ABSTRACT
To assess potential species-specific expression of gonadotropin releasing hormone (GnRH), the distal
human (h) GnRH promoter was cloned, characterized and tested in gene transfer
studies. The nucleotide sequence of
~
3.8 kb of 5
'
-flanking region was determined. Homology to the rat (r) GnRH sequence was
observed in the proximal promoter region between -551 h (-424 r) and the transcriptional start site and within multiple
distal promoter regions. In contrast, there was little similarity in the
sequences between -1131/-551 h and -1031/-424 r. A deletion panel of 5
'
-flanking hGnRH promoter constructs was made and tested in transient
transfection assays in GnRH-producing mouse GT1-7 neuronal cells. The largest hGnRH promoter construct (-3832/+5 h) exhibited high levels of reporter activity,
similar to that observed with the largest rGnRH construct (-3026/+116 r). However, in contrast to the rat gene, deletion of distal
promoter sequences of the hGnRH promoter to -1971, -1131 or -551 did not result in a decrease in luciferase reporter
activity. Further truncation to -350 resulted in a 3-fold decrease in luciferase activity. There was no prefer- ential use of the putative upstream hGnRH start site in
neuronal cells. DNase I protection assays showed unique protection patterns
with nuclear extracts from GT1-7 and Gn10 neuronal cells and the hGnRH and rGnRH promoter fragments.
These data suggest the presence of different
cis
-acting elements and
trans
-acting factors that mediate species-specific neuronal GnRH expression.
Gonadotropin releasing hormone (GnRH) is a hypothalamic releasing factor,
expressed in a small subset of hypothalamic neurons, that controls pituitary
gonadotropin subunit biosynthesis and thus the reproductive axis (
1
). Although the gene for human (h) GnRH has been cloned by several groups (
2
-
5
), neither sequence information beyond -1131 nor functional characterization of the distal promoter beyond -551 is available. Based on the recent observations by Kepa
et al
. (
6
-
7
) and Mellon and co-workers (
8
-
9
) that the rat (r) GnRH promoter contains a neuronal-specific enhancer in the region between -1863 and -1571, we wondered whether the hGnRH promoter would be
similarly organized. Previous work by Radovick and colleagues (
4
,
10
) and Dong
et al
. (
5
) characterized the activity of the hGnRH promoter to -551 in Gn10 neuronal cells (
10
,
11
), derived from a mouse olfactory lobe tumor at the time of migration of GnRH
neurons, and JEG-3 choriocarcinoma cell lines (
4
,
5
). Since previous work characterizing the rGnRH promoter was performed in GT1-7 neuronal cells (
12
), derived from a mouse forebrain tumor post migration of the GnRH neurons, we
wished to compare the activity of the distal hGnRH and rGnRH promoters using
the same neuronal cell type and expression vector system.
We sequenced the distal hGnRH promoter using overlapping sequencing reactions. A
comparison of the two genes showed only partial similarity of the distal (-3800/-1500) and proximal (-350/+1) promoter regions, with marked differences in the
mid-promoter (-1500/-350) region. In gene transfer studies, the largest hGnRH
construct from -3832/+5 h was transcriptionally robust, exhibiting levels similar to that
observed with the largest rGnRH construct (-3026/+116 r). In contrast to the rat, the hGnRH promoter maintained high
levels of reporter activity with successive 5' promoter deletions to -551. Only with deletion to -350 did luciferase activity drop significantly. There was
no activity of hGnRH constructs containing only the upstream transcriptional
start site at -579 in the neuronal cells. In contrast to the ability of distal rat
promoter sequences to confer enhancer (5- to 7-fold) activity to a heterologous neutral promoter (RSV
180
), the distal human promoter sequences had little effect (1- to 2-fold). DNase I footprinting of the rGnRH and hGnRH promoters
suggested that different
cis
-acting elements bind nuclear proteins from GT1-7 and Gn10 neuronal cells. Thus, there appear to be marked
differences in the structural organization of the rGnRH and hGnRH promoters
that are reflected in a similar overall activity in neuronal cells, but a unique arrangement of
cis
-acting elements to direct species-specific and neuronal-specific expression of GnRH.
Sequence analysis was performed using the dideoxynucleotide chain termination
procedure with the Sequenase kit (US Biochemicals) (
13
). DNA fragments containing the distal hGnRH promoter (
Pst
I-
Hin
dIII) (
10
,
11
) were subcloned into pGEM vectors. Restriction sites used for localization of
synthetic oligonucleotide primers used in sequencing and subcloning are
indicated in Figures
1
and
2
respectively. Each DNA segment was sequenced and read from both strands of at
least two different reactions.
