| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Laboratoire INSERM U376 Hôpital Arnaud de Villeneuve 34295 Montpellier cedex 05, France
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
During the last decade, we have extensively studied the rat serine protease inhibitor (spi) gene system, which comprises three members (8) that are differentially expressed, depending on the physiological situation. Two of them, spi 2.1 and 2.2, are maximally expressed in normal animals and are down-regulated during acute inflammation (9). In contrast, the third one, spi 2.3, is virtually silent in control animals and is transiently induced during the acute-phase response (9). Regulation of expression of the spi 2.3 gene is complex. It involves positive promoter-regulatory elements including several interleukin-6-response sites and a glucocorticoid (GC)-response element (GRE) (10), and a negative regulatory element located in the 3'-untranslated region (11), which most likely keeps the gene silent in normal rats. Two types of effectors, GH and GC, appear to be responsible for the high level of spi 2.1 (and 2.2) gene expression in normal animals (12). We (4, 13) and others (5) have characterized a distal GH-response element (GHRE-II) in the spi 2.1 promoter which, in cultured hepatocytes, accounts for approximately 50% of the GH response (4). This element contains two potential binding sites for STAT-5 (5), a protein that belongs to a class of transcription factors activated upon tyrosine phosphorylation by a GH-activated Janus kinase (Jak2) (for a review, see Refs. 14 and 15). We have recently identified the second spi 2.1 promoter GH-response element (GHRE-I). This element also referred to as the GAGA box is structurally unrelated to STAT-binding sites (16) and behaves, at least in cultured hepatocytes, as a bifunctional enhancer controlling basal transcription and responsible for part of the GH response (4, 16). In addition to these GHREs, the spi 2.1 promoter also contains a GRE, which, in vitro, binds specifically GR receptor-GC complexes (GR-GC) and entirely mediates GC response, and several CCAAT/enhancer-binding protein (C/EBP)-binding sites that contribute to basal (i.e. hormone-independent) promoter activity (4, 13).
Specific binding of transcription factors (i.e. C/EBP, STATs, GR-GC complexes and GAGA box-binding proteins, GAGA-BPs) to their cognate sites in the spi 2. 1 promoter has been demonstrated by EMSA and in vitro footprinting (4, 16, 17). However, the physiological role of such factors remains questionable since identical in vitro footprinting patterns were obtained with liver nuclear extracts from control, hypophysectomized (hypox), or inflamed rats (17), which correspond to high, negligible, and intermediate levels of spi 2.1 gene expression, respectively (13). To address this issue, we have used genomic footprinting to examine changes in site occupancy occurring in vivo within the proximal spi 2.1 promoter region as a function of the level of gene transcription. Concomitantly, we have analyzed deoxyribonuclease I (DNase I) sensitivity of the promoter region to evaluate structural chromatin alterations that take place during gene activation. The data suggest that transcription of the spi 2.1 gene in vivo strictly correlates with occupancy of the GAGA box presumably triggered by or associated to a GH-dependent chromatin remodeling event. In addition, our studies largely unravel the mechanism of down-regulation of the spi 2.1 (and 2.2) gene that takes place during acute inflammation.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
shows that such a
requirement is met with the spi 2.1 gene. The high level of
transcription observed in control animals, in which it exceeds that of
the albumin gene, was decreased by more than 95% upon hypophysectomy,
which, in contrast, did not affect transcriptional activity of albumin
and ß-actin genes. In acutely inflamed rats, spi 2.1 gene
transcription was also severely depressed. The 5- to 7-fold decrease in
signal intensity observed in the run-on assay, in fact, underestimates
the inflammation-mediated decrease in spi 2.1 gene
transcription. Indeed, because of the extensive degree of sequence
identity (more than 95%) between the two cDNAs in their 5' and central
regions (9), the spi 2.1 probe used also recognized the
spi 2.3 transcript, which was essentially undetectable in
control animals but is induced during inflammation (Refs. 9 and 10; see
also Fig. 7). As previously observed, acute inflammation negatively
affected albumin gene transcription but had no effect on the expression
of the ß-actin gene (10).
