Báo cáo khoa học: Deadenylation of interferon-b mRNA is mediated by both the AU-rich element in the 3¢-untranslated region and an instability sequence in the coding region pot

Deadenylation of interferon-b mRNA is mediated by both
the AU-rich element in the 3¢-untranslated region and an instability
sequence in the coding region
Muriel Paste
´
, Georges Huez and Ve
´
ronique Kruys
Laboratoire de Chimie Biologique, Institut de Biologie et de Me
´
decine Mole
´
culaires, Universite

´
Libre de Bruxelles, Belgium
Viral infection of fibroblastic and endothelial cells leads to
the transient synthesis of interferon-b (IFN-b). The down-
regulation of IFN-b synthesis after infection results both
from transcriptional repression of the IFN-b gene and rapid
degradation of mRNA. As with many cytokine mRNAs,
IFN-b mRNA contains an AU-rich element (ARE) in its
3¢-untranslated region (UTR). AREs are known to mediate
mRNA deadenylation and destabilization. Depending on
the class of ARE, deadenylation was shown to occur
through synchronous or asynchronous mechanisms. In this

study, we analysed IFN-b mRNA deadenylation in natural
conditions of IFN-b synthesis, e.g., after viral infection. We
show that human IFN-b mRNA follows an asynchronous
deadenylation pathway typical of a mRNA containing a
class II ARE. A deletion analysis of the IFN-b natural
transcript demonstrates that poly(A) shortening can be
mediated by the ARE but also by a 32 nucleotide-sequence
located in the coding region, that was identified previously as
an instability determinant. In fact, these elements are able to
act independently as both of them have to be removed to
abrogate mRNA deadenylation. Our data also indicate that
deadenylation occurs independently of mRNA translation.

Moreover, we show that deadenylation of IFN-b mRNA is
not under the control of viral infection as IFN-b mRNA
derived from a constitutively expressed gene cassette is
deadenylated in absence of viral infection. Finally, an
unidentified nuclear event appears to be a prerequisite for
IFN-b mRNA deadenylation as IFN-b mRNA introduced
directly into the cytoplasm does not undergo deadenylation.
In conclusion, our study demonstrates that IFN-b mRNA
poly(A) shortening is under the control of two cis-acting
elements recruiting a deadenylating machinery whose
activity is independent of translation and viral infection but
might require a nuclear event.

Keywords: mRNA stability; polyadenylation; translation.
The transient expression of human interferon-b (IFN-b)in
response to double stranded RNA or viral infection is a
direct consequence of transcriptional activation [1] and leads
to the accumulation of mRNA. In contrast, the shutoff of
IFN-b gene expression involves the induction of a tran-
scriptional repressor as well as a rapid decay of IFN-b
mRNA [2,3]. The human IFN-b mRNA contains an AU-
rich element (ARE) in its 3¢-untranslated region (3¢UTR).
AREs were first discovered as highly conserved elements
present in the 3¢UTR of mRNAs encoding cytokines and
oncoproteins [4]. These motifs composed of the AUUUA

pentamer, were shown to confer mRNA instability and to
regulate mRNA translation [5]. Indeed, Shaw and Kamen
first reported that the ARE located in the 3¢UTR of the
granulocyte macrophage-colony stimulating factor (GM-
CSF) mRNA was responsible for mRNA rapid degradation
[6]. Later on, the destabilizing activity of several other AREs
was documented (for review, ref [7]). AREs have been
classified into three distinct categories based on the number
and distribution of AUUUA pentamers. Class I AREs are
characterized by the presence of one to three pentamers
distributed into a large part of the 3¢UTR coupled with a
nearby U-rich region. Class II AREs have at least two

overlapping copies of the nonamer UUAUUU(U/A)
(U/A)U in a U-rich environment and class III do not
contain any pentamers but present U-rich stretches. AREs
from all three classes confer mRNA instability in cultured
cells through different mechanisms that all imply mRNA
deadenylation (for review, see [7]). Class II AREs (e.g.,
GM-CSF, TNF-a, and IL-3) induce asynchronous dead-
enylation resulting in the accumulation of poly(A)
–
inter-
mediates. In contrast, class I and class III AREs (e.g., c-fos
and c-jun) direct a synchronous poly(A) shortening. Several

ARE-binding proteins have been identified, among which
AUF1 and the tristetraprolin (TTP) were shown to
participate in the destabilization of ARE-containing
mRNAs. Recently, Chen et al. showed that ARE-binding
proteins such as AUF1 and TTP were able to interact with a
multiprotein complex, called the exosome [8]. This complex
first discovered in yeast [9], is composed of proteins with
ribonuclease activity and is able to direct 3¢)5¢ mRNA
degradation. Therefore, the recruitment of the exosome by
ARE-binding proteins might account for the degradation of
ARE-containing mRNAs.
Correspondence to V. Kruys, Laboratoire de Chimie Biologique,

Institut de Biologie et de Me
´
decine Mole
´
culaires, Universite
´
Libre
de Bruxelles, 12 rue des Profs. Jeener et Brachet, 6041 Gosselies,
Belgium. Fax: +32 2 6509800, Tel.: +32 2 6509804,
E-mail:
Abbreviations: ARE, AU-rich element; IFN-b, interferon-b;UTR,
untranslated region; CRID, coding region instability.