The promoterless vector (pA
3
LUC) and the pRSV
400
LUC vector were constructed as previously described (
6
,
14
). The pA
3
LUC vector contains a trimerized SV40 polyadenylation signal located upstream of
inserted promoter sequences that results in minimal background luciferase
activity (
15
). The largest fragment of the hGnRH construct was used to make a series of
deletion constructs using available restriction sites (
4
,
10
,
11
). The rGnRH promoter constructs were as previously described (
6
,
14
). The heterologous constructs were made using a 180 bp fragment (-130/+50) of the 3' long terminal repeat of the Rous sarcoma virus promoter inserted
downstream of GnRH promoter fragments (pRSV
180
LUC). pSV40[beta]gal was used as an internal control for transfection efficiency.
GT1-7 neuronal cells are an immortalized mouse hypothalamic cell line which
synthesizes and secretes abundant GnRH (
12
) and were kindly provided by P.Mellon (UCSD). Cells were grown in Dulbecco's
modified Eagle's medium (DMEM) and 5% fetal calf serum. Medium was supplemented
with 100 U/ml penicillin, 100 [mu]g/ml streptomycin and 0.25 [mu]g/[mu]l Fungizone. Transient transfections were performed using
electroporation as described (
6
). Ten micrograms of test plasmid and 5 [mu]g pSV40[beta]gal were transfected into GT1-7 cells and harvested after 16-18 h. Lysates were assayed for luciferase and [beta]-galactosidase to control for transfection
efficiency. Resultant luciferase activities were then normalized to the
activity of pRSV
400
LUC transfected in parallel to control for cell number across experiments as
described (
6
,
14
).
Nuclear extracts from GT1-7 and Gn10 neuronal cells were prepared as previously described (
16
). Protein concentration was determined using the BioRad kit, then aliquoted and
kept frozen at -80oC. DNA probes were prepared using restriction fragments of hGnRH (-551/-350 and -350/+5) and rGnRH (-2012/-1597 and -171/+116) promoters. Each
fragment was end-labeled with [[alpha]-
32
P]nucleotides (3000 Ci/mmol; NEN) and reverse transcriptase (
17
). Unincorporated nucleotides were removed using a Sephadex G-50 column. DNase I protection was performed using 1-2 [mu]l purified probe (0.5-1.0 ng or 5-10 000 c.p.m.), 400 ng ssDNA (2 [mu]l), 60 mM KCl, 240 [mu]g GT1-7 or Gn10 nuclear extract and 2.5-1000 ng DNase I. The samples
were run on a 6% acrylamide-8 M urea gel, followed by autoradiography (
17
).
Significant changes in promoter activity between constructs was determined by
one way analysis of variance (ANOVA) using the INSTAT program.
Previous analysis of the hGnRH promoter had included 5'-flanking sequence up to only -1131 (
3
). Thus, we first subcloned fragments of the largest DNA fragment available and
used a series of synthetic oligomers to further sequence the distal hGnRH
promoter (see Fig.
1
). Sequence analysis of overlapping subcloned fragments revealed that the
largest fragment contained sequence from -3832 to +5 of the hGnRH promoter (see Fig.
2
). Computer analysis of the human GnRH 5'-flanking sequence revealed significantly fewer convenient
restriction sites than in the rGnRH promoter. Scanning of the GenBank database
of reported consensus sequences for classes of DNA binding proteins revealed
two activator protein-1 (AP-1) sites at positions -2292 and -472, an imperfect estrogen response element (ERE) at
position -534 (
11
) and several putative DNA binding sites corresponding to members of the POU
domain (
18
) class of DNA binding proteins, including an octamer consensus at -3370. AP-1 consensus sites (
6
) were present in different regions than previously reported for the rGnRH
promoter at -99 (
6
).
To analyze potential similarities and differences in the sequences of the hGnRH
and rGnRH promoters, a sequence alignment (ALIGN) program was utilized.
Interestingly, there were three major regions of homology in the distal
promoter: -3036 to -2923 h with -2053 to -1940 r (region I); -2766 to -2539 h with -1786 to -1559 r (region II); -1775 to -1552 h
with -1311 and -1106 r (region III). Despite homology in the distal promoter
sequences, the putative repetitive sequences suggested by Whyte
et al
. (
8
) to be critical for neuronal-specific expression of the rGnRH promoter were not present in the human
sequence (Fig.
3
A). Between -1552 and -579, the putative upstream start site in the human promoter (
5
), there was little similarity with the rat promoter. However, in the proximal
promoter (region IV, -343 to +8 h and -332 to +96 r), there was marked homology similar to the proximal
mouse (m) GnRH promoter (see Fig.
3
B).
