deiodinase in the Pathophysiology of the 3,5,3'-Triiodothyronine Toxicosis of McCune-Albright Syndrome J. Clin. Endocrinol. Metab., June 1, 2008; 93(6): 2383 - 2389. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Canettieri, A. Franchi, M. D. Guardia, I. Morantte, M. G. Santaguida, J. W. Harney, P. R. Larsen, and M. Centanni Activation of Thyroid Hormone Is Transcriptionally Regulated by Epidermal Growth Factor in Human Placenta-Derived JEG3 Cells Endocrinology, February 1, 2008; 149(2): 695 - 702. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Dentice, C. Luongo, S. Huang, R. Ambrosio, A. Elefante, D. Mirebeau-Prunier, A. M. Zavacki, G. Fenzi, M. Grachtchouk, M. Hutchin, et al. Sonic hedgehog-induced type 3 deiodinase blocks thyroid hormone action enhancing proliferation of normal and malignant keratinocytes PNAS, September 4, 2007; 104(36): 14466 - 14471. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. Gumieniak, T. S. Perlstein, J. S. Williams, P. N. Hopkins, N. J. Brown, B. A. Raby, and G. H. Williams Ala92 Type 2 Deiodinase Allele Increases Risk for the Development of Hypertension Hypertension, March 1, 2007; 49(3): 461 - 466. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Grarup, M. K. Andersen, C. H. Andreasen, A. Albrechtsen, K. Borch-Johnsen, T. Jorgensen, J. Auwerx, O. Schmitz, T. Hansen, and O. Pedersen Studies of the Common DIO2 Thr92Ala Polymorphism and Metabolic Phenotypes in 7342 Danish White Subjects J. Clin. Endocrinol. Metab., January 1, 2007; 92(1): 363 - 366. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. H. A. Kester, G. G. J. M. Kuiper, R. Versteeg, and T. J. Visser Regulation of Type III Iodothyronine Deiodinase Expression in Human Cell Lines Endocrinology, December 1, 2006; 147(12): 5845 - 5854. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Kohrle, F. Jakob, B. Contempre, and J. E. Dumont Selenium, the Thyroid, and the Endocrine System Endocr. Rev., December 1, 2005; 26(7): 944 - 984. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Marrif, A. Schifman, Z. Stepanyan, M.-A. Gillis, A. Calderone, R. E. Weiss, J. Samarut, and J. E. Silva Temperature Homeostasis in Transgenic Mice Lacking Thyroid Hormone Receptor-{alpha} Gene Products Endocrinology, July 1, 2005; 146(7): 2872 - 2884. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. P. Peeters, A. W. van den Beld, H. Attalki, H. v. Toor, Y. B. de Rijke, G. G. J. M. Kuiper, S. W. J. Lamberts, J. A. M. J. L. Janssen, A. G. Uitterlinden, and T. J. Visser A new polymorphism in the type II deiodinase gene is associated with circulating thyroid hormone parameters Am J Physiol Endocrinol Metab, July 1, 2005; 289(1): E75 - E81. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. H. Canani, C. Capp, J. M. Dora, E. L. S. Meyer, M. S. Wagner, J. W. Harney, P. R. Larsen, J. L. Gross, A. C. Bianco, and A. L. Maia The Type 2 Deiodinase A/G (Thr92Ala) Polymorphism Is Associated with Decreased Enzyme Velocity and Increased Insulin Resistance in Patients with Type 2 Diabetes Mellitus J. Clin. Endocrinol. Metab., June 1, 2005; 90(6): 3472 - 3478. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Morimura, K. Tsunekawa, T. Kasahara, K. Seki, T. Ogiwara, M. Mori, and M. Murakami Expression of Type 2 Iodothyronine Deiodinase in Human Osteoblast Is Stimulated by Thyrotropin Endocrinology, April 1, 2005; 146(4): 2077 - 2084. [Abstract] [Full Text |