(Received 30 December 2002, revised 18 February 2003,
accepted 20 February 2003)
Eur. J. Biochem. 270, 1590–1597 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03530.x
Based on its sequence, IFN-b ARE belongs to class II.
Moreover, several reports described IFN-b mRNA insta-
bility and the involvement of the ARE in this process
[10,11]. In addition, another destabilizing sequence was
identified within the coding region of IFN-b mRNA [12,13].
Whereas the independent removal of the ARE or the coding
region instability determinant (CRID) result in a moderate
stabilization of the mRNA, replacement of both elements
by control sequences greatly enhances mRNA half-life [13].

It should be mentioned however, that these observations
were made in heterologous cell systems using reporter DNA
constructs under the control of heterologous or IFN-b
modified promoters. Moreover, the role of the instability
determinants in the control of IFN-b mRNA deadenylation
was not addressed.
In the present study, we analysed the expression of
endogenous IFN-b in human cells upon viral infection. We
also investigated the influence of the ARE and the CRID on
the poly(A) status of the human IFN-b mRNA in natural
conditions of IFN-b synthesis.
Materials and methods

Reagents
All the reagents and enzymes used in this study were
purchased from Roche Molecular Biochemicals and Life
Technologies Inc., unless specified. The Sendaı
¨
virus (Can-
tell strain, ATCC VR-907 Parainfluenza 1) was obtained
from Charles River Laboratories. [a-
32
P]UTP, [a-
35
S]UTP

and [
35
S]-Met were purchased from Amersham-Pharmacia
Biotech. The anti-IFN-b ELISA kit was purchased from
Biosource. Rat monoclonal anti-HA Ig (clone 3F10) was
purchased from Roche Molecular Biochemicals.
Plasmid Construction
The complete sequence of the human IFN-b gene including
the IFN-b promoter (EcoRI/EcoRI fragment described in
reference [14]) was inserted in the pcDNA3 plasmid
(Invitrogen) from which the cytomegalovirus (CMV) pro-
moter was deleted previously. In the pIFNHA construct,

the IFN-b gene was tagged by PCR using an oligonucleotide
containing three repetitions of the sequence corresponding
to the human influenza A virus hemaglutinin (HA) epitope.
The pIFNHAAU
–
construct was generated by deleting the
75-nucleotide region corresponding to the ARE (from
nucleotides 740–815 of the mRNA). The pIFNCRIDHA
and pIFNCRIDHAAU
–
constructs were obtained by
deletion of a 32-nucleotide region (from 513–545) in the

PIFNHA and PIFNHAAU
–
constructs, respectively.
A stable hairpin (hp) structure obtained by oligomeriza-
tion of a SalI linker was inserted in the HincII site of the
pIFNHA construct located at the beginning of IFN-b
mRNA 5¢UTR. To place the IFN-b gene under the
transcriptional control of a constitutive promoter, the
HA-tagged IFN-b cDNA was inserted between the EcoRI
and BamHI sites of the pSG5 plasmid (Stratagene)
(pSG5IFNHA) downstream of the simian virus 40 (SV40)
promoter.

For in vitro transcription, the pBSIFNpA vector was
generated as follows. The poly(A) tail was obtained by
hybridization of a 15-A and a 15-T oligonucleotide. The
single stranded extremities were filled with the Klenow
polymerase before oligomerization. The DNA fragments
were cloned in the T4 DNA polymerase blunted PstIsiteof
the pSP65 vector. The length of the inserted fragments was
estimated on agarose gel and the vector containing a 100–
150 nucleotides insert was selected. The poly(A)
100)150
tail
was then cloned in the HindIII/SalI sites of the pBluescript

SK (Stratagene). The restriction sites between SacIandPstI
were deleted in this pBluescript SK poly(A) and the IFN-b
gene without its promoter (EcoRI/BamHI fragment from
the pSP65IFNc plasmid described elsewhere [5]) was then
cloned between the EcoRI and HindIII sites. The deletion of
the ARE was performed by inserting a EcoRI/NdeI
fragment of the IFN-b gene [5].
Cell culture and treatments
The human endometrial adenocarcinoma cells (Hec-1B,
ATCC number, HTB-113) were maintained in DMEM
containing 10% of fetal bovine serum (FBS; Myoclone
Super Plus, Life Technologies) and 1% of penicillin/

streptomycin. The cells were infected by addition of
80 UÆmL
)1
of Sendaı
¨
virus during 2 h. Actinomycin D
and cycloheximide were used at final concentrations of
5 lgÆmL
)1
and 10 lgÆmL
)1
, respectively.

