In a previous study, we defined the functional destabi-lizing ARE element in the 3¢ UTR PAI-2 as a single nonameric AU-rich sequence UUAUUUAUU located 304 nucleotides upstream of the poly
Trang 1destabilized via a multi-component 3¢ UTR localized
adenylate and uridylate-rich instability element in an
analogous manner to cytokines and oncogenes
Stan Stasinopoulos1, Mythily Mariasegaram1, Chris Gafforini1, Yoshikuni Nagamine2and Robert
L Medcalf1
1 Monash University, Australian Centre for Blood Diseases, Melbourne, Victoria, Australia
2 Friedrich Miescher Institute for Biomedical Research, Basel, Switzerland
Introduction
The generation of the serine protease plasmin by the
plasminogen activator system is a critical event in a
variety of physiological processes, including
fibrino-lysis, development, wound healing and cell migration
[1–4] Plasmin generation is regulated by two
plasmin-ogen activators: urokinase-type plasminplasmin-ogen activator
in the extracellular environment and tissue-type
plas-minogen activator in the circulation The proteolytic activities of both tissue-type plasminogen activator and urokinase-type plasminogen activator are controlled by plasminogen activator inhibitor types 1 and 2 (PAI-1 and PAI-2, respectively) One of the enigmatic features
of PAI-2 is that, although it can inhibit extracellular and receptor-bound urokinase-type plasminogen
Keywords
3¢ untranslated region; adenylate and
uridylate-rich element; mRNA decay;
plasminogen activator inhibitor type 2
Correspondence
R Medcalf, Australian Centre for Blood
Diseases, Monash University, 6th Floor
Burnet Building, AMREP, 89 Commercial
Road, Melbourne 3004, Australia
Fax: +61 3 9903 0228
Tel: +61 3 9903 0133
E-mail: robert.medcalf@med.monash.edu.au
(Received 21 August 2009, revised 23
December 2009, accepted 28 December
2009)
doi:10.1111/j.1742-4658.2010.07563.x
Plasminogen activator inhibitor type 2 (PAI-2; SERPINB2) is a highly-regulated gene that is subject to both transcriptional and post-transcrip-tional control For the latter case, inherent PAI-2 mRNA instability was previously shown to require a nonameric adenylate-uridylate element in the 3¢ UTR However, mutation of this site was only partially effective at restoring complete mRNA stabilization In the present study, we have identified additional regulatory motifs within the 3¢ UTR that cooperate with the nonameric adenylate-uridylate element to promote mRNA destabi-lization These elements are located within a 74 nucleotide U-rich stretch (58%) of the 3¢ UTR that flanks the nonameric motif; deletion or substitu-tion of this entire region results in complete mRNA stabilizasubstitu-tion These new elements are conserved between species and optimize the destabilizing capacity with the nonameric element to ensure complete mRNA instability
in a manner analogous to some class I and II adenylate-uridylate elements present in transcripts encoding oncogenes and cytokines Hence, post-tran-scriptional regulation of the PAI-2 mRNA transcript involves an interaction between closely spaced adenylate-uridylate elements in a manner analogous
to the post-transcriptional regulation of oncogenes and cytokines
Abbreviations
ARE, adenylate and uridylate rich element; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GM-CSF, granulocyte macrophage-colony-stimulating factor; IL, interleukin; PAI-2, plasminogen activator inhibitor type 2; REMSA, RNA electrophoretic mobility shift assays; RPA, RNase protection analysis.
