(Received for publication, April 11, 1997)
From the Division of Pulmonary and Critical Care Medicine, Departments of Medicine and Pathology, University of Maryland School of Medicine, Baltimore, Maryland 21201, Medical Service, Baltimore Veterans Administration Medical Center, Baltimore, Maryland 21201, and the University of Maryland at Baltimore Cytokine Core Laboratory, Baltimore, Maryland 21201
In unstimulated RAW 264.7 macrophage-like cells,
tumor necrosis factor- (TNF-
) mRNA was transcribed and
accumulated in the cytoplasm, but the TNF-
transcripts failed to
associate with polysomes, and TNF-
protein was not detected.
Stimulation with lipopolysaccharide (LPS) induced an increase in
TNF-
transcription, cytoplasmic TNF-
mRNA accumulation,
polysome association, and secretion of TNF-
protein. This process
was associated with a 200-nucleotide increase in the apparent length of
the TNF-
mRNA. The difference in TNF-
mRNA size was
caused by marked truncation of the 3
poly(A) tail in unstimulated
cells. Fully adenylated TNF-
mRNA appeared within 15 min of
LPS stimulation. We speculate that removal of the poly(A) tail blocks
initiation of TNF-
translation in unstimulated macrophages. LPS
inactivates this process, allowing synthesis of translatable
polyadenylated TNF-
mRNA.
Tumor necrosis factor-
(TNF-
)1 plays a pivotal
role in inflammation and host defense (1). TNF-
stimulates effector
cell microbicidal and tumoricidal activity, enhancing survival of the infected host (2, 3) and causing regression of some tumors (4). On the
other hand, persistent or inappropriately high TNF-
expression has
grave consequences, including multiorgan failure and death (5, 6).
Thus, stringent control of TNF-
regulation is critical for host
survival and has led to the evolution of multiple levels of macrophage
TNF-
regulation (7).
The TNF- gene lies between the lymphotoxin-
and lymphotoxin-
genes in a tightly linked array, but these genes are regulated by
distinct promoters (8, 9). TNF-
transcription occurs in
plastic-adherent but otherwise unstimulated macrophages in the absence
of detectable translation (10, 11). The TNF-
promoter shares several
regulatory elements with other genes, including, but not limited to,
four
B-like elements, a GC box/Sp-1 binding site, a Y-box like
decanucleotide, a cAMP-responsive element-binding protein binding site,
and binding sites for AP-1 and AP-2 (12). Several macrophage activators
induce a modest increase in TNF-
transcription (13), which is
blocked by agents that enhance intracellular cyclic AMP levels (14). In
the ANA-1 macrophage cell line, TNF-
transcript elongation is
blocked in unstimulated cells. Bacterial endotoxin lipopolysaccharide
(LPS) stimulation both increases TNF-
transcription initiation and
reverses the transcript elongation block (15). Once transcribed,
TNF-
mRNA is stabilized by the same stimuli that induce TNF-
release in macrophages, including LPS (16). Enhancement of
transcription initiation, release of the transcript elongation block,
and transcript stabilization contribute to the accumulation of TNF-
mRNA in stimulated macrophages. The TNF-
mRNA that is
present in unstimulated macrophages is translationally silent.
Macrophage activation induces both accumulation and translation of
TNF-
mRNA (11, 13, 17). Cis acting elements in the TNF-
3
-untranslated region are sufficient for both translational silencing
in unstimulated macrophages and the LPS-induced activation of TNF-
translation (17).
We (18) and others (19) have reported the translationally silent
TNF- mRNA in unstimulated macrophages migrates faster than the
translationally active TNF-
transcripts in LPS-stimulated macrophages in Northern blot analyses. Possible causes of the observed
differences in TNF-
transcript size in unstimulated and stimulated
macrophages include: 1) isoforms of the TNF-
gene; 2) use of an
alternative poly(A) signal site(s); 3) mRNA splice variants in
unstimulated macrophages lacking one or two internal exons; or 4)
heterogeneity in poly(A) tail length. Current data do not support the
first two possible possibilities. Previous searches of genomic
libraries did not reveal any genes with homology to TNF-
(8) and
only a single AAUAAA canonical nuclear poly(A) signal site is present
in mouse and human TNF-
sequences (20).