Isolation of total RNA and RNase H treatment
Total RNA was prepared by the Trizol method (Life
technologies, Inc.). RNase H treatment was performed
according to the method described by McGrew et al. [15].
Northern blot analysis
Northern blot analysis was performed as described by
Kruys et al. [16]. Total RNA (10 lg per lane) was separated
by electrophoresis in a 2.2% agarose gel, electrotransferred
to nylon membrane and cross-linked by UV-irradiation.
Blots were hybridized with antisense [a-
32
P]UTP or

[a-
35
S]UTP labelled riboprobes.
In vitro
Transcription and translation
DNAs were linearized at unique restriction sites and capped
mRNA were generated by in vitro transcription with T3 or
Sp6 polymerases. RNA was quantified by absorbance at
260 nm and its integrity was verified by agarose gel
electrophoresis followed by ethidium bromide staining.
Translation was carried out in rabbit reticulocyte lysate
(Promega) in the presence of

35
S-labelled Met (Amersham
Pharmacia Biotech).
DNA and RNA transfection
Hec-1B cells were transfected with DNA using the Fugene
reagent (Life technologies) following the procedure provided
by the supplier. RNA transfections were carried out using
the lipofectine reagent (Life technologies) as described by the
supplier. In brief, cells were grown to 50% confluency in six-
well plates before transfection. The culture medium was then
replaced by serum-free medium and the transfection mix was
Ó FEBS 2003 Sequences governing IFN-b mRNA deadenylation (Eur. J. Biochem. 270) 1591

added. The transfection mix contained 10 lgofRNA
(between 10–100 ng of the in vitro transcribed mRNA
supplemented by a carrier tRNA) and 10 lL of lipofectine in
a total volume of 200 lL of serum free-medium.
In both cases, the culture media were harvested to
measure IFN-b concentration by ELISA and the cells were
harvested for total RNA extraction.
Metabolic protein labeling and immunoprecipitation
Hec-1B cells were plated in six-well plates at 200 000 cells
per well and were incubated for 6 h before transfection.
After transfection, the cells were incubated for another 24 h
and then infected by Sendaı

¨
virus for 2 h. Cells were washed
and preincubated in a Met and Cys-depleted medium for
1 h. Metabolic labeling was performed by adding
500 lCiÆmL
)1
of
35
S-labelled Met and Cys in the cell
culture for 5 h. The cell culture medium was harvested for
immunoprecipitation. Immunoprecipitation was performed
in RIPA buffer (25 m

M
Tris pH 8.2, 50 m
M
NaCl, 0.5%
Nonidet P40, 0.5% deoxycholate, 0.1% SDS) using an anti-
HA Ig and protein A-Sepharose. Proteins were analysed by
SDS/PAGE followed by autoradiography.
Results
Deadenylation of the human IFN-b mRNA
in virus-infected cells
So far, all the studies aimed at understanding the post-
transcriptional regulation of human IFN-b mRNA have

been performed in heterologous cell systems. Therefore, we
chose to analyse the regulation of human IFN-b mRNA in
human cells (Hec-1B) that naturally produce IFN-b upon
viral infection [17]. We first performed a kinetic analysis of
IFN-b production by Hec-1B cells after infection by the
Sendaı
¨
virus for 2 h. IFN-b appeared in the cell culture
2–5 h after the infection, reaching a maximum between
8–11 h and subsequently levels droped at later times
(Fig. 1A). We then analysed, by Northern blot, the induc-
tion and decay of IFN-b transcript in the same conditions.

AsshowninFig.1B,IFN-b mRNA was detectable 4 h
after the beginning of the infection, reached a maximum
after 6–7 h and then rapidly disappeared thereafter. Inter-
estingly, two IFN-b mRNA species were observed, the
shorter form appearing later in the infection process. As
class II AREs are known to mediate mRNA degradation by
Fig. 1. Interferon-b production by Hec-1B cells infected by Sendaı
¨
virus. Hec-1B cells were infected for 2 h by the Sendaı
¨
virus, the cells were then
washed with NaCl/P

i
and fresh medium was added. (A) Every 3 h, the supernatant was sampled and replaced by fresh culture medium. The IFN-b
was quantified in the supernatants by ELISA. (B) Cells were harvested for total RNA extraction 0, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13 h after infection.
The amount and the length of IFN-b mRNA was analysed by Northern blot, using a
32
P-labelled antisense IFN-b riboprobe. As a control, the
membrane was hybridized with a GAPDH antisense riboprobe. (C) Total RNA of cells infected with the Sendaı
¨
for various lengths of time was
digested (or not) by RNase H in the presence of oligo(dT). Treated and untreated RNAs were analysed by Northern blotting. (D) Hec-1B cells were
infected by the Sendaı
¨

virus and further cultured with cycloheximide (10 lgÆmL
)1
). At the indicated times after infection, total RNA was extracted
and analysed by Northern blot. (E) Hec-1B cells were infected by the Sendaı
¨
virus and further cultured with cycloheximide (10 lgÆmL
)1
)and
actinomycin D (5 lgÆmL
)1
). At the indicated times after induction, total RNA was extracted and analysed by Northern blot with a
32