|
|
shows that digestion of genomic DNA
with BamHI generated a 2.6-kb fragment (panel A) spanning
the -862/+1725 gene region (panel B). DNase I treatment of liver
nuclei from normal rats led to the appearance of three minor fragments
of approximately 0.8, 0.75, and 0.67 kb that were not detected in DNA
isolated from hepatic nuclei of hypox animals. This revealed the
presence of three DNase I-hypersensitive sites (HSS I, II, and III)
that were seen only when the gene is efficiently transcribed and that
map in the proximal promoter region containing the two GHREs previously
identified by functional in vitro studies (4, 5, 13). This
is consistent with the observation made by Yoon et al. (19),
who first reported a GH-dependent induction of DNase I-hypersensitive
sites in the 5'-flanking spi 2.1 gene region. A similar
analysis has also been performed with liver nuclei from inflamed rats,
yielding results comparable to those obtained with nuclei from hypox
animals (data not shown).
|
suggested that chromatin
remodeling occurred in the spi 2.1 promoter region, when
switching from a situation of very low level (i.e. hypox
rats) to a situation of very high level (i.e. control
animals) of gene transcription. However, they did not provide any
information on the relationship that may exist between the degree of
occupancy of putative functional sites within the promoter in
vivo and the level of gene transcription. Our previous in
vitro binding studies (i.e. EMSA and footprinting
experiments) have not allowed us to correlate changes in the level of
spi 2.1 gene transcription in vivo with changes
in DNA-protein interactions more particularly in the region of the GAGA
box (17) which proved to be the major spi 2.1 promoter
regulatory element in isolated hepatocytes (4, 16). We thus turned to
the genomic footprinting technique that allows a precise analysis of
nuclear protein-DNA interactions in vivo. Figure 3 displays the results of an analysis of
the lower strand of the spi 2.1 promoter region from
positions -160 to +10 relative to the cap site. This region
was previously found to contain the elements required for maximal basal
as well as hormone-stimulated promoter activity in cultured hepatocytes
(4, 13). A comparison of patterns of bands obtained with naked DNA
(in vitro) and with the three DNase I-treated nuclei-derived
templates (in vivo) revealed major differences throughout
the promoter. In control rats, a unique region encompassing the GAGA
box and presumably extending toward the 3'-region, as suggested by the
slight decrease in intensity of the -37/38 and -26/-27 bands, was
strongly protected against DNase I attack. In contrast, essentially no
protection of the GAGA element occurred in hypox or acutely inflamed
rats. It should also be noted that hypophysectomy and, to a lesser
extent, inflammation induced a strong DNase I-reactive site at
positions -37/-38 (presumably a doublet) and weaker sites at
positions -26/-27 and +10. Unlike what was observed with the GAGA
box, no DNase I reactivity of the adjacent C/EBP site spanning the -66
to -58 region was detected, regardless of the situation analyzed. This
obviously precludes drawing any conclusion on the in vivo
role of this site shown to be functional in vitro (4, 13).
|
)
which, however, displayed marked differences in DNase I reactivity as a
function of the level of gene transcription. Thus, a strong reactive
site occurring at position -86 within the GRE undetected with a naked
template (i.e. in vitro) and a minor one flanking
this motif occurring at position -92 or -93 (or possibly both) also
seen in vitro became apparent in control animals but almost
totally disappeared upon hypophysectomy or inflammation. In contrast,
enhanced DNase I reactivity was detected at positions -90, -95, -96,
and -98 (i.e. 5' to the GRE) in nuclei from hypox and, to a
lesser extent, inflamed animals, compared with both in vivo
controls and naked DNA. Finally, the fairly strong nuclease reactive
site found at position -80 in DNase I-treated naked DNA and in DNA
isolated from control nuclei digested with the enzyme was significantly
attenuated in hypox and acutely inflamed rat nuclei.
The second region differentially affected in both
pathophysiological models relative to the control encompasses the
spi GHRE-II (4) and includes the potential STAT-5 binding
sites present at positions -124 to -132 and -141 to -147. In
addition to other minor differences that proved poorly reproducible,
two major observations were consistently made: 1) unlike what was
observed with the GAGA box, no clear protection of any of the
STAT-binding sites could be detected, whatever the state of the animal,
by comparing in vitro and in vivo footprint
patterns; 2) in the in vivo patterns, several DNase
I-hyperreactive sites (positions -126, -128, and -137) were visible
with hypox and, although much less pronounced (positions -126 and
-128), inflamed rat liver nuclei-derived templates but not with
controls. In addition, two minor nuclease-reactive sites (positions
-118 and -120) that were seen with both a naked template and DNA
prepared from DNase I-digested control nuclei became barely detectable
upon hypophysectomy or inflammation. Figure 4 shows a similar analysis performed with
the upper strand. The most striking observation made with naked DNA as
well as with templates derived from DNase-treated nuclei was the almost
complete lack of DNase I reactivity of the GAGA box region and, as
already noted with the lower strand, of its 5'-flanking C/EBP site. As
observed with the lower strand, no clear protection but rather
significant changes in DNase I reactivity of the promoter were detected
in the GRE region. Thus, a set of hyperreactive sites internal to the
GRE (positions -82 and -77 to -72) that were not observed with naked
DNA but could be seen with templates prepared from control nuclei, were
undetectable except for two of them (positions -74 and -73), in
templates prepared from nuclei of hypox or inflamed rats. In addition,
two other minor changes including the appearance of a weak reactive
site at position -84 and attenuation of stronger ones at positions
-92 and -91 were also detected in this region, with in
vivo hypox and inflamed rat-derived templates compared with
controls or in vitro DNase I-treated DNA.