The largest hGnRH promoter fragment was placed into the pA
3
LUC vector and tested in a transient transfection assay optimized for rGnRH
promoter activity in GT1-7 neuronal cells. pA
3
LUC (promoterless vector) exhibited low activity (4-500 light units; LU). In contrast, pRSV
400
LUC, containing a neutral promoter, was dramatically expressed (7 * 10
6
LU). The -3832/+5 hGnRH construct was significantly expressed in the cells (53 000
LU), similarly to the largest rGnRH construct, -3026/+116 (80 000 LU).
A series of deletion constructs of the hGnRH promoter was made to assess the
cis
-acting regions that direct hGnRH gene expression in neuronal cells (Fig.
4
A). Deletion from -3832 to -1971 resulted in no change in luciferase activity (35 000 LU).
Further truncation to -1131 increased reporter activity slightly to 45 000 LU, which was
maintained with deletion to -551. Additional deletion to -350 decreased luciferase activity by 70% to 14 000 LU.
Interestingly, this region of the hGnRH promoter between -551 and -350 has little similarity to the rat gene. These results with the
hGnRH promoter were in marked contrast to results with the rGnRH promoter
constructs tested in parallel, containing a neuronal-specific enhancer region between -1863 and -1571 and multiple proximal regions that mediate neuronal
expression (see Fig.
4
B;
6
-
9
).
Since the distal promoter sequences of the hGnRH and rGnRH genes were
significantly homologous despite differences in their functional activities, we
asked whether distal hGnRH promoter fragments could confer enhancer activity on
a heterologous promoter. Various promoter fragments were inserted in front of
the neutral promoter RSV
180
and tested in gene transfer studies. The hGnRH fragments -3832/-2412, -2412/-3832, -2412/-2196 and -2196/-1971 of the hGnRH
promoter were able to enhance reporter activity only minimally, 1- to 2-fold above that seen with pRSV
180
LUC (Fig.
5
A). In contrast, the fragment -2012/-1597 of the rGnRH promoter increased the activity of the RSV
180
promoter 5- to 7-fold in an orientation-independent manner (Fig.
5
B). Together, these studies confirmed a species-specific difference in the structural organization of the distal GnRH
promoter.
To assess whether the differences in the functionally active
cis
-acting elements of the hGnRH and rGnRH promoters were accompanied by
species-specific variation in interactions with
trans
-acting nuclear proteins from neuronal cells, we performed DNase I
footprinting of DNA fragments shown to direct neural-specific promoter activity. Nuclear extracts from GT1-7 and Gn10 cells were incubated with -551/-350 and -350/+5 of the hGnRH promoter (Fig.
6
A and B) and -2012/-1597 and -117/+116 of the rGnRH promoter (Fig.
7
A and B). Extracts from both neuronal cells were tested to assess whether a
different complement of
trans
-acting factors was present in the neuronal cell line derived from a tumor
in the olfactory lobe during GnRH neuronal migration (Gn10), which produces low
levels of GnRH mRNA and protein, in comparison with a cell line from a
forebrain tumor post migration of GnRH neurons (GT1-7), which makes abundant mRNA and protein.
The hGnRH promoter fragment -551/-350 (Fig.
6
A) was protected in the region proximal to the upstream start site at -579. In Gn10 cells, this A-T region produced a potential footprint extending from -489 to -508, whereas GT1-7 extract protected a region from -479 to -535. Interestingly, this fragment
contains a putative hERE, an imperfect palindrome GGTATATAGTGTC at -534/-521, as compared with the consensus ERE GGTCANNNTGACC described by
Radovick and co-workers to be active in Gn10 cells (
10
).
The proximal promoter fragment -350/+5 h was heavily protected, but to a greater degree with GT1-7 than Gn10 extracts (Fig.
6
B). The CAAT motif at -162/-159 h and -53/-50 h as well as the TATA box at-32/-25 h was not footprinted with extracts
from Gn10 neuronal cells. These differences in the complement of active
transcription factors may explain the markedly different levels of endogenous
GnRH production in the GT1-7 compared with the Gn10 neuronal cells.
Although the hGnRH and rodent GnRH genes were cloned in the 1980s, little
information is available concerning a species comparison of promoter sequences
and functional activity in neuronal cells. Recent studies have characterized
the distal rGnRH promoter and shown the presence of a potent neural-specific enhancer between -1863 and -1571 (
6
-
9
). To see if similar regions exist in the human gene to mediate neuronal
expression, we sequenced >3 kb of the distal hGnRH promoter and tested its
activity in GT1-7 neuronal cells.
Sequence comparison showed few regions of significant homology in the distal
promoter regions. Despite sequence homology to the region of the rat enhancer
between -2000 and -1600, the human promoter sequence lacks the critical boxed repeats
suggested to be important by Mellon and co-workers for neuronal-specific expression (
8
,
9
) (see Fig.