P-labelled
antisense IFN-b riboprobe. The results presented in A, B and C are representative of more than five independent experiments. Data in D and E are
representative of two independent experiments.
1592 M. Paste
´
et al. (Eur. J. Biochem. 270) Ó FEBS 2003
promoting deadenylation, we investigated whether the two
transcripts differed by the length of their poly(A) tail.
Therefore, total RNA of Hec-1B cells infected by the Sendaı
¨
virus was treated (or not) by RNAse H in the presence of
oligo(dT) before the Northern blot analysis with an IFN-b

probe. This treatment led to the detection of a single band
comigrating with the short IFN-b transcript in untreated
samples (Fig. 1C), indicating that the large and short
transcripts observed in untreated samples corresponded to
polyadenylated and deadenylated mRNAs, respectively.
The poly(A) tail of IFN-b mRNA was estimated to be about
200 nucleotides long based on the difference of electropho-
retic migration between the adenylated and deadenylated
IFN-b mRNA (Fig. 1B). We then analysed the effect of
cycloheximide on the accumulation of the two IFN-b
transcripts to determine whether IFN-b mRNA deadenyla-
tion process required ongoing translation. As shown in

Fig. 1D, the addition of the translation inhibitor after the
infection of the Hec-1B cells did not prevent the appearance
of the IFN-b short transcript. Moreover, as reported
previously, cycloheximide led to a marked increase of
IFN-b mRNA accumulation at later times of infection
resulting from mRNA stabilization and/or absence of
transcriptional repression [10,12]. Treatment by both actino-
mycin D and cycloheximide did not prevent IFN-b mRNA
deadenylation either as the polyadenylated IFN-b transcript
accumulated in response to the viral infection was also
shortened before being degraded. Moreover, the transcrip-
tional blockade by actinomycin D abrogated the increase of

mRNA accumulation due to the cycloheximide (Fig. 1E).
This latter observation indicates that increased accumulation
of IFN-b mRNA in cycloheximide-treated cells is due to the
absence of transcriptional repression of the IFN-b promoter.
Altogether, these results indicate that viral infection
triggers the synthesis of a polyadenylated IFN-b mRNA
that is deadenylated rapidly before degradation. Moreover,
this deadenylation process does not require IFN-b mRNA
translation and/or protein synthesis as it is effective in the
presence of a translational inhibitor.
Deadenylation of IFN-b mRNA occurs when IFN-b
synthesis is induced by other agents such as synthetic

double-stranded polyriboinosinic polyribocitydylic acid
(poly rI.rC) and was observed in other cell types such as
Namalwa B cells (data not shown). These observations
indicate that deadenylation is a general mechanism con-
trolling the length of IFN-b mRNA poly(A) tail.
IFN-b mRNA deadenylation is mediated by both
the ARE and the CRID
The IFN-b mRNA contains in its 3¢UTR an AU-rich
element (ARE) which is very similar to AREs present in
other unstable mRNAs [4]. As AREs present in other
cytokine mRNAs were demonstrated to induce mRNA
degradation by triggering poly(A) shortening, we first

analyzed the role of such an element in the deadenylation
process of IFN-b mRNA. To this end, two DNA
constructs were generated in which the IFN-b gene
contained or not the ARE (pIFNHA and pIFNHAAU
–
).
In addition, the sequence encoding the HA epitope was
inserted at the end of the IFN-b coding sequence to
distinguish the products resulting from the expression of the
DNA constructs and the endogenous gene (Fig. 2). These
constructs were transfected in Hec-1B cells and the cells
were subsequently infected with the Sendaı

¨
virus for 2 h.
Cells were lyzed 3 or 8 h after infection to extract the RNA
which was treated (or not) by RNAse H in the presence of
oligo(dT) before the Northern blot analysis with a HA
antisense riboprobe. As shown in Fig. 3A (lanes 3 and 7),
the HA-tagged IFN-b transcript underwent significant
deadenylation 8 h after infection independently of the
presence or the absence of the ARE. Another RNA
instability determinant was identified in the 3¢-end of the
IFN-b coding region [13]. This element named CRID
(coding region instability determinant), has been mapped

between nucleotides 513–545, the first nucleotide corres-
ponding to the adenosine of the initiation codon. Deletion
of this element by itself from the IFN-b gene (pIFNCRI-
DHA) did not abolish mRNA deadenylation (Fig. 3B,
compare lanes 2 and 3). However, deletion of both the
ARE and the CRID (pIFNCRIDHAAU-) led to a
blockade of the deadenylation process (Fig. 3B, compare
lanes 6 and 7). These results demonstrate that deadenyla-
tion is controlled by both the ARE and the CRID.
Fig. 2. Schematic representation of the DNA
constructs.
Ó FEBS 2003 Sequences governing IFN-b mRNA deadenylation (Eur. J. Biochem. 270) 1593