Trang 2activator [5,6], it exists primarily as a nonglycosylated
intracellular protein Over the past decade, evidence
has accumulated to suggest a role for PAI-2 in
intra-cellular events associated with apoptosis [7–11],
prolif-eration and differentiation [4,12], and the innate
immune response [7,13–15] PAI-2 has also generated a
substantial level of interest because of its impressive
regulatory profile It is one of the most responsive
genes known (i.e it can be induced over 1000-fold),
and is regulated in a cell type-dependent manner by
phorbol esters [16,17], the phosphatase inhibitor,
oka-daic acid [18], tumour necrosis factor a [19,20],
lipo-polysaccharide [21,22] and elevated levels of serum
lipoprotein (a) [23] Although there is a significant
transcriptional component to the regulation of PAI-2
expression by these agents, in recent years, the role of
post-transcriptional regulation has come to the fore
because a number of studies have shown that the
half-life of PAI-2 mRNA can also be altered in a treatment
and cell type-dependent manner [19,22,24–26]
Post-transcriptional control of gene expression is
particularly important for controlling the levels of
tran-siently induced transcripts Many of these transcripts
have extremely short half-lives, and this is usually
attrib-uted to the presence of adenylate and uridylate-rich
instability elements (AREs) located with the 3¢ UTR
[27] Instability regions in the 3¢ UTR can comprise
single- or multiple-ARE elements that either interact
with each other or act independently to define the fate
of a transcript in response to a specific physiological
state [28–33] AREs are usually 50–100 nucleotides in
length and contain single or multiple copies of the
consensus motif AUUUA, UUAUUUA(U⁄ A)(U ⁄ A) or
UUAUUUAUU embedded within a U-rich sequence
[34,35] AREs have been classed into three groups
(groups I, II and III), depending on their particular
AU-rich sequence content [35]
Functional studies have indicated that AREs initially
accelerate mRNA deadenylation, which is then followed
by the degradation of the mRNA body [28,36,37]
A number of in vitro studies have also reported that
both AREs and ARE-binding proteins can interact with
the exosome, which then degrades the body of the
transcript with 3¢- to 5¢ polarity [38–40] Recent in vivo
studies, however, have elucidated a mammalian 5¢- to 3¢
ARE decay pathway that is localized to P-bodies via an
ARE interaction with tristetraprolin and BRF1 [41–44]
However, both 5¢- to 3¢ and 3¢- to 5¢ pathways can be
simultaneously engaged in mRNA decay in an
ARE-mediated manner [45], suggesting that the pathway
of mammalian ARE-mediated mRNA decay can be
flexible Recently, an excellent database compiling ARE
containing transcripts was established [46] and it has
been predicted that approximately 8% of human genes code for transcript that contain AREs [47]
In a previous study, we defined the functional destabi-lizing ARE element in the 3¢ UTR PAI-2 as a single nonameric AU-rich sequence (UUAUUUAUU) located
304 nucleotides upstream of the poly(A) tail [24,48] and suggested that tristetraprolin was a candidate PAI-2-nonameric element binding protein involved in desta-bilizing the PAI-2 mRNA transcript [49] However, subsequent work from our group demonstrated that mutagenesis of the nonameric element only partially sta-bilized the b-globin-PAI-2 3¢ UTR transcript [48], sug-gesting the presence of additional functional destabilizing regions within the PAI-2 3¢ UTR In the present study, we reveal that the nonameric ARE resides within a 108 nucleotide U-rich (54%) region consisting
of three pentameric AU elements (one of which is a no-nameric motif) and one atypical AU-rich region, and that this extended region fully accounts for the complete destabilizing activity of the PAI-2 3¢ UTR Further-more, functional mapping within the 108 AU-rich region revealed that the essential destabilizing sequences, con-sisting of the first two pentameric motifs and the atypical AU-rich region, resided within a continuous 74 nucleo-tide region, which we now define as the functional PAI-2 mRNA ARE element The nonameric motif indeed com-prises the core sequence that is essential for constitutive mRNA decay; however, its optimal destabilizing activity
is only achieved in a cooperative manner with either one
of two auxiliary AREs The results obtained support the concept that AU-rich instability elements can be composed of multiple AREs that act in a synergistic manner to destabilize or stabilize transcripts depending