The objectives of this study are to determine the cause of the TNF-
mRNA size heterogeneity in unstimulated and LPS-stimulated macrophages and to evaluate its relevance to the regulation of TNF-
expression. We show that the TNF-
mRNA in adherent, unstimulated monocytes and macrophages is approximately 200 nucleotides shorter than
those that appear after LPS stimulation. The size difference is due to
markedly shorter or absent 3
poly(A) tails on the TNF-
mRNA in
the unstimulated macrophages. We show that the 3
poly(A) tail
shortening is not caused by either ongoing translation or dissociation
of TNF-
mRNA from polysomes and suggest that 3
poly(A) tail
metabolism may play a critical role in TNF-
regulation.
Phenol-extracted LPS prepared from Escherichia coli 0111:B4 was obtained from Difco. Unless otherwise specified, all other reagents were obtained from Sigma.
Cells and Culture ConditionsRAW 264.7 (RAW), J774A.1, and PU5-1R mouse macrophage cell lines were obtained from the American Type Culture Collection (Rockville, MD). The P388D1 mouse macrophage cell line was the kind gift of Dr. R. Nordan (National Institutes of Health, Bethesda, MD). All cell lines were maintained in either RPMI 1640 (Mediatech; Fairfax, VA) supplemented with 50 units/ml penicillin, 50 µg/ml streptomycin, 2 mM L-glutamine, 1 mM sodium pyruvate, 10 mM Hepes buffer, pH 7.3 (Life Technologies, Inc.) (CRPMI), containing heat-inactivated 10% newborn calf serum (Hyclone; Logan, UT) or in Macrophage-SFMTM (Life Technologies) serum-free medium without supplementation (SFM) at 37 °C in 5% CO2 enriched air. Cells were routinely tested for Mycoplasma sp. infection using a commercial assay system (MycoTect; Life Technologies), and new macrophage cell line cultures were established monthly from frozen stocks. All media and reagents contained <0.1 ng/ml endotoxin as determined by a Limulus polyphemus amebocyte lysate assay (Associates of Cape Cod; Falmouth, MA). Whole blood was obtained from healthy volunteers following informed consent using a protocol approved by the University of Maryland at Baltimore Institutional Review Board. Peripheral blood mononuclear cells were isolated by Ficoll-hypaque density centrifugation (21), resuspended in CRPMI containing 10% autologous human serum, and cultured in Teflon beakers (Nalgene; Rochester, NY) for 7 days before use (22).
RNA IsolationTotal RNA was isolated by an acidic
guanidinium thiocyanate-phenol-chloroform extraction as described (23).
Cytoplasmic RNA was obtained by lysing cells in polysome buffer (PB)
containing 10 mM Tris, pH 7.3, 150 mM NaCl, 10 mM MgCl2, 0.65% Nonidet P-40 and collecting
the postmitochondrial supernatant fraction after centrifugation at
10,000 × g for 15 min at 4 °C. The
postmitochondrial supernatants were either applied directly to sucrose
gradients for polysome separations or were extracted with
SDS/phenol/chloroform/isoamyl alcohol followed by ethanol precipitation
of the cytoplasmic RNA. Precipitated RNA was washed in ethanol,
dissolved in water, quantitated by absorbance at 260 nm
(A260), and stored at 80 °C (24).
Ten micrograms of denatured RNA/lane as well as a RNA molecular size standard (Life Technologies) were separated by electrophoresis through either 1.5 or 3% agarose (SeaKem Gold, FMC Bio Products; Rockland, ME), 0.74 M formaldehyde (Fisher) gels. The separated RNA was transferred to nylon membrane (Duralon-UV, Stratagene; La Jolla, CA) by positive pressure (PosiBlot, Stratagene), UV-cross-linked (UV Stratalinker, 120,000 µJ, Stratagene), prehybridized in QuikHyb (Stratagene) for 15 min at 68 °C, and then hybridized for 1 h at 68 °C in QuikHyb containing 0.67 µg/ml denatured salmon testes DNA and 32P-labeled cDNA probe. Following washes of increasing stringency, membranes were analyzed with a PhosphorImager (Molecular Dynamics; Sunnyvale, CA) and autoradiographed. These blots were stripped and reprobed with a second radiolabeled cDNA where indicated according to standard methods (24). Equal RNA loading was determined by UV visualization and photography of 18 and 28 S ribosomal RNA bands in ethidium bromide-stained gels and Northern blots as well as quantification of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) housekeeping gene transcripts on Northern blots.