Deadenylation of IFN-b mRNA is uncoupled
from its translation
The role of translation in ARE-mediated mRNA deadeny-
lation and subsequent decay is still a subject of controversy.
Indeed, whileseveralreports supportatranslation-dependent
mechanism [18–20], other observations deny any coupling
between the recruitment of the mRNA into polysomes and
its deadenylation/degradation [7].
Here, we analysed whether ongoing translation is a
prerequisite for IFN-b mRNA deadenylation. Therefore,
we generated a IFN-b gene construct containing a stable
hairpininthe5¢UTR (pIFNHAhp, Fig. 2) and the deadeny-

lation of the derived mRNA was compared to that of the
mRNA lacking such a secondary structure (pIFNHA). As
shown in Fig. 4A, the presence of the hairpin in the 5¢UTR
does not influence the deadenylation process. To verify that
the hairpin effectively prevented the translation of the
mRNA, the secretion of HA-tagged IFN-b was monitored
in the culture medium of cells transfected with these con-
structs. Whereas cells transfected with the construct lacking
the hairpin produced detectable amounts of HA-tagged
IFN-b, no translation product was detectable with the
construct containing the hairpin in the 5¢UTR (Fig. 4B).
Deadenylation of IFN-b mRNA occurs independently

of viral infection
We then analysed whether IFN-b mRNA deadenylation
resulted from the infection of the cells by the Sendaı
¨
virus.
In order to ensure the production of IFN-b transcripts in
absence of infection, the HA-tagged IFN-b gene was placed
downstream of the SV40 early promoter (pSG5IFNHA,
Fig. 2). Hec-1B cells were transfected with the pSG5IF-
NHA construct and were subsequently infected (or not)
with the Sendaı
¨

virus. Deadenylation of the HA-tagged
IFN-b mRNA was monitored in the presence of actino-
mycin D to block further accumulation of HA-tagged
IFN-b mRNA. As shown in Fig. 5, deadenylation of the
HA-IFN-b transcript occurs even in absence of viral
infection.
Deadenylation of IFN-b mRNA requires a nuclear event
We next determined whether IFN-b mRNA deadenylation
requires a nuclear event. To approach this question, a
synthetic IFN-b transcript containing a poly(A) tail of
 100–150 residues was generated by in vitro transcription
in the presence of

32
P-labelled UTP (see Materials and
Fig. 4. Deadenylation of IFN-b mRNA is independent of translation. (A) Deadenylation analysis of the PIFNHA and PIFNHA hp transcripts.
Hec-1B cells were transfected with the PIFNHA and PIFNHA hp DNA constructs. Cells were harvested for RNA extraction 3 h (lanes 1, 2, 5, 6)
and 8 h (lanes 3, 4, 7, 8) after infection with the Sendaı
¨
virus. Half of each RNA sample was treated with RNAse H (lanes 1, 4, 5, 8) before Northern
blot analysis with a
35
S-labelled HA antisense riboprobe. (B) Cells transfected with the PIFNHA construct (lane 1), and PIFNHAhp construct (lane
2) were cultured in methionine and cysteine-depleted medium in the presence of 500 lCiÆmL
)1

of
35
S-labelled Met and Cys. The supernatants were
immunoprecipitated witn the anti-HA Ig and the radiolabelled proteins were analysed by SDS/PAGE. The results presented in this figure are
representative of three independent experiments.
Fig. 3. Deadenylation of IFN-b mRNA is abolished upon deletion of
both the ARE and the CRID. (A) Hec-1B cells were transfected with
the PIFNHA and PIFNHAAU
–
DNA constructs for 24 h and were
subsequently infected during 2 h by the Sendaı
¨

virus. Cells were har-
vested for RNA extraction 3 h (lanes 1, 2, 5, 6) and 8 h (lanes 3, 4, 7, 8)
after infection with the Sendaı
¨
virus. Half of each RNA sample was
treated with RNAse H (lanes 1, 4, 5, 8) before Northern blot analysis
with a
35
S-labelled HA antisense riboprobe. (B) The PIFNCRIDHA
and PIFNCRIDHAAU
–
constructs described in Fig. 2, were trans-

fected in Hec-1B cells. The cells were infected with the Sendaı
¨
virus for
2 h and were harvested 3 (lanes 1, 2, 5 and 6) and 8 h (lanes 3, 4, 7 and
8) after infection for RNA analysis by Northern blot using a
35
S-
labelled HA riboprobe. Lanes 1, 4, 5 and 8 correspond to deadenylated
RNAs obtained after RNase H treatment. The results presented in this
figure are representative of three independent experiments.
1594 M. Paste
´

et al. (Eur. J. Biochem. 270) Ó FEBS 2003
methods) (Fig. 6A). Hec-1B cells were transfected with this
synthetic transcript for 2 h, and total RNA was extracted at
various times after transfection to be analysed by agarose
electrophoresis and autoradiography. As shown in Fig. 6B,
the IFN-b transcript is rapidly degraded without prior
deadenylation, suggesting that IFN-b mRNA must origin-
ate from the nucleus to be a substrate of the deadenylation
process. To verify the poly(A) status of the IFN-b
transcript, we compared its migration in agarose gel to
poly(A)
–