|
) and specifically seen with hypox and
inflamed rat-derived templates was also observed. Several other
nuclease-reactive sites were induced in the two pathophysiological
animal models examined, both in this distal region of the
spi 2.1 promoter (positions -165, -106, and -104) and in
a region closer to the cap site (positions -34 to -32). It
is also remarkable that the cap site region (positions +1 to
-15) displayed enhanced DNase I reactivity in all in vivo
conditions analyzed, compared with naked DNA.
Analysis of GAGA-BPs and STAT-5 Contents of Liver Nuclei
by EMSA
The almost complete absence of detectable in
vivo interactions between GAGA-BPs and the GAGA box in hypox and
acutely inflamed animals revealed by genomic footprinting could be
merely due to a decrease in amounts of active GAGA-BPs in nuclei
preparations. To test this hypothesis, we made extracts from liver
nuclei prepared in the same way as those used for DNase I treatment and
analyzed by EMSA the capacity of nuclear proteins to bind to the GAGA
box. Figure 5A shows that no significant
qualitative or quantitative difference could be detected between
control, hypox, and inflamed rat liver nuclear extract preparations. As
already observed with control extracts (16), formation of a major (a)
and a minor (b) complex, previously shown to represent specific binding
of nuclear factors to the spi GAGA box, occurred with all
three different sets of nuclear proteins.
|
does not support this hypothesis. Indeed, significant amounts
of STAT-5 present in control nuclear extracts bind to a specific STAT-5
probe derived from the ß-casein gene promoter. In addition to STAT-5,
C/EBP factors massively represented in liver nuclei have also been
shown to bind the spi 2.1 GHRE-II in vitro (17).
However, based on our genomic footprinting data, they do not appear to
form detectable stable complexes with this promoter region in
vivo.
In Vivo Effect of Lipopolysaccharides (LPS)
and in Vitro Effect of Cytokines on GH-Dependent or
Independent spi 2 mRNA Levels
An important observation made in the genomic
footprinting experiments was that the lack of protection of the GAGA
box was associated with a very strong down-regulation of the
spi 2.1 gene in inflamed animals. Since GH appears to
strictly control spi 2.1 GAGA box occupancy in
vivo (see Discussion), a perturbation occurring in the
GH transduction pathway subsequent to the development of a GH-resistant
state could account for this observation. In keeping with this
possibility, some of us recently showed that treatment of primary
hepatocytes with interleukin (IL)-1ß or tumor necrosis factor-
(TNF) (23), which are two major lymphokines implicated in the liver
acute-phase response (for a review, see Refs. 24 and 25), or injection
of inflammatory compounds (LPS) to living animals (26) strongly
inhibited GH-dependent insulin-like growth factor-I (IGF-I) message
synthesis. It thus became important to check whether this type of
effect could represent a more general phenomon and apply to the
spi 2 gene system. To this aim, we investigated the effect
of LPS (in vivo) or cytokines (in vitro) on GH
action on the level of spi 2 messages. As previously
observed (9), acute inflammation reduced by about 50% spi
2.1 and 2.2 mRNA liver steady-state levels and, inversely, induced the
spi 2.3 message (Fig. 6A, compare lanes 1–3 to lanes 4–6; and Fig. 6B). Injection of GH alone
slightly increased the level of spi 2.1 and 2.2 messages and
had no effect on that of spi 2.3 (Fig. 6, A and B).