3
A). The mid-promoter regions are dissimilar between the two GnRH genes. Unlike the rat
promoter, the human promoter contains a second upstream start site at -579, suggested to be important in GnRH expression in non-neuronal reproductive tissues (
5
). Perhaps upstream start site-specific control regions in the hGnRH promoter account for the divergent
mid-promoter region sequences in comparison with the rat gene.
In gene transfer studies in neuronal cells, the largest hGnRH construct, -3832/+5, exhibited activity similar to the largest rGnRH construct, -3026/+116. Unexpectedly, however, serial truncation of the distal
hGnRH promoter restored high reporter activity. This is in sharp contrast to
the pattern of activity of rGnRH deletion constructs. Only deletion to -350 of the hGnRH promoter resulted in a significant decrease in
luciferase activity. In contrast to the ability of the rGnRH promoter fragment -2012/-1597 to confer enhancer activity to a potent heterologous
promoter, various distal human promoter fragments had little effect on the RSV
180
minimal promoter. Together, these data suggest a species-specific difference in the structural organization of
cis
-acting promoter elements used to confer neural-specific expression.
DNase I footprinting with extracts from GT1-7 and Gn10 neuronal cells and proximal hGnRH and rGnRH promoter fragments
confirmed a different pattern of protein-DNA interactions. With the functional data, these studies localize the
neural-specific human GnRH enhancer to -535 and -479. Although gene transfer studies in Gn10 cells previously
defined a functional ERE at -534/-521 in the proximal hGnRH promoter (
10
), no clear protection of this region was observed with Gn10 extracts as was
documented with extracts from GT1-7 cells. The low abundance of estrogen receptor (ER) in Gn10 cells (
10
), the lack of optimal ligand concentration in the preparation of Gn10 extracts
and the difficulty of demonstrating DNase I protection by steroid receptors secondary to their rapid on/off rate may explain these
findings. The extended protected area produced with GT1-7 extracts most likely contains multiple nuclear proteins, however, none
are ER, since our subline of these cells do not contain ER (
7
).
The intense protection of functionally significant
cis
-acting elements of both the rat and human promoters with GT1-7 in comparison to Gn10 nuclear extracts may reflect the different
complement of active transcription factors that mediate the marked difference
in the level of endogenous GnRH expression between the two cell lines. Although
several protected areas were similar with the two neuronal extracts, unique
footprints support the potential different complements of
trans
-acting proteins expressed in the migratory versus post-migratory GnRH neurons. Together, these studies support the
hypothesis that neuronal proteins interact in a species-specific manner on the hGnRH and rGnRH promoter to ensure neural-specific expression.
Why there would be species-specific differences in the structural organization of the GnRH gene is
unclear. No sequence or functional analysis of the distal mouse GnRH promoter
is currently available to ask whether the differences we observed reflect a rat-specific or rodent-specific pattern of organization. Recent studies by Radovick and
colleagues have shown the ability of the DNA sequences -3832/+5 and -1131/+5, but not -484/+5, of the hGnRH promoter to target luciferase
expression to hypothalamic GnRH neurons in transgenic mice (
20
). Together with data from the functional assays and DNase I footprinting, these
data suggest that sequences between -535 and -479 contribute significantly to the expression of hGnRH
in vivo
and
in vitro
. The characterization of the complement of nuclear proteins binding to this
region will be the focus of future investigation.
The structural organization of other hypothalamic releasing hormone genes is
currently under active investigation. The corticotrophin releasing factor (CRF)
gene contains both proximal and distal transcription initiation sites with
tissue-specific differential utilization resulting in multiple mRNAs of variable
translational efficiency (
21
-
23
). The growth hormone releasing hormone (GHRH) gene has been shown to have two
distinct promoter regions to direct hypothalamus- versus placenta-specific expression (
24
). Cross-species comparisons in functional assays of hypothalamic releasing factor
promoters other than GnRH in neuronal cells, however, has not been possible due
to the lack of available model systems. Future studies will be necessary to see
if species-specific promoter organization exists within other hypothalamic releasing
hormone genes.
In summary, these studies identify the locus of neuronal-specific expression of the hGnRH promoter and provide a foundation to fine
map the differences in hGnRH and rGnRH promoter activity in neuronal tissues.
Future studies are needed to elucidate the ramifications of species-specific differences in the structural organization of the hGnRH and rGnRH
promoters.
We greatly acknowledge the secretarial support of Ms Gloria Smith. The medium
was provided by the Tissue Culture Core of the UCHSC Cancer Center (NCI P30
CA46934). The GT1-7 cells were provided by P.Mellon, UCSD. This work was supported by NIH
grant HD31191 (MEW), a Veterans Affairs Merit Review (MEW), NRSA (JKK) and HD-30040 (SR). Dr Wierman is a Clinical Investigator in Veterans Affairs. The
novel sequence reported in this paper has been deposited in the GenBank
sequence database, accession no. 456735.
REFERENCES
Return