IFN-b transcripts, containing (or not) the ARE
after transfection into Hec-1B cells. The migration of the
different transcripts confirmed that the poly(A)
+
IFN-b
mRNA bore a 100–150 nucleotides long poly(A) tail
(Fig. 6C). Moreover, in order to verify the effective
introduction of the synthetic mRNA into cells, IFN-b
production was assayed in the culture medium after
transfection. As shown in Fig. 6D (lane 1), transfection of
polyadenylated mRNA led to IFN-b synthesis (Fig. 6D,
lane 1) in contrast to the poly(A)

–
transcripts which were
poorly translated (Fig. 6D lanes 2 and 3). Similar results
were obtained when cells were infected by the Sendaı
¨
virus
before RNA transfection (data not shown).
Discussion
In the present study, we analysed the expression and the
poly(A) status of human IFN-b mRNA in human endo-
thelial Hec-1B cells in response to infection by the Sendaı
¨

virus. As observed in other cell types, IFN-b synthesis is
transiently induced and results from a strong accumulation
of IFN-b mRNA that rapidly disappears at later times of
infection [12,21]. We showed that the disappearance of
IFN-b mRNA is accompanied by the shortening of its
poly(A) tail. As described for certain class II ARE-
containing mRNAs (e.g., GM-CSF, IL-3) [22,23], IFN-b
mRNA is deadenylated asynchronously with the formation
of poly(A)
–
intermediates. However, IFN-b mRNA dead-
enylation is not solely under the control of the ARE. Indeed,

poly(A) shortening is abolished only upon deletion of both
the ARE and the CRID (Fig. 3). The CRID was identified
previously as an instability determinant that, in combina-
tion with the 3¢UTR, mediates the rapid decay of human
IFN-b mRNA in NDV-infected NIH/3T3 cells [13]. Both
sequences were shown by UV-crosslinking experiments to
recruit a cytosolic 65-kDa protein of unknown identity.
Fig. 6. Deadenylation of IFN-b mRNA is independent of viral infection.
(A) Schematic representation of the DNA constructs used to generate
in vitro transcribed IFN-b mRNA. (B)
32
P-labelled IFN-b mRNA

containing a 100 nucleotide poly(A) tail was generated by in vitro
transcription. The RNA was transfected for 2 h into Hec-1B cells and
total RNA was extracted at the indicated times after the end of
transfection.
32
P-labelled IFN-b mRNA was analysed by agarose gel
electrophoresis and autoradiography. (C) Polyadenylated IFN-b
mRNA AU
+
was transcribed from the pBSIFNpA. The poly(A)
–
IFN-b mRNAs, AU

+
pA
–
and AU
–
pA
–
, were transcribed from the
pSP65IFN construct linearized by BamHI and NdeI, respectively. The
[
32
P]–labelled transcripts were transfected into Hec-1B cells for 8 h

before total RNA extraction, agarose gel electrophoresis and auto-
radiography. (D) The IFN-b was assayed in cell culture medium by
ELISA. The results presented in this figure are representative of four
independent experiments.
Fig. 5. In vitro transcribed IFN-b mRNA is not deadenylated upon
transfection into Hec-1B cells. The pSG5IFNHA construct was
transfected into Hec-1B cells. The cells were infected (or not) with the
Sendaı
¨
virus for 2 h and actinomycin D (5 lgÆmL
)1
)wasaddedinthe

culture medium. Total RNA was extracted at the indicated times and
analysed by Northern blot using a
35
S-labelled HA riboprobe. The
RNase H-treated samples position the fully deadenylated mRNA. The
same blot was rehybridized with a
35
S-labelled GAPDH riboprobe.
The results presented in this figure are representative of two inde-
pendent experiments.
Ó FEBS 2003 Sequences governing IFN-b mRNA deadenylation (Eur. J. Biochem. 270) 1595
These and our observations suggest that the binding of this

65-kDa protein to one of these elements might be required
to induce IFN-b mRNA deadenylation and subsequent
degradation of the RNA body.
c-fos, c-myc and plasminogen activator inhibitor (PAI-2)
messenger RNAs are other ARE-containing mRNAs
bearing instability determinants in their coding region
[24–27]. Moreover, in the case of c-fos, it was shown that
ARE mediates mRNA deadenylation by a translation-
independent mechanism, while the coding region instabi-
lity determinant facilitates mRNA deadenylation by a
mechanism coupled to translation [25,28]. Here, we show
that IFN-b mRNA deadenylation occurs independently of

the translational status of the mRNA. This observation
correlates with the fact that IFN-b mRNA destabilization
at later times of infection occurs even when mRNA
translation is abrogated by the insertion of a stop codon
immediately after the initiation codon [13]. It remains to
be established, however, whether any of the two elements
taken separately is translation-dependent in promoting
mRNA deadenylation.
IFN-b mRNA deadenylation is a constitutive mecha-
nism. Indeed, the IFN-b transcript derived from a consti-
tutively transcribed gene cassette undergoes deadenylation
in absence of viral infection. However, poly(A) shortening is