on the physiological status of the cell Finally, our studies show that PAI-2 mRNA harbours a spatial and functional class I ARE profile that is more analogous to that of highly-regulated cytokines and oncogenes, including granulocyte macrophage-colony-stimulating factor (GM-CSF), interleukin (IL)-8 and c-fos This may explain why the regulation of the PAI-2 gene differs so vastly from the broader family of serine proteases
Results
Mutation of the PAI-2 3¢ UTR nonameric sequence results in only partial mRNA stabilization
To re-assess the mRNA destabilizing characteristics
of the PAI-2 3¢ UTR, we established a tetracycline-regulated system to accurately determine mRNA decay rates Accordingly, we used a HT-1080 fibrosarcoma
Trang 3TET-OFF system (Clontech, Mountain View, CA, USA)
in combination with a plasmid pTETBBB (referred to as
pTETGLO), which contains the gene for b-globin under
the control of a tetracycline-regulated promoter that
allows transcription only in the absence of tetracycline or
a derivative (e.g doxycycline) [50] We cloned the
full-length wild-type PAI-2 3¢ UTR, and a mutant PAI-2
3¢ UTR containing a four nucleotide substitution within
the nonameric ARE (UUAUUUAUU to UUAAAG
CUU) sequence into the unique BglII site in the b-globin
3¢ UTR of plasmid pTETGLO to create plasmids pTET
GLOPAI)2 and pTETGLOARE II-MUT, respectively
These plasmids, including the empty vector, pTETGLO,
were transiently transfected into HT-1080 fibrosarcoma
TET-OFF cells and the decay characteristics (t1⁄ 2min)
of the various transcripts were determined after the
addi-tion of doxycycline by RNase protecaddi-tion analysis (RPA)
As shown in Fig 1, the half-life of the wild-type b-globin
transcript was greater than 480 min, beyond the end
point of the experiment (based on the composite curve of
three separate experiments presented in Fig 1),
demon-strating the high stability of this transcript The half-life
of the b-globinPAI)2transcript was reduced to158 min,
whereas the half-life of the b-globinARE II-MUTtranscript
only increased to301 min (Fig 1) This demonstrates
that mutation of this element only partially stabilized the
b-globinARE II-MUT transcript, which is in agreement
with previous studies from our laboratory using a
differ-ent mRNA decay system [48] and also supports the
hypothesis that the PAI-2 3¢ UTR contains
uncharacter-ized functional instability elements
The PAI-2 3¢ UTR mRNA destabilizing elements
are localized to a 108 nucleotide U-rich (54%)
sequence
Analysis of the PAI-2 3¢ UTR sequence (Fig 2)
revealed that the nonameric element resided within a
108 nucleotide U-rich (54%) sequence and was flanked
at the 5¢ and 3¢ ends by two classical pentameric ARE
(AUUUA) motifs and an atypical AU-rich region
(AUUUUAUAUAAU) immediately abutting 3¢ to the
nonamer This 108 nucleotide ‘extended ARE’ can, by
structure and sequence homology, be categorized as a
class I ARE element [35] Furthermore, these classical
pentameric elements could be the source of the
addi-tional destabilizing sequences within the ‘extended
ARE’ (Fig 2), which could act independently or in a
cooperative manner with the nonameric ARE
To determine whether the ‘extended ARE’ possessed
all of the destabilizing elements within the PAI-2
3¢ UTR, the entire 108 nucleotide sequence was deleted
from the 3¢ UTR to create plasmid
pTET-GLO3¢ UTRDARE This plasmid was transiently trans-fected into HT1080 TET-OFF cells and the half-life of the b-globinARED transcript was shown to be
> 480 min (Fig 3) Hence, this deletion resulted in significant mRNA stabilization, with mRNA decay kinetics reminiscent of the wild-type b-globin transcript (Fig 1) In addition, replacement of the 108 nucleotide
‘extended ARE’ with an equivalent length of an irrele-vant sequence also substantially stabilized the tran-script (data not shown) to an extent similar to that seen previously with the b-globin and the b-globinARED transcripts Moreover, cloning the 108 nucleotide
‘extended ARE’ into the BglII site in the b-globin 3¢ UTR, creating plasmid pTETGLOEXT.ARE, resulted
A
B
C
Fig 1 The PAI-2 3¢ UTR localized nonameric sequence only par-tially contributes to PAI-2 mRNA instability (A)
Rabbit-b-globin-PAI-2 3¢ UTR constructs prepared for the transient transfection of HT1080-TET OFF cells (B) HT1080-TET OFF cells were transfected with the TET-responsive b-globin reporter plasmids described in (A) After 16 h of incubation, doxycycline was added and total RNA was isolated at the indicated times and analysed by RPA The graph in (C) corresponds to the experiments shown in (B) (n = 3–6) Each point represents the mean ± SE.
