(Received for publication, January 24, 1997, and in revised form, March 21, 1997)
From the Beatrice and Samuel A. Seaver Laboratory, Department of Medicine, Cornell University Medical College, New York, New York 10021
The gene encoding the high output isoform of
nitric oxide synthase represents a large class of alarm and defense
genes transcriptionally induced in response to bacterial
lipopolysaccharide (LPS). The promoters of most of these genes contain
at least two LPS-response elements, one of which commonly binds
transcription factors of the NF-B/Rel family. Here a novel
LPS-response element is identified in the inducible nitric oxide
synthase promoter, termed LREAA, which contains
critical adenosine residues lying 19-20 base pairs downstream of the
proximal NF-
B binding element (NF
Bd). Both NF
Bd and
LREAA are required for LPS-induced promoter activity. A protein partially recognized by antibody against transcription factor
Oct-1 binds to the LREAA element constitutively in
untreated macrophages while contributing to a DNA-protein complex that
includes NF-
B p50 in macrophages treated with LPS. NF-
B p50 and
the LREAA-binding proteins may together recruit an
LPS-triggered transactivator of transcription.
Endotoxic lipopolysaccharide (LPS),1
the major constituent of the cell walls of Gram-negative bacteria, is
one of the most potent agonists in biology. Mammalian, avian, and
insect phagocytes (1-4) respond to LPS by altering the expression of
numerous genes. This response can help protect the host from infection
but can also cause systemic inflammation (1, 2). A recent review catalogued 21 mammalian genes induced by LPS, and in 19 of these LPS-response elements have been characterized (5). Transcription factors binding to these promoter elements have been identified with
the following frequency: NF-B/Rel, 14 genes; NF-IL6, 9 genes; AP-1,
2 genes; and IL1
-UNF1, NF-1
, NF-
A, c-Jun/ATF2, IRSE-binding protein, NF-M, ATF family members, and c-Jun, 1 gene each.
LPS-inducible genes include NOS2 (iNOS), encoding the high output
isoform of nitric oxide synthase (6, 7), which contributes to both
antimicrobial defense (8) and LPS-induced hypotension (9).
Transcriptional induction of iNOS by LPS (6, 10, 11) requires binding
of transcription factor NF-B/Rel to a promoter element, NF
Bd
(12), at position
85 to
76 (13-16). Another NF-
B binding site
NF
Bu (12) at position
971 to
962 (13, 14) also plays a role
(16).
Although NF-B/Rel is essential for the LPS induction of iNOS, it is
not sufficient (12, 17). Additional transcription factors are required
(12), which is consistent with the precedent that at least two-thirds
of LPS-induced genes studied have more than one LPS-response element
(5). In the present work, a second LPS-response element
(LREAA) is identified in the iNOS promoter that has not
been previously recognized in any gene.
The macrophage cell line RAW 264.7 (American Type Culture Collection) was cultured in complete medium (RPMI 1640 supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and 200 µg/ml penicillin and streptomycin) as described (13).
ReagentsPoly(dI-dC)·poly(dI-dC) was from Pharmacia
Biotech Inc. and isotopes were from Amersham Corp. Santa Cruz
Biotechnology, Inc. (Santa Cruz, CA) supplied affinity-purified goat or
rabbit IgGs against synthetic peptides derived from NF-B/Rel
proteins p50, p52, p65 (RelA), c-Rel (C), and RelB, and from
transcription factors Oct-1, Oct-2, Pit-1, Stat1(p91), c-Jun/AP-1,
c-Fos, Ets-1/Ets-2, and C/EBP
. Rabbit IgG against purified
recombinant murine IRF-1 was prepared by Teruaki Nomura and Heinz
Ruffner and kindly provided by Dr. Luis Reis (University of Zurich,
Zurich, Switzerland). LPS from Escherichia coli 0111:B4 was
from Sigma.
Plasmids p7iNOS-CAT and
p8.11iNOS-CAT contain the iNOS promoter and a reporter gene
for chloramphenicol acetyltransferase (CAT). p7iNOS-CAT (18)
contains a full-length promoter, including a basal promoter and an
upstream enhancer; p8.11iNOS-CAT (12) contains only the
basal promoter. Polymerase chain reaction-engineered mutations of
element NFBd (from GGG to CTC) and of element LREAA (from AA to CG) were individually introduced into
p8.11iNOS-CAT to form p8.11
Bm and
p8.11LREm, respectively. Similarly, mutant constructs
p7
Bm and p7LREm were formed from
p7iNOS-CAT. DNA sequence analysis confirmed the mutation of
these elements without unwanted mutations elsewhere.