detectable only after addition of actinomycin D that blocks
the accumulation of newly synthesized polyadenylated
IFN-b mRNA (Fig. 5). This observation suggests that the
deadenylation machinery is pre-existing in the cells and
deadenylates IFN-b mRNA as soon as its synthesis is
induced by stimulating agents. The fact that the 65-kDa
protein binds ARE and CRID in UV-crosslinking experi-
ments, performed with cytosolic extracts from both non-
infected and infected cells [13], further supports the
involvement of this protein in the deadenylation process
of IFN-b mRNA. Interestingly, IFN-b ARE does not
recruit other ARE-binding factors (data not shown),

thereby emphasizing the role of this 65-kDa RNA-binding
protein [13] whose identity and function remain to be
investigated.
IFN-b mRNA deadenylation seems to be conditioned by
a nuclear event. Indeed, a synthetic IFN-b mRNA bearing a
100 nucleotide poly(A) tail escapes the deadenylation
process when transfected in Hec-1B cells (Fig. 6). The
nondeadenylation of this synthetic transcript does not
however, protect it from rapid decay, thereby suggesting
that it becomes a target of an alternative poly(A)-independ-
ent degradative pathway. Although the nuclear event
conditioning IFN-b mRNA deadenylation remains to be

established, we provide the first evidence indicating such
requirement for this mRNA degradative process. It seems
however, possible that such a nuclear event might also be
required for other mRNAs undergoing specific deadenyla-
tion. Indeed, most RNA-binding proteins mediating
mRNA deadenylation/degradation shuttle between the
nucleus and the cytoplasm [29]. The association of specific
transcripts with such factors in the nuclear compartment
might thus condition their cytoplasmic fate.
Altogether, our results and previous observations
[12,13,30] demonstrate clearly that IFN-b mRNA behaves
similarly to class II ARE-containing mRNA prototypes

(e.g., GM-CSF, IL-3). However, the deadenylation and
degradation of IFN-b mRNA is under the control of two
independent elements, one of which is located in the mRNA
coding region.
The coexistence of two independent but apparently
redundant instability determinants might reflect the need
for stringent control of IFN-b gene, whose prolonged
expression might be detrimental to the organism.
Acknowledgements
This work was funded by the EC contract (QLK3-2000-00721), the
Fund for Medical Scientific Research (Belgium, grant 3.4618.01), and
the ÔActions de Recherches Concerte

´
esÕ (grant 00-05/250).
References
1.Hayes,T.G.,Yip,Y.K.&Vilcek,J.(1979)Leinterferonpro-
duction by human fibroblasts. Virology 98, 351–363.
2. Tan, Y.H. & Berthold, W. (1977) A mechanism for the induction
and regulation of human fibroblastoid interferon genetic expres-
sion. J. General Virol. 34, 401–411.
3. Raj, N.B. & Pitha, P.M. (1981) Analysis of interferon mRNA in
human fibroblast cells induced to produce interferon. Proc. Natl
Acad. Sci. U.SA 78, 7426–7430.
4. Caput, D., Beutler, B., Hartog, K., Thayer, R., Brown-Shimer, S.

& Cerami, A. (1986) Identification of a common nucleotide
sequence in the 3¢-untranslated region of mRNA molecules spe-
cifying inflammatory mediators. Proc. Natl Acad. Sci. USA. 83,
1670–1674.
5. Kruys,V.,Wathelet,M.,Poupart,P.,Contreras,R.,Fiers,W.,
Content, J. & Huez, G. (1987) The 3¢ untranslated region of the
human interferon-beta mRNA has an inhibitory effect on trans-
lation. Proc.NatlAcad.Sci.USA.84, 6030–6034.
6. Shaw, G. & Kamen, R. (1986) A conserved AU sequence from the
3¢ untranslated region of GM- CSF mRNA mediates selective
mRNA degradation. Cell. 46, 659–667.
7. Chen, C.Y. & Shyu, A.B. (1995) AU-rich elements: characteriza-

tion and importance in mRNA degradation. Trends Biochem. Sci.
20, 465–470.
8. Chen, C.Y., Gherzi, R., Ong, S.E., Chan, E.L., Raijmakers, R.,
Pruijn, G.J., Stoecklin, G., Moroni, C., Mann, M. & Karin, M.
(2001) AU binding proteins recruit the exosome to degrade ARE-
containing mRNAs. Cell 107, 451–464.
9.Allmang,C.,Mitchell,P.,Petfalski,E.&Tollervey,D.(2000)
Degradation of ribosomal RNA precursors by the exosome.
Nucleic Acids Res. 28, 1684–1691.
10. Raj, N.B. & Pitha, P.M. (1983) Two levels of regulation of beta-
interferon gene expression in human cells. Proc. Natl Acad. Sci.
USA 80, 3923–3927.