Trang 4in decay characteristics similar to the b-globinPAI)2
wild-type transcript (t1⁄ 2178 min and t1⁄ 2198 min,
respectively) (Fig 4) Collectively, these experiments
demonstrate that the 108 nucleotide ‘extended ARE’
contains all the essential destabilizing elements in the
PAI-2 3¢ UTR
ARE I and III are not independent functional destabilizing elements
To assess the relative contribution of these additional ARE elements, the essential residues in ARE I and III were mutated either alone or in combination to create plasmids pTETGLOARE I-MUT, pTETGLOARE III-MUT and pTETGLOARE I+III-MUT(Fig 2) within the context
of the full-length PAI-2 3¢ UTR, and the influence of these mutations on the mRNA decay characteristics was determined As shown in Fig 5, the estimated half-lives
of these transcripts were231 min for the b-globinPAI)2 wild-type transcript,204 min for the b-globinARE I-MUT,
193 min for the b-globinARE III-MUT and 224 min for the b-globinARE I+III-MUT These experiments dem-onstrate that ARE I and III, both of which are composed of classical pentameric sequence AUUUA,
do not independently contribute to the instability of the PAI-2 transcript
ARE I acts as a functional auxiliary element to the core destabilizing ARE II site
To assess the possibility that the destabilizing activity exhibited by the ‘extended ARE’ was the result of
A
B
C
Fig 2 The PAI-2 3¢ UTR contains a 108 nucleotide functional ‘extended ARE’ A diagrammatic representation of the PAI-2 3¢ UTR showing the location and sequence of the AU-rich regions of interest within the ‘extended ARE’, and the sequences of the various ‘extended ARE’ mutants that were generated.
Fig 3 Deletion of the ‘extended’ ARE from the PAI-2 3¢ UTR results in a stabilization reminiscent to the wild-type b-globin tran-script (A) Rabbit-b-globin-PAI-2 3¢ UTR constructs prepared for the transient transfection of HT1080-TET OFF cells Plasmids pTET-GLOPAI)2, containing the full-length PAI-2 3¢ UTR, and pTET-GLO3¢ UTRDARE in which the ‘extended ARE’ is deleted (B, C) HT1080-TET OFF cells were transfected with the TET-responsive b-globin reporter plasmids described in (A) and the b-globin mRNA decay curves were quantified by RPA as described in Fig 1 and the Experimental procedures The experiments shown in (C) were repeated three times and each point represents the mean ± SE.