Interestingly, the inflammatory drug largely prevented GH to restore
normal spi 2.1 and 2.2 mRNA levels (Fig. 6A, lanes 10–12;
and Fig. 6B).
|
, lane 1). This level could be
increased, however, by GH by 2- to 3-fold (lane 2), an effect that was
almost completely obliterated by IL-1ß (lanes 3 and 4), TNF (lanes
7 and 8), or a combination of both agents (lanes 11 and 12). In
contrast, IL-6 did not appear to signicantly influence GH action on
spi 2.1 and 2.2 messages but, as expected (10), strongly
induced the spi 2.3 mRNA (lanes 5 and 6). The latter, which
was already slightly expressed under unstimulated conditions, was also
increased by IL-1ß (compare lanes 1 and 3) but not TNF (compare
lanes 1 and 7). This effect of IL-1ß, which was not previously
observed by using a different culture system [i.e. plastic
dishes and a serum-containing medium (10) instead of matrigel-coated
dishes and a serum-free medium, this study], can presumably be
accounted for by the presence of a nuclear factor 
B (NF-B) site
within the spi 2.3 promoter (8, 10). IL-1 has indeed been
shown to activate NF-B (for a review, see Ref.27), which, in turn,
stimulates transcription of specific genes such as the nitric oxide
synthase gene in pancreatic ß-cells (28). To further characterize the
antagonistic action of IL-1ß on the GH effect, we performed
dose-dependent studies. Induction of spi 2.1 and 2.2 mRNA
species by GH in cultured hepatocytes was inhibited in a dose-dependent
manner by IL-1ß, with a maximal effect (nearly 100% inhibition)
occurring with 10 ng/ml (Fig. 8, panel
A). A similar relationship was also observed for induction of the
spi 2.3 mRNA. Production of spi 2.1 and 2.2
messages proved highly sensitive to GH since a concentration of 100
ng/ml (
5 nM) was maximally efficient (Fig. 8, panel B).
However, regardless of the GH concentration used, the effect was
totally inhibited by IL-1ß (Fig. 8, panel B). Thus, data obtained by
analyzing IGF-I (23, 26) and spi 2 messages (present study)
strongly support the notion that acute inflammation, through the action
of IL-1ß (and possibly TNF), causes the appearance of a strong
GH-resistant state in the hepatocyte.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
The most striking observation arising from genomic footprinting analysis of the spi 2.1 promoter is the strict correlation existing between the stable interaction of nuclear proteins with the proximal GHRE (GHRE-I) recently identified as the GAGA box (16), and the level of gene transcription in vivo. Such a correlation fully agrees with the observation that, in cultured hepatocytes, a mutation of the GAGA box that is lethal for both basal and hormone (i.e. GH and GC)-dependent promoter activities (4) also abolishes binding of as yet unidentified GAGA-BPs (16). This also further supports the notion that the GAGA box, a purine-rich promoter motif fully conserved in all three spi 2 genes (8), is a key control element working both in vivo and in vitro as an on/off transcriptional switch.
Clearly, transcriptional activation of the spi 2.1 gene
in vivo is strictly GH-dependent, since only GH can
reinduce gene expression in hypox rats (12). This strongly
suggests that proper assembly of GAGA-BPs-GAGA box complexes,
which is mandatory for spi 2.1 gene promoter activity
in vivo, requires that the signal initiated by GH binding to
its cell surface receptors is normally processed to the nuclei. Thus,
in any situation where a severe impairment in the GH-signaling pathway
occurs, formation of GAGA-BPs-GAGA box complexes should be prevented
and spi 2.1 gene transcription shut off. This is the case,
as expected, in hypox rats due to the absence of GH, but also in
acutely inflamed animals. Indeed, lipopolysaccharides (in
vivo) and inflammatory cytokines (IL-1ß and TNF, ex
vivo) induce a strong GH-resistant state in hepatocytes,
presumably leading to a blockade of GAGA box-GAGA BPs complex formation
and subsequent inactivation of the spi 2.1 gene. Cytokines
have generally been found to directly regulate transcription of
acute-phase genes via their specific promoter response elements (24, 25). Our data reveal that cytokines can also regulate acute-phase gene
expression indirectly, by disturbing the functioning of a key promoter
element such as the spi GAGA box.
How GH contributes to bringing GAGA-BPs to the promoter, however,
remains an open question. Experiments performed with cell-free systems
strongly suggest that binding of GAGA-BPs (see Fig. 5A) and subsequent
transcriptional activation (16) can occur independently of GH. This
seems to indicate that only the recruitment of GAGA-BPs, presumably
linked to the chromatin-remodeling event shown to take place during
spi 2.1 gene activation, requires GH in vivo. At
least two hypotheses can be put forward: 1) GH might first activate a
chromatin-remodeling system independent of GAGA-BPs that would locally
alter the nucleosomal structure and permit binding of those
transcription factors to the promoter; 2) GAGA-BPs themselves could
bear a chromatin-remodeling activity stimulated by GH (e.g.
through posttranslational modifications). In keeping with the latter
possibility, trans-displacement of histones has been
proposed as one mechanism for trancription factor-targeted generation
of a nucleosome-free region in chromatin (29). Assessing these
hypotheses must await the cloning and characterization of GAGA-BPs that
should allow setting up reconstitution experiments.