11. Mosca, J.D. & Pitha, P.M. (1986) Transcriptional and post-
transcriptional regulation of exogenous human beta interferon
gene in simian cells defective in interferon synthesis. Mol. Cell Biol.
6, 2279–2283.
12. Whittemore, L.A. & Maniatis, T. (1990) Postinduction turnoff of
beta-interferon gene expression. Mol. Cell. Biol. 10, 1329–1337.
13. Raj, N.B. & Pitha, P.M. (1993) 65-kDa protein binds to destabi-
lizing sequences in the IFN-beta mRNA coding and 3¢ UTR.
FASEB J. 7, 702–710.
14. Derynck, R., Content, J., DeClercq, E., Volckaert, G., Tavernier,
J., Devos, R. & Fiers, W. (1980) Isolation and structure of a
human fibroblast interferon gene. Nature, 19, 542–547.

15. McGrew, L.L., Dworkin-Rastl, E., Dworkin, M.B. & Richter,
J.D. (1989) Poly (A) elongation during Xenopus oocyte maturation
is required for translational recruitment and is mediated by a short
sequence element. Genes Dev. 3, 803–815.
1596 M. Paste
´
et al. (Eur. J. Biochem. 270) Ó FEBS 2003
16. Kruys, V., Thompson, P. & Beutler, B. (1993) Extinction of
the tumor necrosis factor locus, and of genes encoding the
lipopolysaccharide signaling pathway. J. Exp. Med. 177, 1383–
1390.
17. Chen,H.Y.,Sato,T.,Fuse,A.,Kuwata,T.&Content,J.(1981)

Resistance to interferon of a human adenocarcinoma cell line,
HEC-1, and its sensitivity to natural killer cell action. J. Gen. Virol.
52, 177–181.
18. Curatola, A.M., Nadal, M.S. & Schneider, R.J. (1995) Rapid
degradation of AU-rich element (ARE) mRNAs is activated by
ribosome transit and blocked by secondary structure at any
position 5¢-to the ARE. Mol. Cell Biol. 15, 6331–6340.
19. Veyrune, J.L., Carillo, S., Vie, A. & Blanchard, J.M. (1995) c-fos
mRNA instability determinants present within both the coding
and the 3¢ non coding region link the degradation of this mRNA
to its translation. Oncogene 11, 2127–2134.
20. Winstall, E., Gamache, M. & Raymond, V. (1995) Rapid mRNA

degradation mediated by the c-fos 3¢ AU-rich element and that
mediated by the granulocyte-macrophage colony-stimulating fac-
tor 3¢ AU-rich element occur through similar polysome-associated
mechanisms. Mol. Cell Biol. 15, 3796–3804.
21. Dehlin, E., von Gabain, A., Alm, G., Dingelmaier, R. &
Resnekov, O. (1996) Repression of beta interferon gene expression
in virus-infected cells is correlated with a poly (A) tail elongation.
Mol. Cell. Biol. 16, 468–474.
22. Carballo, E., Lai, W.S. & Blackshear, P.J. (2000) Evidence that
tristetraprolin is a physiological regulator of granulocyte-macro-
phage colony-stimulating factor messenger RNA deadenylation
and stability. Blood 95, 1891–1899.

23.Nair,A.P.,Hirsch,H.H.,Colombi,M.&Moroni,C.(1999)
Cyclosporin A promotes translational silencing of autocrine
interleukin-3 via ribosome-associated deadenylation. Mol. Cell
Biol. 19, 889–898.
24. Wisdom, R. & Lee, W. (1991) The protein-coding region of c-myc
mRNA contains a sequence that specifies rapid mRNA turnover
and induction by protein synthesis inhibitors. Genes Dev. 5, 232–
243.
25. Schiavi, S.C., Wellington, C.L., Shyu, A.B., Chen, C.Y., Green-
berg, M.E. & Belasco, J.G. (1994) Multiple elements in the c-fos
protein-coding region facilitate mRNA deadenylation and decay
by a mechanism coupled to translation. J. Biol. Chem. 269, 3441–

3448.
26. Langa, F., Lafon, I., Vandormael-Pournin, S., Vidaud, M., Bab-
inet, C. & Morello, D. (2001) Healthy mice with an altered c-myc
gene: role of the 3¢ untranslated region revisited. Oncogene 19,
4344–4353.
27. Tierney, M.J. & Medcalf, R.L. (2001) Plasminogen activator
inhibitor type 2 contains mRNA instability elements within exon 4
of the coding region. Sequence homology to coding region
instability determinants in other mRNAs. J. Biol. Chem. 276,
13675–13684.
28. Chen, C.Y., Xu, N. & Shyu, A.B. (1995) mRNA decay mediated
by two distinct AU-rich elements from c-fos and granulocyte-

macrophage colony-stimulating factor transcripts: different dead-
enylation kinetics and uncoupling from translation. Mol. Cell Biol.
15, 5777–5788.
29. Shyu, A.B. & Wilkinson, M.F. (2000) The double lives of shuttling
mRNA binding proteins. Cell 102, 135–138.
30. Mosca, J.D. & Pitha, P.M. (1986) Transcriptional and post-
transcriptional regulation of exogenous human beta interferon
gene in simian cells defective in interferon synthesis. Mol. Cell Biol.
6, 2279–2283.
Ó FEBS 2003 Sequences governing IFN-b mRNA deadenylation (Eur. J. Biochem. 270) 1597