Trang 5cooperation between the classical AU-rich elements,
double-ARE mutants (ARE II and ARE I; or ARE II
and ARE III) were created within the context of
the full-length PAI-2 3¢ UTR to create constructs
pTETGLOARE I+II-MUT and pTETGLOARE II+III-MUT
and their decay characteristics were determined The
b-globinARE I+II-MUT transcript was significantly
stabilized (t1 ⁄ 2 > 480 min; Fig 6), to a level
reminis-cent to that seen for the b-globin (Figs 1 and 4) and
b-globinARED (Fig 3) transcripts, compared to the
wild-type b-globinPAI)2 ( 192 min) transcript in this series of experiments (Fig 6) This result suggests that ARE I is an essential functional auxiliary element
to the core destabilizing ARE II sequence and that this combination of AU-rich elements (ARE I⁄ ARE II) plays a central role in determining the half-life of the PAI-2 mRNA transcript under physiological conditions
Curiously, the half-life of the b-globinARE II+III double mutant transcript was only partially stabilized
to 347 min (Fig 6), which is also reminiscent of the half-life of 333 min for the b-globinARE II-MUT transcript (Fig 1) This implies that ARE III is unlikely to cooperate with the AREII⁄ nonameric element to contribute to the destabilizing activity
of the ‘extended ARE’ in the presence of an active ARE I
A
B
C
Fig 4 The 108 nucleotide ‘extended ARE’ independently confers
mRNA instability in an analogous manner to the PAI-2 full-length
3¢ UTR (A) Rabbit-b-globin-PAI-2 3¢ UTR constructs prepared for
the transient transfection of HT1080-TET OFF cells Plasmids
pTET-GLO, Plasmids pTETGLOPAI)2, containing the full-length PAI-2
3¢ UTR, and pTETGLO EXT.ARE containing the 108 nucleotide
‘extended ARE’ (B, C) HT1080-TET OFF cells were transfected
with the TET-responsive b-globin reporter plasmids described in (A)
and the b-globin mRNA decay curves were quantified as described
in Fig 1 and according the northern hybridization protocol (see
Experimental procedures) The experiments shown in (C) were
repeated three times and each point represents the mean ± SE.
The dotted line represents 50% mRNA remaining.
A
B
Fig 5 The PAI-2 ‘extended ARE’ contains two classical
pentamer-ic sequences (AUUUA), designated as ARE I and III, that do not independently function as instability elements (A) Rabbit-b-globin-PAI-2 3¢ UTR constructs prepared for the transient transfections of HT1080-TET OFF cells The full-length PAI-2 3¢ UTR was cloned into the 3¢ UTR of b-globin creating plasmid pTETGLOPAI)2 A five nucleotide substitution (as shown in Fig 2) was introduced into the ARE I and the ARE III pentameric sequences, individually or in com-bination, to create pTETGLO ARE I-MUT , pTETGLO ARE III-MUT and pTETGLOARE I+III-MUT (B) HT1080-TET OFF cells were transfected with the TET-responsive b-globin reporter plasmids described in (A) and the b-globin mRNA decay curves were quantified by RPA as described in Fig 1 and the Experimental procedures The experi-ments shown in (B) were repeated two or three times and each point represents the mean ± SE.
Trang 6The ‘extended ARE’ contains an alternate atypical
AU-rich auxiliary element that interacts with the
core ARE II sequence
Comparison of the human PAI-2 3¢ UTR with
those from a number of mammalian species (Fig 7)
using clustalw [50a] analyses revealed a high degree
of conservation between the ARE II (nonameric)
sequences and a 12 nucleotide atypical AU-rich
sequence (labelled ARE IV) immediately 3¢ to ARE II
To determine the extent to which this sequence
contributed to the decay rate, the same seven nucleo-tide substitution (gUUAUUUAUUaugcauuccuau) was introduced into the abutting atypical ARE IV site within the context of the full-length 3¢ UTR (Fig 2)
to create the plasmid pTETGLOARE IV-MUT As determined by our TET-regulated globin mRNA decay system, disruption of this element resulted in a half-life
of the b-globinARE IV-MUT transcript of 223 min (Fig 8) compared to the half-life of the b-globinPAI)2 wild-type transcript ( 182 min), which is unlikely to
be a significant difference Hence, ARE IV is unlikely
to function as an independent PAI-2 mRNA destabi-lizing element
To determine whether the adjacent elements (ARE