ProbesA 0.8-kb cDNA probe for human GAPDH was obtained
from M. Olman (University of California; San Diego, CA). A 2.0-kb
cDNA for chicken -actin was obtained from D. Cleveland (Johns
Hopkins University School of Medicine; Baltimore, MD). A 1.6-kb
cDNA for mouse TNF-
was provided by B. Sherry (Picower
Institute; Manhasset, NY). A fragment of mouse TNF-
cDNA
comprising exon 4 was provided by S. Nedospasov (National Cancer
Institute; Frederick, MD). For Northern blot analysis, cDNA inserts
were purified and labeled with 32P using a random primer
kit (Life Technologies) according to the manufacturer's
instructions.
Twenty micrograms of
cytoplasmic RNA was denatured in a reaction mixture containing 0.6 µg
oligo(dT)12-18 (Life Technologies), 20 mM
Tris, pH 7.5, 100 mM KCl, 10 mM
MgCl2, 0.1 mM dithiothreitol, and 5% sucrose
at 70 °C for 5 min. The mixture was placed on ice for 15 min before
the addition of 2.5 units RNase H (Life Technologies) and incubation at
37 °C for 30 min. The digest was extracted with phenol/chloroform/isoamyl alcohol followed by sodium acetate/ethanol precipitation of the RNA. Precipitated RNA was washed in ethanol, dissolved in water, quantitated, and stored at 80 °C.
Heat-denatured RNA was reverse transcribed in a
reaction mixture containing 50 mM Tris, pH 8.3, 75 mM KCl, 3 mM MgCl2, 0.5 µg of
random hexamers (Life Technologies), 0.5 mM dNTP mix
(Pharmacia Biotech Inc.), 0.01 mM dithiothreitol (Life
Technologies), 40 units of RNasin (Promega; Madison, WI), 2 µg of
acetylated bovine serum albumin, and 200 units of Moloney murine
leukemia virus reverse transcriptase (Life Technologies) at 37 °C
for 1 h. The reaction was terminated by heating at 95 °C for 5 min, chilled on ice, and stored at 20 °C. One microliter of the
first strand product was amplified using 12.5 µM primer
pairs and 0.625 units Taq DNA polymerase (AmpliTaq, AS;
Perkin-Elmer) in a reaction mixture containing 2.5 mM
MgCl2, 10 mM Tris, pH 8.3, 50 mM
KCl, and 0.5 mM dNTP mix (Pharmacia). Each cycle consisted
of a 30-s denaturation at 95 °C, 1 min of annealing at 54 °C, and
3 min of extension at 72 °C. A total of 35 cycles were performed
followed by a 10-min terminal extension (RoboCycler, Stratagene).
Amplification products and a DNA molecular size standard (Life
Technologies) were analyzed by electrophoresis in 1.5% agarose (Life
Technologies) with UV visualization and photography. PCR primers were
constructed: upstream sense primer, 5
-TCAGCGAGGACAGCAAGGGA-3
;
downstream antisense primer, 5
-CCAAGCGATCTTTATTTCTCTCA-3
.
Twelve-milliliter 15-50% linear sucrose gradients in PB lacking Nonidet P-40 were prepared, and approximately 0.5 ml of postmitochondrial supernatant was layered on the gradient and centrifuged for 1 h, 45 min at 39,700 rpm (270,000 × g) in a SW41Ti rotor (Beckman Instruments, Inc.) at 4 °C. Fractions were collected, using a density gradient fractionator (Isco; Lincoln, NE; Brandel Inc.; Gaithersburg, MD), into equal volumes of isopropyl alcohol while monitoring A254. RNA in the fractions was pelleted by centrifugation, extracted sequentially in acidic guanidinium thiocyanate (23) and phenol/chloroform/isoamyl alcohol, precipitated with isopropyl alcohol, washed in ethanol, dissolved in water, quantified by A260, and analyzed by Northern blotting.
TNF-TNF- protein was measured by
sandwich enzyme-linked immunosorbent assay using commercial reagents
(Minikit, Endogen; Cambridge, MA) and expressed in pg/ml as calculated
from a recombinant TNF-
standard curve. This assay has a sensitivity
of 8 pg/ml.