Unlike what was observed with the spi 2.1 GHRE-I (i.e. the GAGA box), we never detected any protection of the STAT-5-binding sites contained in the GHRE-II in vivo. This was true in control animals, despite the fact that a significant level of active nuclear STAT-5 could be demonstrated, and also in hypox rats treated for 2 h with GH, a condition under which the nuclei content of active STAT-5 exceeds that in controls (C. Legraverend, unpublished results). This, however, does not eliminate the possibility that STAT-5 can bind to the GHRE-II in living animals. It is indeed conceivable that activation of STAT-5 by GH could lead to a transient interaction with its cognate binding sites that would not be detected by the genomic footprinting technique but that could subsequently modify the local chromatin structure and contribute to transcriptional activation. Such a structural modification appears to take place upon gene activation, at least at the level of the more proximal STAT-5-binding site, as evidenced by the decrease in local DNase I reactivity detected with control templates. However, an alternative explanation to such a change in DNase I reactivity that was also observed in other parts of the promoter when switching from a transcriptionally inactive (i.e. hypox or inflamed rats) to a transcriptionally active (control animals) state could be the displacement or the structural alteration of a nucleosome. Thus, our genomic footprinting data do not provide a clear answer to the question of whether STAT-5 can form a complex with the spi 2 GHRE-II in vivo. Whatever the answer to this question might be, however, it should be kept in mind that STAT-5 could contribute to transcriptional activation of the spi 2.1 promoter without directly contacting the DNA. This transcription factor has indeed been shown to associate with GR receptors and potentiate their effect in vitro (30).
The absence of detectable C/EBP footprints, whatever the in
vivo situation analyzed, appears at first to be totally
paradoxical, based on our previous in vitro footprinting
assays (4, 17) and heterologous transactivation experiments (11).
Unlike what is happening in vitro (see EMSA in Fig. 8), the
results of genomic footprinting clearly indicate that there is no
stable interaction of C/EBP with the -130 to -122 site internal to
the GHRE-II, in vivo. As pointed out for STAT-5, however, we
cannot formally rule out that transient and/or very loose binding of
C/EBP might occur in the context of intact chromatin and confer some
functionality to this site in vivo. Due to the lack of DNase
I sensitivity of the other potentially important C/EBP-binding site
immediately flanking the GAGA box (-66/-58), it is not possible to
exclude this site as being functional in vivo.
Unlike what has been observed in vitro (4), no protection of the spi 2.1 promoter GRE could be observed in vivo by genomic footprinting, either with control rats having normal circulating GC levels or with inflamed animals known to overproduce GCs (24, 25). However, a differential alteration in DNase I reactivity of the GRE region was observed as a function of the level of gene transcription by the technique of in vivo footprinting. The most striking change is represented by the presence of several DNase I-hyperreactive sites mapping within the GRE or in its close vicinity in chromatin from control liver nuclei and their absence in naked DNA and in chromatin from nuclei of hypox or inflamed rats. Similar changes in reactivity to DNase I occurring upon gene activation mediated by GCs have been observed with the rat tyrosine aminotransferase promoter (21) and the mouse mammary tumor virus long terminal repeat (22). They have been interpreted as the signature of a transient association of GR-GC complexes with GREs. Thus, the possibility seems to exist that those hypersensitive sites specifically detected in the GRE region in an actively transcribing state (i.e. control animals) are, at least in part, the consequence of an interaction with GR-GC complexes.
The genomic footprinting patterns of bands generated within the GRE region by DNase I treatment of nuclei isolated from hypox and inflamed rat liver clearly differed from those obtained with control nuclei but, surprisingly, were qualitatively similar. This was rather unexpected, considering that hypophysectomy and acute inflammation represent two opposed situations with respect to endogenous levels of circulating GCs (i.e. almost total absence of hormone in hypox rats vs. enhanced levels in inflamed rats). This makes it unlikely that GR-GCs complexes mediate changes occurring in DNase I reactivity in the spi 2.1 gene GRE region in either of these animal models compared with controls. Those alterations could more likely be related to an overall modification of the nucleosomal structure occurring in the promoter upon gene activation and revealed by the appearance of DNase I-hypersensitive sites. Further studies involving in vitro reconstitution of the nucleosomal structure will be necessary, however, to assess this hypothesis.