II and IV) could destabilize the transcript in an additive or cooperative manner in an analogous way
to the AREI⁄ AREII region, both the ARE II and the abutting ARE IV sequence were mutated (gUUAAAGCUUaugcauuccuau) within the context of the full-length 3¢ UTR to create the plasmid pTET-GLOARE II+IV-MUT This plasmid was transiently transfected into HT1080 TET-OFF cells and the half-life of the b-globinARE II+IV-MUT transcript was sub-stantially increased (t1⁄ 2 > 480 min) (Fig 8), which is equivalent to the high level of stability of the b-globin, the b-globinARED and the b-globinARE I+II-MUT tran-scripts (Figs 1, 3 and 6, respectively)
RNA electrophoretic mobility shift assays (REMSA) were next performed to determine whether these adja-cent ARE sites played a role in protein binding activ-ity Initial experiments confirmed that the extended wild-type ARE sequence provided specific protein binding sites for cytoplasmic proteins extracted from HT1080 TET-OFF cells (Fig S1A) Subsequent analy-ses further indicate that mutations introduced into ARE II substantially reduced protein binding activity, which is consistent with our previous results using shorter RNA probes [48] However, mutations intro-duced into the adjacent ARE IV had only a minimal effect on binding activity When both the ARE II and
IV sites were mutated simultaneously, binding activity was reduced to the level seen with mutations in ARE
II alone (Fig S1B) Hence, ARE IV does not appear
to modulate protein binding activity to the ‘extended ARE’, despite the fact that it contributes to mRNA stability Whether this is a consequence of the limita-tion of the REMSA approach or the influence of alternative functional AREs (e.g ARE I) remains unknown
Taken together, the results obtained in the present study suggest that the functional PAI-2 3¢ UTR insta-bility sequence consists of an essential core nonameric sequence, for which the optimal destabilizing activity
A
B
C
Fig 6 The ARE I pentameric sequence can optimize the mRNA
destabilizing activity of the ARE II nonameric sequence (A)
Rabbit-b-globin-PAI-2 3¢UTR constructs prepared for the transient
transfec-tion of HT1080-TET OFF cells Double ARE mutants were
con-structed by combining ARE I-MUT and ARE II-MUT to create
plasmid pTETGLO ARE I+II-MUT and by combining ARE II and ARE III
to create plasmid pTETGLOARE II+III-MUT(Fig 2) (B, C) HT1080-TET
OFF cells were transfected with the TET-responsive b-globin
repor-ter plasmids described in (A) and the b-globin mRNA decay curves
were quantified by RPA as described in Fig 1 and the Experimental
procedures The graph in (C) corresponds to the experiments
shown in (B) (n = 3–5) Each point represents the mean ± SE.
Trang 7depends on the cooperative activity of two auxiliary
elements One is a pentameric motif (ARE I) located
55 nucleotides upstream of the core ARE II element,
and the second is an atypical AU-rich sequence (ARE
IV) abutting 3¢ to the core ARE II sequence (Fig 7)
This functional multidomain ARE structure has been
observed in a variety of class I and II ARE elements,
including those of c-fos, GM-CSF, IL-8 [28,29,33]
Discussion
PAI-2 is a serine protease inhibitor and is a
highly-reg-ulated member of the plasminogen activator system,
and is one of the most highly inducible genes known
Its expression can be dramatically increased in
response to cytokines, growth factors, hormones,
lipopolysaccharides and tumour promoters
[16,18,20,21,51] Although the impressive induction of
PAI-2 has been attributed to transcriptional events,
work from the early to mid-1990s demonstrated that
PAI-2 gene expression could be regulated
post-trans-criptionally via the modulation of mRNA stability
[19,24]
We previously demonstrated that human PAI-2
mRNA was inherently unstable, with a half-life of
1 h and that most of the destabilizing activity was
attributed to the 3¢ UTR [24] and, to a lesser extent,
an instability element within exon 4 of the coding
region [52] It was originally predicted the nonameric
ARE (UUAUUUAUU) located 304 nucleotides
upstream of the poly(A) tail was largely responsible
for the 3¢ UTR driven-instability of the PAI-2
tran-script However, mutagenesis of this nonameric ARE