RAW 264.7 cells were transfected by a modification of the DEAE-dextran procedure (19). The DNA used for transfection was prepared by the EndoFree plasmid kit (QIAGEN Inc., Chatsworth, CA). Media and DNA at the concentrations used contained <50 pg/ml LPS (QCL-1000 kit, BioWhittaker, Inc., Walkersville, MD). Cells were washed twice with serum-free RPMI and suspended at 5-9 × 106/ml in RPMI prewarmed to 37 °C containing DEAE-dextran (250 µg/ml) and 50 mM Tris (pH 7.4). 10 µg of DNA was added to 2 ml of cell suspension at 37 °C for 45 min with occasional shaking. Cells were shocked with 10% Me2SO for 1 min at room temperature, washed, and distributed to two 100-mm plates with 10 ml of complete medium. After 24 h at 37 °C, 5% CO2, LPS (100 ng/ml) was added to some of the plates. About 16 h later, the cells were washed with ice-cold phosphate-buffered saline, resuspended in 0.25 mM Tris (pH 8.0), and frozen and thawed 3 times. Lysates were centrifuged (11,700 × g, 10 min, 4 °C), and the supernatant was heated at 65 °C for 10 min to inactivate CAT inhibitors and then centrifuged as above. The supernatant (1 or 10 µg) was assayed for CAT by a TLC method (20), and protein content was determined (21).
Oligonucleotides and ProbesSingle-stranded
oligonucleotides (Oligos Etc., Inc., Guilford, CT) were annealed with
the complementary strand by polymerase chain reaction to form
double-stranded oligomers with 5 overhang. To prepare probes (see
Table I), double-stranded oligomers were filled in by the Klenow
fragment of DNA polymerase I with [
-32P]dCTP and the
three other nonradiolabeled dNTPs. To prepare competitors, all four
dNTPs were nonradiolabeled.
|
Binding was tested in
15 µl of solution by incubating 5 µg of nuclear extract (22) with
reaction buffer (20 mM HEPES, pH 7.9, 1 mM
EDTA, 60 mM KCl, 12% glycerol, 1 mM
dithiothreitol, 2 µg of poly(dI-dC)·poly(dI-dC)) in the presence or
absence of competitor or antibody for 20 min, followed by a 20-min
incubation at room temperature with probe (20,000 cpm). Products were
electrophoresed at 30 mA for at least 3 h on 4.8% polyacrylamide
gels in high ionic strength buffer (50 mM Tris, 380 mM glycine, 2 mM EDTA, pH ~8.5) (23), and
dried gels were analyzed by autoradiography.
50 µg of nuclear extract and
106 cpm of bromodeoxyuridine-containing probe
DBm were reacted in a total volume of 60 µl. The
electrophoretic mobility shift assay gel was exposed to x-ray film for
10 min to localize the complex of interest. Excised gel was
UV-irradiated (366 nm) 5 cm from an inverted transilluminator at
4 °C for 60 min, boiled with sample buffer for 5 min, and analyzed by 10% SDS-polyacrylamide gel electrophoresis and autoradiography.
Construct
p8.11iNOS-CAT contains a fragment of the 5-flanking region
of the mouse iNOS gene (
85 to +161) that includes the NF
Bd element
(
85 to
76) and confers LPS-inducible promoter activity on
transfected RAW 264.7 mouse macrophages (12). LPS responsiveness was
lost in p8.13iNOS-CAT, from which the the NF
Bd element
was eliminated (12). In this work, two mutated constructs derived from
p8.11iNOS-CAT were prepared by (Fig.