In conclusion, we would like to propose a mechanism for spi 2.1 gene activation in vivo that includes first, a GH-dependent remodeling of the chromatin in the promoter region caused by, or concomitant to, the recruitment of GAGA-BPs. This would then allow association of GR-GC complexes with the GRE, which in turn could boost GAGA-BP binding through a dynamic interplay, thus accounting for the potentiating effect of GCs on GH action. At this point, the contribution of STAT-5 to GH-mediated spi 2.1 gene activation in vivo remains to be clarified. It should be pointed out, however, that unlike activation of STAT-5, which appears to be sex-dependent (31), functioning of the GAGA box as well as expression of spi 2.1 and 2.2 genes are essentially identical in the male and the female (our unpublished observations). This strongly argues for a predominant role of the GAGA box in controlling GH-dependent regulation of spi 2 gene expression in vivo. Finally, a very important outcome of these studies is the elucidation of the indirect mechanism of cytokine-mediated down-regulation that accounts for the transcriptional shut-off of spi 2.1 and 2.2 gene expression induced by acute inflammation.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-32P]ATP (3000 Ci/mmol) and
[-32P]UTP (800 Ci/mmol) were obtained from Du Pont NEN
(Paris, France). Vent Exo- and Amplitaq DNA polymerases
were purchased from New England Biolabs (Beverly, MA) and Perkin Elmer
(Paris, France), respectively, and DNase I (3000 U/mg) was from
Worthington (Freehold, NJ). Collagenase (0.65 U/mg) was from Boehringer
Mannheim (Mannheim, Germany) and bacterial (Escherichia
coli) LPS were obtained from Life Technology (Cergy-Pontoise,
France). Bovine GH (bGH) was a kind gift of Mosanto (St-Louis, MO). Rat
GH (rGH) (NIH-B-13, AFP-87401) was a gift from the National Institute
of Diabetes and Digestive and Kidney Diseases, and recombinant murine
IL-1ß, IL-6, and TNF were purchased through R & D systems
(Abingdon, U.K.). Antibodies to STAT-5b were purified Ig preparations
obtained from Santa Cruz Biotechnology Co. (Santa Cruz, CA). Nylon
membrane (Hybond N+) was from Amersham Life Science
(Amersham, U.K.). Female Wistar rats were obtained from the Katholieke
Universiteit Leuven (Heverlee, Belgium) and male Wistar rats were
purchased from IFFA-CREDO (l’Arbresle, France). Probes used for EMSA listed below (genomic sequences are in uppercase letters) include: (i) an oligonucleotide (GA-ext) (8) containing the spi GAGA box (underlined) and bearing its natural 6 and 7 nt-long 5'- and 3'-flanking sequences, respectively: 5'-tcgaTTCCTAAGAGGAGGGAGGAGCCTTTGGT; (ii) an oligonucleotide comprising the high-affinity STAT-5-binding site (underlined) present in the ß-casein gene promoter (32): 5'-gTCCCTTAATTCCAAGAAGTCC.
The following spi 2.1 promoter-derived oligonucleotides were used in genomic footprinting to amplify: (i), the upper strand:
L1 (-238/-218), 5'-GTACACTCTACTTTTGCTTTG;
L2 (-211/-187): 5'-TCCCACTTTCCTCATTGACTTTGAC;
L3 (-201/-170): 5'-CTCATTGACTTTGACCACTCAATAAATAAAAGG; and (ii), the lower strand:
L11 (+73/+56), 5'-TGATACCAGCCAGCTGCC;
L21 (+50/+28), 5'-CTGTGTTTGCTGACACCTGATGT;
L31 (+42/+18), 5'-GCTGACACCTGATGTTCAGGGTTGT.
It should be mentioned that, despite the extensive sequence conservation of the three spi 2 gene family members, more particularly in the promoter region (8), sets of oligonucleotides used in our experiments were totally specific, since they allowed exclusive amplification of the spi 2.1 sequence. The linker used was the same as that described by Mueller and Wold (7).
Nuclear Run-On Assays
Nuclei isolated from the liver (33) of normal, hypox, or
LPS-treated (killed 18 h after a single ip injection of 750
µg/100 g body weight) adult male rats were used to evaluate the
in vivo rate of transcription initiation. The procedure for
run-on analysis, which includes incubation of the nuclei with
[-32P]UTP, purification of total cellular RNA, and
hybridization to templates immobilized on a nylon membrane, has been
previously described (34). The following plasmids were used as
hybridization templates: pUC19 bearing 0.52 kb of the spi
2.1 cDNA 5'-region (9); pBR322 bearing 1.6 kb of the mouse ß-actin
cDNA (35); pBR322 containing 1.2 kb of the rat albumin cDNA (36). The
amount of labeled transcripts used for hybridization corresponded to
20 x 106 trichloroacetic acid-precipitable
counts.