TNF- and GAPDH
transcription rates were measured using a modification of previously
described nuclear run-on analysis (25). RAW cells were grown to
approximately 80% confluence. LPS (100 ng/ml) was added, and the
incubation was continued for the indicated duration. At each indicated
time point, cells were lysed in 10 ml of lysis buffer, containing 10 mM Tris, pH 8.0, 2.5 mM MgCl2, 0.25% Triton X-100, 0.3 M sucrose, 1 mM
dithiothreitol, and nuclei were collected by centrifugation for 5 min
at 500 × g. Isolated nuclei were manually counted, and
equal numbers from each culture were incubated with 100 µCi of
[
-32P]UTP (NEN Life Science Products) for 30 min at
37 °C. Plasmid containing GAPDH, full-length cDNA for TNF-
,
or the exon 4 TNF-
cDNA fragment or plasmid without insert was
alkaline-denatured and immobilized on nitrocellulose (Protran;
Schleicher and Schuell) by vacuum filtration using a slot-blot manifold
(Hoeffer) or a dot blot manifold (Bio-Rad). The membrane was rinsed
with 6 × SSC, air-dried, and UV-cross-linked. Membranes were
prehybridized overnight at 42 °C in 50% formamide, 5 × SSC,
2 × Denhardt's solution, 0.1% SDS, and 100 µg/ml denatured
salmon testes DNA and then hybridized at 42 °C for 3 days in
prehybridization buffer containing 32P-labeled RNA
extracted from the run-on reaction mixtures. Membranes were extensively
washed (four washes with 2 × SSC, 0.1% SDS at room temperature
for 5 min each and then 0.1 × SSC, 0.1% SDS at 60 °C for 30 min) and air-dried. The membranes were then analyzed with a
PhosphorImager (Molecular Dynamics) before exposing to XAR film.
RAW cells were plated in six-well polystyrene tissue culture plates (Costar; Cambridge, MA) and allowed to reach 90% confluence. Monolayers were incubated with 67 µCi of [5,6-3H]uridine (NEN Life Science Products) in SFM for 10 min at 37 °C and were then treated with 5 µg/ml actinomycin D. At various times, aliquots of cells were washed with ice-cold SFM and lysed with SDS, and cell lysates were precipitated with ice-cold trichloroacetic acid. The trichloroacetic acid-insoluble precipitate was collected on glass fiber filters (GF/B, Whatman International, Ltd.; Maidstone, United Kingdom), air-dried, transferred to scintillation fluid (Ready Safe, Beckman Instruments, Inc.), and counted (LS5000TD, Beckman Instruments, Inc.).
To understand the anomaly
in mRNA size and its role in TNF- gene expression, we first
measured the expression pattern of TNF-
in several macrophage cell
lines as well as in primary cultured human monocyte-derived macrophages
(Fig. 1A). In four mouse
macrophage cell lines (lanes 1-8) and in primary cultured
human macrophages (lanes 9 and 10), a discrete
TNF-
mRNA band was evident on Northern blots of RNA from
adherent, unstimulated cells (lanes 1, 3,
5, 7, and 9). Stimulation with LPS
(100 ng/ml) in serum-free medium for 1 h caused two effects in
these cells: 1) an increase in TNF-
mRNA band intensity, and 2)
a decrease in electrophoretic mobility and broadening of the TNF-
band on denaturing agarose/formaldehyde gels, indicating an increase in
transcript size and heterogeneity after LPS stimulation (lanes
2, 4, 6, 8, 10).
Similar changes occurred in cells incubated in serum-containing medium
(data not shown). Higher resolution Northern analysis of RAW
macrophage-like cell line TNF-
mRNA on 3% agarose/formaldehyde
gels (Fig. 1A, inset) indicated an approximately
200-240-nucleotide larger TNF-
mRNA in the LPS-stimulated
macrophages.
Nuclear run-on analysis (Fig. 1B) showed TNF- was
transcribed in unstimulated adherent RAW cells at 29% of the maximal
rate achieved after LPS stimulation. As previously reported (15), elongation of TNF-
transcripts is partially blocked in unstimulated macrophages, leading to 5
fragments of TNF-
pre-mRNA. To
examine the transcription rate of full-length TNF-
mRNA, we also
used a 3
cDNA fragment representing exon 4 to detect radiolabeled transcripts (Fig. 1B, hatched bars). The results
of the run-on assays using the full-length TNF-
cDNA and the
exon 4 fragment were virtually identical.
LPS induced a 5-fold increases in TNF- mRNA levels 1 h
post-stimulation (Fig. 1A). TNF-
secretion, which was
undetectable in unstimulated cells, increased to 4.87 ± 0.61 ng
of TNF-
/million cells 6 h after LPS treatment (Fig.