1A) and tested for their promoter activity as
induced by LPS (Fig. 1B). In p8.11
Bm, three
nucleotides of the NF
Bd element were mutated, from
GGGACTCTCC to CTCACTCTCC. This reduced
LPS-induced promoter activity to 3.2 ± 1.3% of wild type
(mean ± S.D., three experiments), consistent with an earlier
finding that nuclear protein failed to bind to the NF
Bd element when
it carried the same mutation (12). Surprisingly, mutation of two
nucleotides (from AA to CG) located 19-20 base pairs downstream from
the 3
end of the NF
Bd element, as seen in p8.11LREm,
also nearly abolished LPS-induced promoter activity leaving only
2.0 ± 0.3% as much activity as in wild type (mean ± S.D.,
three experiments). This new LPS-response element will be called
LREAA. When the same mutations of elements NF
Bd and LREAA were individually introduced into the full-length
promoter construct p7iNOS-CAT (Fig. 1A) to form
p7kBm and p7LREm, LPS-induced transcription was
also substantially reduced (Fig. 1C). Therefore, both
elements play an important role in the full-length promoter as well as
in small fragments.
Contribution of Both LREAA and NF
Oligonucleotides derived from the iNOS promoter
with or without mutation (Table I) were used as probes
or competitors in electrophoretic mobility shift assays to analyze the
DNA binding activity of nuclear extracts from cells cultured with or
without LPS. As described earlier (12), a cycloheximide-sensitive
DNA-protein complex termed "X" was formed upon LPS induction when
probe B was used (Fig. 2A, lane
2). Two other faster migrating complexes, designated "Z" and
"Y" in Fig. 2A (not labeled on Fig. 2A in
Ref. 12), were present whether or not the cells were exposed to LPS. Without LPS, complex Z was strong and complex Y faint; LPS increased the amount of complex Y without appreciably changing the amount of
complex Z (see Fig. 2 in Ref. 12).
Probe D is 19 nucleotides shorter than probe B at the 3 end but still
contains elements NF
Bd and LREAA. Probe D had the same
binding activity as probe B, forming complexes X, Y, and Z with nuclear
extracts from LPS-induced cells (Fig. 2A, lane 4). Thus, the nucleotides between
85 and
50 of the iNOS
promoter are sufficient to sustain the formation of all three
complexes.
Competition assays established that LREAA together with
NFBd contributed to the formation of these complexes. Complexes X and Y but not Z were competed by excess nonlabeled oligomer A containing element NF
Bd but not LREAA (Fig.
2B, compare lanes 2 and 4). Such
competition did not occur with oligomer AkBm, whose NF
Bd
element was mutated (Fig. 2B, compare lanes 2 and
6). On the other hand, oligomer C, which contained the
LREAA element, blocked the formation of complex Z (Fig.
2B, compare lanes 1 and 7 without LPS
and lanes 2 and 8 with LPS) and most of complex X
(compare lanes 2 and 8). No competition was seen
using oligomer CLREm with a mutated LREAA
element (lanes 9 and 10).
Results of transcription (Fig. 1B) and competition assays
(Fig. 2B) were mirrored by tests of direct binding to
mutated probes (Fig. 2, C-E). Taking the formation of
complexes X, Y, and Z upon LPS induction as wild type DNA binding
activity (Fig. 2C, lane 2), the incomplete
formation of these complexes resulted from mutation of either NFBd
(lane 3) or LREAA (lane 4) in the
context of probe D. Probe DkBm containing a point mutation
of element NF
Bd did not sustain formation of complex X or Y but
still formed complex Z (Fig. 2C, lane 3; note its
slightly slower migration). On the other hand, complex Z and most of
complex X were absent using probe DLREm containing the
mutated element LREAA; in the meantime, complex Y became
stronger and was accompanied by a complex migrating more slowly than Y
(lane 4). Probe DkBmLREm, in which both elements
are mutated, formed none of the complexes (lane 5). Probe C,
containing LREAA but not NF
Bd, formed complex Z alone
(Fig. 2F, lanes 1 and 2), whereas no
complex formed on the mutated probe CLREm (lanes
7 and 8). Thus, LREAA and NF
Bd were the
only two elements required for the binding of nuclear factors on the
iNOS promoter between nucleotides
85 and
50.
The complex Z that formed with probe C or the mutated probe DkBm shared the same characteristics as the complex Z that formed with wild type probe D. Its formation was independent of exposure to LPS (Fig. 2, D and F, lanes 1 and 2) but dependent on the presence of element LREAA. Formation of complex Z on probe D was competed by excess unlabeled oligomer C containing the element LREAA (Fig. 2, D and F, lanes 3 and 4) but not by the oligomer CLREm in which LREAA is mutated (Fig. 2, D and F, lanes 5 and 6).