EMSA
Liver nuclear extracts were prepared from control, LPS-treated
(18 h), or hypox rats, according to the method of Gorski et
al. (33) as modified by Sierra (37). The following minor changes
were introduced: 0.5 mM phenymethylsulfonylfluoride (PMSF),
14 µg/ml aprotinin, 1 mM Na3VO4,
and 10 mM NaF were added to the homogenizing buffer; 0.1
mM PMSF, 0.1 mM Na3VO4,
and 0.1 mM NaF were added to the nuclear lysis buffer; and
0.1 mM Na3VO4 and 0.1
mM NaF were added to the dialysis buffer. Double-stranded
oligonucleotides were end-labeled using T4 polynucleotide kinase and
[-32P]ATP. Probes were purified on nondenaturing
polyacrylamide gels before use. The incubation mixture (20 µl)
contained 10 µl liver nuclear extracts (10 µg proteins diluted in
dialysis buffer), 1 µl (2 µg)
poly(deoxyinosinic-deoxycytidylic)acid, 1 µl (100 µg) BSA, 4 µl
of 5x buffer (25 mM MgCl2, 20 mM
spermidine, 4.4 mM dithiothreitol, 0.25 mM
EDTA, 50% glycerol, 0.25 mM
Na3VO4, 0.25 mM NaF), and 2 µl
water or antibodies. After a 15-min preincubation at room temperature
(RT), the labeled probe (0.2 ng of oligonucleotide, 50–100.000 cpm)
was then added in 2 µl, and the binding reaction was allowed to
proceed for 15 min at RT. For supershift experiments, nuclear proteins
were first incubated with preimmune or immune serum for 1 h at 4
C. Samples were electrophoresed for 3 h at RT on 5% nondenaturing
polyacrylamide gels in 0.5 x TBE (44.5 mM Tris base,
44.5 mM boric acid, 1 mM EDTA) at 130 V.
In Vivo Footprinting of the spi 2.1
Promoter
Preparation of DNase I-Treated Templates
Nuclei isolated from the liver of control, hypox, or inflamed
(i.e. LPS-treated for 18 h) rats were incubated for 10
min on ice in a mixture containing 10 mM Tris-HCl (pH 7.4),
15 mM NaCl, 60 mM KCl, 0.15 mM
spermine, 0.5 mM spermidine, 0.5 mM
CaCl2 and 1 mM MgCl2, in the
presence of DNase I (10 µg/ml). Reactions were stopped by adding 4
vol of a lysis solution containing 1 mg/ml proteinase K, 30
mM EDTA, and 2.5% SDS, and nuclear lysates were incubated
for 4 h at 45 C. Naked DNA was concomitantly treated with DNase I.
Genomic DNA was purified as described by Rigaud et al. (21)
by phenol extraction and was ethanol precipitated and dissolved in
water.
Genomic Sequencing with the LMPCR Procedure
This was performed using a modification of the method of Mueller and
Wold (7) described by Grange et al. (38), except for
ligation of the linker that was carried out for 1.5 h at RT, in
the presence of 15% polyethylene glycol (39). Extension of primers 1
(L1 or L11) was performed at 50 C, exponential amplification (30
cycles) with primers 2 (L2 or L21) was performed at 66 C, and extension
(9 cycles) of 32P-labeled primers 3 (L3 or L31) for
detection of LMPCR products was performed at 68 C. Chemical sequencing
reactions were performed by the Maxam and Gilbert procedure (40).
Labeled products were separated on 6% polyacrylamide-urea sequencing
gels that were dried and autoradiographed without an intensifying
screen for 12–24 h at RT. Autoradiographs shown in Results
are raw data from a representative experiment. In in vivo
footprinting, only alterations in the pattern of bands that have been
reproduced with at least three independent template preparations have
been considered as real.
Analysis of DNase I-Hypersensitive Sites
Conditions used to incubate nuclei isolated from the liver of
control or hypox male rats with increasing concentrations of DNase I
(0–20 µg/ml) and the procedure to purify genomic DNA were as
described for the preparation of DNase I-treated templates for LMPCR.