1C). These data suggest that LPS induces both TNF-
mRNA stabilization and TNF-
translation.
Analysis of published
TNF- cDNA and genomic sequences revealed that mature TNF-
mRNA comprises 4 exons and 3 introns (20). To determine if one or
both of the internal exons were excluded from the TNF-
mRNA in
unstimulated macrophages, random hexamer-primed reverse transcribed
cDNA from adherent, unstimulated, and LPS (100 ng/ml for 1 h)-stimulated RAW cells were PCR-amplified (RT-PCR) using a primer pair
that spanned the full length of the TNF-
mRNA (Fig.
2A). Exclusion of an internal
exon from unstimulated macrophage TNF-
mRNA would result in
amplification of a smaller RT-PCR product in these cells. In four
experiments, only a single 1.6-kb product was amplified in unstimulated
macrophages that was identical in size to the product in LPS-stimulated
cells (Fig. 2B), thus excluding alternative TNF-
splicing
as a cause for the observed transcript heterogeneity.
Contribution of 3
Having excluded the possibility that a difference
in internal TNF- sequence between unstimulated and LPS-stimulated
macrophages could account for the heterogeneity, the contribution of
heterogeneous TNF-
transcript 3
poly(A) tail length to TNF-
mRNA size heterogeneity was analyzed using an oligo(dT)/RNase H
digestion assay (26) (Fig. 2C). The size of undigested
(lanes 1, 3, 5, and 7) and
digested (lanes 2, 4, 6, and
8) TNF-
mRNA was determined by Northern blot analysis
on 3% agarose/formaldehyde gels that had a resolution of approximately
20-30 nucleotides. Digesting TNF-
mRNA from LPS-stimulated
macrophages (lanes 4, 6, and 8)
reduced its apparent size to that of TNF-
mRNA in unstimulated
cells (lane 1). By contrast, digesting TNF-
mRNA from
unstimulated cells did not cause any apparent change in TNF-
transcript size (lanes 1 and 2). In comparison,
GAPDH mRNA in unstimulated and LPS-stimulated macrophages were of
similar size, and both were shortened equally by oligo(dT)/RNase H
digestion. These results demonstrate that the TNF-
mRNA 3
poly(A) tail is either absent or markedly truncated in unstimulated
macrophages. Henceforth, we will refer to the short TNF-
mRNA in
unstimulated macrophages as hypoadenylated mRNA.
The 3 poly(A) tail has been described as a cis
acting transcript-stabilizing factor (27). Classically, 3
poly(A)
tails shorten during translation. When its length is reduced below
approximately 25 nucleotides, the minimal length required for poly(A)
binding protein binding, degradation of the transcript body ensues
(28). Taken together, the 3-4-fold increase in transcription rate and the 5-fold increase in steady state TNF-
mRNA levels suggests that TNF-
transcripts would be at least as stable in LPS-stimulated macrophages as in unstimulated cells. We measured the structural half-life of TNF-
mRNA by blocking transcription with 5 µg/ml actinomycin D, which completely blocked [3H]uridine
incorporation into RNA 10 min after treatment (Fig. 3A). The apparent half-life of
the hypoadenylated TNF-
mRNA in unstimulated TNF-
mRNA
was surprisingly long (30-49 min) (Fig. 3, B and
C) compared with a 22-min half-life 1 h after LPS
stimulation, which we have previously reported (25). These data
indicate that a pool of relatively stable cytoplasmic hypoadenylated
TNF-
mRNA may be present in unstimulated macrophages. Of note,
in the absence of LPS, the intensity of the TNF-
mRNA band
gradually decreased without a change in the apparent transcript length
as would be expected if its degradation began with exonucleolytic shortening of the poly(A) tail (Fig. 3B, lanes
1-9). This transcript degradation pattern differs from the
gradual shortening of TNF-
mRNA that occurs after LPS
stimulation (25).
Polysome Association of TNF-
The 3 poly(A) tail enhances
recruitment of ribosomes to mRNA and increases translational
efficiency (29). The polysome distribution of cytoplasmic TNF-
mRNA was analyzed using sucrose density gradient centrifugation
(Fig. 4). In LPS-stimulated RAW cells,
55% of the TNF-
mRNA was associated with polysomes containing >5 ribosomes (Fig. 4, A and B, fractions
6-8). A minor component of predominantly hypoadenylated
(poly(A)
) TNF-
mRNA was found in the
ribonucleoprotein/monosome fractions (fractions 2 and
3) in the LPS-stimulated cells. On the contrary, virtually
all of the TNF-
mRNA in lysates from unstimulated cells was
detectable in the ribonucleoprotein fractions (Fig. 4, C and
D, fractions 2 and 3). No detectable
TNF-
mRNA was polysome-bound in these cells (lanes
6-8). These data show that the hypoadenylated TNF-
mRNA in
unstimulated macrophages is not associated with polysomes and is
consistent with its translational dormancy as previously reported
(17).