On the other hand, the complex Y that formed on probe DLREm
was more abundant than that formed on wild type probe D (Fig. 2C, compare lanes 2 and 4). Formation
of complex Y required only the element NFBd, because complex Y that
was formed with probe DLREm disappeared in a competition
assay with NF
Bd containing oligomer A (Fig. 2E,
lane 4) but not with oligomer AkBm containing the mutated NF
Bd element (not shown).
With probe DLREm, a small amount of complex X was still seen after LPS induction (Fig. 2E, lane 2). This residual complex X was competed by oligomer C (Fig. 2E, lane 6).
p50 and Oct-1-like Proteins Contribute to the DNA-Protein ComplexesSupershift assays gave information about some of the
proteins comprising complexes X, Y, and Z. No reaction was detected
with antibodies against NF-B p52, p65, c-Rel, or RelB, or against IRF-1, STAT1 (p91), c-Jun, c-Fos, C/EBP, Ets-1/Ets-2, NF-AT, Oct-2, or
Pit 1, except for a partial supershift of complex X with large amounts
of anti-c-Rel (not shown). In contrast, anti-NF-
B p50 (a reagent
that immunoblotted and supershifted authentic p50 overexpressed upon
cell transfection; not shown) completely supershifted complex X (Fig.
3A, lane 2) while leaving complex
Z untouched (Fig. 3, A and B, lanes 1 and 2). Moreover, anti-p50 completely supershifted complex Y
(marked by an asterisk in lane 2 of Fig.
3A), and such supershift was even clearer when complex Y was
formed with probe DLREm (Fig. 3C, lane
3). Thus, p50 is a component of complexes X and Y but is not
present (or is not accessible to the antibody) in complex Z.
Anti-Oct-1 (a reagent that supershifted authentic Oct-1 overexpressed
upon cell transfection; not shown) partly supershifted complexes X and
Z (Fig. 3, A and B, lanes 3 and
4). In contrast, anti-Oct-1 was nonreactive with complex Y
as formed with probe DLREm (containing NFBd but not
LREAA) (Fig. 3C, lane 4). As noted above, probe DkBm only supported the formation of complex Z
and did so whether or not the cells had been exposed to LPS (Fig. 3B, lanes 5 and 6).
Complex Z was subjected to UV cross-linking followed by
SDS-polyacrylamide gel electrophoresis, revealing that the region containing element LREAA bound at least three nuclear
proteins with apparent molecular masses of ~160, ~100, and ~60
kDa (Fig. 4).
The most frequently implicated LPS-response element in mammalian
promoters is the 10-base pair B element (GGGRNNYYCC) that binds
transcription factors of the NF-
B/Rel family (5, 24). NF-
B
frequently associates with other transcription factors to impart
specific regulation (25). The present work identifies an LPS-response
element termed LREAA including the dinucleotide AA
downstream of NF
Bd in the mouse iNOS promoter. An NF-
B-like binding site followed closely by an LPS-response element containing an
AA dinucleotide was reported in the regulatory region of the major
histocompatibility complex class II A
k gene
(26) and later in the mouse granulocyte colony-stimulating factor
promoter (27), but in neither gene was the AA dinucleotide noted.
The discovery of LREAA focuses attention on the identity of
the transcription factors that bind to it and the nature of their interaction with NF-B. The requisite dinucleotide AA of
LREAA is embedded in an octamer-like sequence,
ATGCAAAA. This sequence departs from the canonical octamer
(ATGCAAAT) at the eighth position. Conversion of the eighth nucleotide
from T to A may reduce or abolish the binding of the ubiquitous
transcription factor Oct-1, as evidenced in the human Ig heavy chain
gene enhancer, where the Oct-4 element (ATGCAAAA) bound octamer-binding
proteins with only very low affinity (28). Similarly, Oct-1 may bind
little to the iNOS LREAA. Other transcription factors known
to bind the octamer motif include Pit-1 and the B cell-specific Oct-2
(29). However, with oligonucleotide probes derived from iNOS promoter and nuclear extracts from RAW cells, antibodies against Pit-1 and Oct-2
had no effect, while the antibody against Oct-1 caused a partial
supershift in the two complexes (Z and X) whose formation depended on
LREAA. Thus, an Oct-1 like
protein (OLP) is a candidate for one of the transcription
factors interacting with LREAA. UV cross-linking analysis
indicated that the LREAA-dependent complex Z
included at least three nuclear proteins with molecular masses of
~160, ~100, and ~60 kDa. None of these correspond to the
molecular mass of Oct-1 (~70 kDa). It is not known which, if any, of
these three proteins binds anti-Oct-1 antibody to cause a partial
supershift of complex Z. The ~100-kDa species corresponds in size to
a component previously detected in complex X (12). The ~100 kDa
protein that binds the LREAA may be the same as the ~100
kDa protein that binds NF
Bd.