DNA samples (25 µg) were then digested with BamHI, and
fragments were resolved on a 1% agarose gel and transferred onto a
nylon membrane. The membrane was hybridized to a uniformly
32P-labeled single-stranded RNA probe corresponding to the
BamHI-PstI fragment (positions -862 to -479) of
the spi 2.1 gene (8) and submitted to autoradiography at
-80 C for 48 h with an intensifying screen.
Animal Treatment, Hepatocyte Culture, and Northern Blotting
Analysis of spi 2 mRNAs
All animal studies were conducted in accord with the principles
and procedures outlined in "Guidelines for Care and Use of
Experimental Animals." Four groups (n = 6/group) of 4-week-old
female Wistar rats were used to analyze the effect of LPS and GH on
liver spi 2 mRNA levels in vivo. The first group
received one ip injection of LPS (750 µg/100 g body weight) and one
sc injection of saline; the second group received one sc injection of
bGH (200 µg/100 g body weight) and one ip injection of saline; the
third group was administered LPS and bGH; and the last group received
two injections of saline. All animals were killed 10 h after
injections.
Rat hepatocytes cultured on matrigel-coated plates for 48 h in serum-free medium were exposed during 24 h to cytokines and rGH at different doses. Procedures used to culture hepatocytes and extract total RNAs from the liver of adult rats or from cultured cells were as previously described (23). For Northern blotting analysis, 20-µg samples of total RNA were resolved on 1% denaturing agarose gels and transferred onto nylon membranes (Hybond) by vacuum blotting. Blots were then hybridized under standard conditions with a random-primed 32P-labeled spi 2.1 cDNA probe (fragment 60–1800) that was previously shown to recognize all spi messages (9). The 2.3-kb band corresponds to the spi 2.3 mRNA species, whereas the 1.8-kb band represents undistinguishable, size-matched spi 2.1 and 2.2 messages which, on the basis of the abundance of corresponding proteins, appeared to be equally represented in the normal liver (12).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Present address: Unité de Diabétologie et Nutrition,
School of Medicine, The University of Louvain, B-1200, Brussels,
Belgium. 
Received for publication September 10, 1997. Revision received December 15, 1997. Accepted for publication December 16, 1997.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-activated sites. J Biol Chem 270:24903–24910-activated sequence-like DNA elements.
J Biol Chem 269:26208–26214 of the insulin-like growth factor I messenger
ribonucleic acid response to growth hormone in rat hepatocyte primary
culture. Endocrinology 138:1078–1084B
factor. In: Cohen P, Foulkes JG (eds) The Hormonal Control of Gene
Transcription. Elsevier, Amsterdam, pp 423–446
B in the
signaling mechanism. Endocrinology 136:4790–4795[Abstract]
This article has been cited by other articles:
 |
|
 |
|
  M. D. Buzzelli, M. Nagarajan, J. F. Radtka, M. L. Shumate, M. Navaratnarajah, C. H. Lang, and R. N. Cooney Nuclear Factor-{kappa}B Mediates the Inhibitory Effects of Tumor Necrosis Factor-{alpha} on Growth Hormone-Inducible Gene Expression in Liver Endocrinology, December 1, 2008; 149(12): 6378 - 6388. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. A. Ahmed, M. D. Buzzelli, C. H. Lang, J. B. Capen, M. L. Shumate, M. Navaratnarajah, M. Nagarajan, and R. N. Cooney Interleukin-6 inhibits growth hormone-mediated gene expression in hepatocytes Am J Physiol Gastrointest Liver Physiol, June 1, 2007; 292(6): G1793 - G1803. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Ahmed, G. Yumet, M. Shumate, C. H. Lang, P. Rotwein, and R. N. Cooney Tumor necrosis factor inhibits growth hormone-mediated gene expression in hepatocytes Am J Physiol Gastrointest Liver Physiol, July 1, 2006; 291(1): G35 - G44. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Volpi, Y. Liu, and G. Aguilera Vasopressin increases GAGA binding activity to the V1b receptor promoter through transactivation of the MAP kinase pathway. J. Mol. Endocrinol., June 1, 2006; 36(3): 581 - 590. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. L. Shumate, G. Yumet, T. A. Ahmed, and R. N. Cooney Interleukin-1 inhibits the induction of insulin-like growth factor-I by growth hormone in CWSV-1 hepatocytes Am J Physiol Gastrointest Liver Physiol, August 1, 2005; 289(2): G227 - G239. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Ling and P. E. Lobie RhoA/ROCK Activation by Growth Hormone Abrogates p300/Histone Deacetylase 6 Repression of Stat5-mediated Transcription J. Biol. Chem., July 30, 2004; 279(31): 32737 - 32750. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 |