Effects of Translational Inhibitors on TNF-
It is
possible the increased TNF- transcription induced by LPS overwhelms
a limited capacity to remove 3
poly(A) tail. This possibility was
addressed by analyzing TNF-
mRNA content and size after treating
RAW cells with cycloheximide, an agent that induces TNF-
transcription (30) and increases TNF-
mRNA level (31).
Cycloheximide (CHX) increased TNF-
transcription rate 80% (Fig. 5A), similar to the
increases stimulated by LPS, and caused increased steady state TNF-
mRNA levels, albeit less than with LPS. Interestingly, all of the
TNF-
mRNA present in cycloheximide-treated macrophages was the
size of hypoadenylated TNF-
mRNA in unstimulated cells (Fig.
5B, lanes 2-6), suggesting that a discrete
signal provided by LPS and not by cycloheximide is needed to allow
formation of polyadenylated TNF-
mRNA. Alternatively,
cycloheximide could directly block the capacity of the macrophage to
accumulate polyadenylated TNF-
mRNA. To test this possibility,
RAW cells were first treated with 10 µg/ml cycloheximide for 3 h
and then with LPS. At this concentration, cycloheximide inhibited
protein synthesis >96% in these cells (data not shown). The
cycloheximide-induced TNF-
mRNA was the same size as
hypoadenylated TNF-
mRNA, but the TNF-
transcripts lengthened
by approximately 200 nucleotides after the addition of LPS to the
cycloheximide-pretreated cells (Fig. 5C, lanes
2-4). Thus, the cycloheximide-treated cells were still capable of
synthesizing fully polyadenylated TNF-
mRNA in response to
subsequent LPS stimulus. Furthermore, these data also suggest that
regulation of TNF-
poly(A) tail length can be dissociated from
TNF-
transcription and is independent of protein synthesis.
The loss of 3 poly(A) tails on TNF-
transcripts in unstimulated
macrophages may be the result (32), rather than the cause of their
failure to associate with polysomes and initiate translation. If
translational silencing and dissociation from polysomes were the
primary event causing deadenylation of TNF-
mRNA, then treatment with the polysome-disrupting agent puromycin would block the appearance of polyadenylated TNF-
mRNA after LPS treatment. In fact,
treating RAW cells with puromycin at concentrations (100 µg/ml) that
blocked protein synthesis (data not shown) did not prevent the
appearance of fully polyadenylated TNF-
mRNA after LPS
stimulation (Fig. 5D). These data demonstrate that
translational inhibitors that preserve (cycloheximide) or disrupt
(puromycin) polysome integrity fail to block accumulation of
polyadenylated TNF-
mRNA after stimulation with LPS. Thus, it is
unlikely that the deadenylation of TNF-
mRNA in unstimulated
macrophages is caused by either ongoing translation or its
dissociation from polysomes.
Regulation of transcript poly(A) tail length is used in
the developing oocyte and early embryo to orchestrate transcript
translation (33). In these developmental systems, the truncated 3
poly(A) tail of translationally silent transcripts is lengthened prior to an increase in their translational efficiency (34-37). The effect of LPS stimulation on the size of preexisting cytoplasmic TNF-
mRNA was analyzed in RAW cells treated with 5 µg/ml actinomycin D
to block new TNF-
mRNA synthesis (Fig.
6, lanes 6-10). Blocking transcription unmasked a LPS-induced lengthening of the cytoplasmic pool of hypoadenylated TNF-
mRNA (lanes 8-10).