Sequence context strongly influences the composition of promoter
binding complexes. The practice of using minimal probes to sustain
complex formation militates against detecting factors that impart
specificity to the induction of genes regulated by widely shared
transcription systems such as NF-B. The present study used
relatively long probes. Results with such probes were consistent with
findings from reporter constructs including those representing point
mutants in the full-length promoter. In earlier work (12) probe A,
containing only the NF
Bd element, supported the LPS-activated
binding of p50/p65 and p50/c-Rel. However, this did not fully explain
LPS-induced promoter activity of iNOS. First, activation of NF-
B/Rel
is not sensitive to cycloheximide (12), but synthesis of iNOS mRNA
in the cells under study was cycloheximide-sensitive (30). Second,
activation of p50/p65 and p50/c-Rel peaked at 0.5 h after LPS
induction and then decreased (12, 16), but synthesis of iNOS mRNA
continued for more than 24 h (11, 18). Finally, NF-
B/Rel was
activated by LPS in an LPS-hyporesponsive macrophage cell line from
C3H/HeJ mice, but iNOS was not induced. Thus, NF-
B/Rel was not
sufficient for LPS induction of iNOS (17). In contrast, probes B and D
(Ref. 12 and present study) included not only the NF
Bd element but
also downstream sequences that appear to be relatively specific for the
iNOS gene. The complexes formed with probes B and D after LPS induction
were different from those formed with probe A. In particular, complex X
required both NF
Bd and LREAA, contained additional
protein(s) besides those of the NF-
B/Rel family, lacked p65 and
c-Rel, and was sensitive to cycloheximide (12).
Sequence analyses suggested that elements for binding of NF-IL6 are
present in the iNOS promoter at positions 74 to
66 and
150 to
142 (14); the latter was protected by in vivo footprinting during LPS induction (15). However, mice rendered genetically deficient
in NF-IL6 produced iNOS normally in response to LPS and interferon-
(31). Footprinting also showed protection of nucleotide at
58 (within
the octamer-like sequence ATGCAAAA) after LPS induction
(15). The present report demonstrates that the dinucleotide AA at
56
and
55 is critical to the formation of complex Z, which is
independent of LPS induction, and that the mutation of AA to CG
eliminates both protein binding and promoter activity.
Based on reporter constructs, binding assays, competition experiments,
and antibody supershifts, it is hypothesized that both constitutive
complex Z and inducible complex X required LREAA and
contained OLP, whereas inducible complexes X and Y required NFBd and
contained NF-
B p50. Since no direct interaction between NF-
B and
an Oct-1-containing complex has been reported and since p50 lacks a
transactivation domain (25), it is postulated that LPS causes a
distinct protein to bridge p50 and OLP, contributes to the formation of
complex X, and either furnishes or recruits the transactivation domain
that stimulates transcription of iNOS. The bridging protein may be the
~100 kDa species that seems common to complex X and complex Z, or the
transactivating protein may bind the ~100 kDa species when it is
bound to NF
Bd (an LPS-induced event) as well as to LREAA
(a constitutive event).
Together, NF-B p50 and the activated OLP-containing complex are
proposed to recruit a transactivator to complex X, much as Bcl-3
supplies transactivating capacity by binding p50 dimers on DNA (32,
33). Cloning of the proteins in complexes X and Z may provide fresh
approaches to the pharmacologic control of iNOS expression and other
responses to LPS.
Carl Nathan provided critical discussions and editorial assistance, Mary Leung gave excellent technical help, and Lei Chen critiqued the manuscript.