The TNF-
transcripts reached a maximal size that was intermediate
between hypoadenylated and fully adenylated TNF-
mRNA. The
TNF-
mRNA levels in these cells were not sufficient for polysome
analysis or oligo(dT)/RNase H digestion, but these results suggest that
LPS may stimulate the readenylation of preexisting cytoplasmic pools of
TNF-
mRNA. Despite >97% inhibition of transcription in RAW
cells treated with 5 µg/ml actinomycin D for 10 min, these cells were
still capable of limited TNF-
secretion (Table
I).
|
Macrophage activation causes dramatic changes in
posttranscriptional processing of TNF- mRNA (16, 17). As
previously reported in other macrophage models (11, 16, 17), basal TNF-
protein secretion was not detectable in adherent RAW
macrophages (Fig. 1C) in the face of active TNF-
transcription (Fig. 1B) and the accumulation of cytoplasmic
TNF-
mRNA (Fig. 1A). We also could not detect TNF-
protein in unstimulated RAW cell lysates (data not shown). Han et
al. (17) directly showed that translation of TNF-
was blocked
in unstimulated RAW cells and LPS stimulation reversed the inhibition
of TNF-
protein synthesis. We have extended these data by showing
that virtually all detectable TNF-
mRNA in unstimulated RAW
macrophages was not associated with polysomes (Fig. 4), indicating
TNF-
translation initiation was blocked in these cells (38). LPS
treatment induced both the appearance of polysome-associated TNF-
mRNA (Fig. 4) and onset of TNF-
secretion (Fig.
1C).
We previously reported preliminary evidence suggesting TNF-
transcripts are shorter in unstimulated adherent RAW macrophages than
in LPS-stimulated cells (18) and now report similar findings in human
monocyte-derived macrophages and three other mouse macrophage cell
lines (Fig. 1). Ulich et al. (19) reported similar TNF-
mRNA size heterogeneity in rat alveolar macrophages in
vitro and in rat spleen and liver in vivo, but the
mechanisms by which these shifts in apparent TNF-
mRNA length
occur were not described. In each of these cases TNF-
protein was
not detectable in the absence of longer TNF-
transcripts. In the
present study, enzymatic removal of the 3
poly(A) tail reduced the
apparent length of TNF-
mRNA in LPS-stimulated cells by
approximately 200 nucleotides. The same digestion of RNA from
unstimulated cells did not cause any detectable change in the apparent
1.6-kb TNF-
mRNA (Fig. 2C). Oligo(dT)/RNase H
digestion shortens poly(A) tails to within 20 nucleotides of the
transcript body (35), and the resolution of our electrophoresis system
was 20-30 nucleotides. Thus, these data indicate that the TNF-
mRNA 3
poly(A) tail is <30 nucleotides long in unstimulated
macrophages. Our inability to amplify TNF-
from unstimulated RAW
cells using oligo(dT)12-18-primed RT-PCR (data not shown)
suggests the TNF-
poly(A) tail may be even shorter in these cells.
In contrast, poly(A) tail length of the constitutively expressed
housekeeping gene transcript GAPDH was unaffected by the macrophage
activation state (Fig. 2C). A similar analysis of other
cytokine transcripts, including granulocyte/macrophage colony-stimulating factor, interleukin-6, and interleukin-1
, was not
possible in this study because these transcripts were not detectable in
unstimulated RAW cells (data not shown).
In unstimulated RAW cells, the apparent TNF- transcription rate
measured by nuclear run-on assay was almost one-third the maximal rate
achieved in LPS-stimulated cells (Fig. 1B). Biragyn and
Nedospasov (15) recently reported that elongation of TNF-
transcripts is partially blocked in unstimulated macrophages despite modest rates of transcription initiation. These authors suggested that
this process may limit effective transcription in unstimulated macrophages. Because the TNF-
gene contains a single poly(A) signal
site near its 3
terminus, incomplete transcription could account for
the lack of 3
polyadenylation. Three lines of evidence suggest that
this is not the major mechanism responsible for the accumulation of
hypoadenylated TNF-
mRNA in unstimulated macrophages. First,
Northern analysis of cytoplasmic RNA from unstimulated RAW cells only
showed a single 1.6-kb TNF-
band. Second, RT-PCR amplification of
cytoplasmic RNA using a primer complementary to the TNF-
3
terminus
generated an amplicon of the expected size in unstimulated macrophages.
Finally, the nuclear run-on assay that demonstrated basal TNF-
transcription utilized an exon 4 TNF-
cDNA fragment to detect
radiolabeled transcripts (Fig. 1B).
Detection of only 1.6-kb TNF- mRNA in unstimulated macrophages,
in the face of ongoing TNF-
transcription, suggests either a block
in nuclear 3
polyadenylation of TNF-
mRNA or rapid TNF-
transcript deadenylation in these cells. Specific cytoplasmic deadenylation of polyadenylated mRNA has been described in
mammalian cells (39, 40). Poly(A) tails of mRNA are nearly
ubiquitous in higher eukaryotes. The only mRNAs known to lack
poly(A) are those encoding the major histones that possess unique 3
secondary structure that is not present in TNF-
mRNA (41). To
our knowledge, synthesis and nuclear export of both adenylated and
deadenylated forms of the same transcript have not been reported.
Therefore, we believe that TNF-
mRNA is synthesized and exported
to the cytoplasm with a 3
poly(A) tail, which is rapidly removed in the cytoplasm of unstimulated macrophages.
The process responsible for the accumulation of deadenylated
TNF- mRNA in unstimulated RAW cells appears to be distinct from the gradual poly(A) tail shortening of polyadenylated TNF-
mRNA in LPS-stimulated cells. Gradual TNF-
mRNA poly(A) tail
shortening is evident within 30 min of LPS stimulation (Fig.
2C) and proceeds over the next 8 h in RAW cells as we
have previously described (25), indicating a typical distributive,
translation-dependent deadenylation process (42). By
comparison, only hypoadenylated, ribonucleoprotein-associated TNF-
mRNA was detectable in unstimulated macrophages (Figs. 1,
2C, and 4).
Surprisingly, hypoadenylated TNF- mRNA in unstimulated RAW cells
was more stable (Fig. 3C) than polyadenylated TNF-
mRNA in LPS-stimulated cells (25). Generally, mRNA
deadenylation is followed by 5
cap removal and transcript degradation
(43). The relative stability of the hypoadenylated TNF-
mRNA
pool may be due to its translational dormancy, since at least some
RNases are polysome-associated (44). By failing to associate with
polysomes, hypoadenylated TNF-
mRNA in unstimulated macrophages
may not be accessible to these RNases.
The data presented in this report suggest that unstimulated macrophages
transcribe TNF- but convert TNF-
mRNA to a relatively stable,
hypoadenylated, translationally dormant form. The possible importance
of this process as a negative regulator of TNF-
expression in
unstimulated macrophages is apparent. The advantage to the host of this
ostensibly inefficient process of synthesizing translationally dormant
TNF-
mRNA is less obvious, but its presence suggests that the
pool of hypoadenylated TNF-
mRNA may serve an important function.
Gene regulation in developmental systems may provide clues about the
role of TNF- mRNA deadenylation in macrophages. Translational activation of certain maternal genes during oocyte maturation, fertilization, and early embryogenesis is orchestrated by cytoplasmic deadenylation and poly(A) elongation (33). For certain genes, premature
expression is prevented by truncating the poly(A) tails of newly
synthesized transcripts before translation is initiated (45). Stable
pools of these translationally dormant, hypoadenylated transcripts are
stored for eventual expression. Translation is subsequently initiated
by poly(A) tail elongation and continues until the 3
poly(A) tail is
removed. In many ways TNF-
mRNA in macrophages behaves like the
developmental mRNAs that are regulated by cytoplasmic
polyadenylation. Unstimulated macrophages contain only hypoadenylated
cytoplasmic TNF-
mRNA, which is relatively stable and
translationally silent (17) (Figs. 1, 2, 3, 4). Moreover, in macrophages
pretreated with actinomycin D to block synthesis of new TNF-
mRNA, LPS stimulated an elongation of TNF-
mRNA (Fig. 6) and
low level secretion of TNF-
protein (Table I), suggesting TNF-
mRNA may undergo polyadenylation and translation in the macrophage
cytoplasm. This process may provide the macrophage with a means of
rapid, limited TNF-
synthesis. Furthermore, this alternative pathway
may allow TNF-
synthesis to occur in the face of transcriptional
arrest. Cytoplasmic deadenylation/polyadenylation in developmental
systems requires the canonical nuclear polyadenylation signal, AAUAAA,
and a second, less defined U-rich 3
-untranslated region element that
is similar to several stretches of AU-rich sequences in the TNF-
3
-untranslated region (46).
We have identified two possible new regulatory sites in TNF-
expression, 3
poly(A) tail metabolism and association of TNF-
mRNA with polysomes. Additionally, macrophages may be able to readenylate cytoplasmic pools of deadenylated TNF-
mRNA, which may provide an alternative, transcription-independent source of translationally active TNF-
mRNA (Fig.
7).
We thank Drs. Dhan Kalvakolanu, Moon Shin, and Sergei Nedospasov for reviews and suggestions.