only partially stabilized both a HGH-PAI2-3¢ UTR
chimeric transcript [48] and a b-globin-PAI-2 3¢ UTR
chimeric transcript (present study) Work from other
groups has demonstrated that the presence of a single
nonameric element [UUAUUUA(U⁄ A)(U ⁄ A)] within a
3¢ UTR has a modest effect on the stability of a
repor-ter transcript [34,53]; as such, we predicted that the
PAI-2 3¢ UTR contained additional functional instabil-ity elements, AU-rich or otherwise [37,54,55], that could contribute to the overall decay rate of the tran-script
Our analysis of the PAI-2 mRNA 3¢ UTR sequence revealed that the nonameric element was present in the centre of a 108 nucleotide class I type of ARE element consisting of three copies of the AUUUA motif that were evenly distributed within a U-rich (54%) region and an atypical ARE (AREIV) immedi-ately adjacent to the nonameric element Moreover, this region did not contain three to six clustered AUUUA motifs, which is indicative of class II ARE elements [35,56] On the basis of this sequence analy-sis, we hypothesized that this 108 nucleotide AU-rich sequence contained all the essential destabilizing ele-ments in the PAI-2 3¢ UTR, and we confirmed this
by demonstrating that either deleting the 108 nucleo-tide ARE (Fig 3) or replacing it with an irrelevant sequence of equivalent length (data not shown) stabi-lized the transcript to a level equivalent to that seen for the wild-type b-globin transcript Furthermore,
we also demonstrated that the 108 nucleotide ARE was sufficient to destabilize the b-globin transcript with kinetics similar to those seen with the PAI-2 full-length 3¢ UTR
The pentameric motif (AUUUA) is the minimal active destabilizing sequence element when present within an appropriate AU-rich or U-rich environment
We therefore tested the hypothesis that each of these pentameric AREs (ARE I, II and III) contributed equally to the overall transcript instability and, as such, were functionally equivalent in an analogous manner to the three pentameric motifs located in the c-fos transcript [33] However, this set of experiments (Fig 5) demonstrated that ARE I and III did not con-tribute to transcript instability, either individually or in combination (Fig 5) We then sought an alternative model to explain the destabilizing characteristics of the PAI-2 ‘extended ARE’ element
Fig 7 CLUSTALW analyses of the PAI-2 ‘extended ARE’ from different mammalian species reveals a high degree conservation in the ARE II and ARE IV regions; human (accession no J02685), Pan troglodytes (accession no XM_001148307), mouse (accession no X16490), and rat (accession no X64563) The open boxes indicate the relative positions of the human ARE I, II, III and IV elements.
Trang 8We next investigated the possibility that the
struc-ture of the PAI-2 ‘extended ARE’ was based on a
multidomain model consisting of an essential,
func-tional destabilizing core domain (e.g the ARE II
nonameric sequence), for which the destabilizing
activ-ity was optimized by the presence of nearby auxiliary
AU-rich sequences Figures 6 and 8 demonstrate that
the b-globin-PAI-2 3¢ UTR chimeric transcript was
only stabilized in an manner comparable to the
b-glo-bin and the b-glob-glo-binDARE transcripts, upon the
intro-duction of two different sets of double mutations (e.g ARE II + ARE IV mutant and ARE II + ARE I mutant) Taking into consideration the fact that muta-genesis of either ARE I or ARE IV in isolation (Figs 5 and 8, respectively) did not influence the transcript’s decay rate, we propose that the ARE II nonameric sequence forms the core destabilizing domain of the PAI-2 ARE and that its optimal destabilizing activity requires the contribution of either the 3¢ abutting ARE
IV (AUUUUAUAUAAU) sequence or the 5¢ ARE I pentameric motif Apart from optimizing the destabi-lizing activity of the core ARE II sequence, the ARE
IV and ARE I elements also appear to buffer effects
of mutations in the core ARE II nonameric sequence, thereby retaining the AREs destabilizing activity, albeit less efficiently (Fig 1) Subsequently, we suggest that the PAI-2 ARE IV and ARE I elements can act as auxiliary elements to the PAI-2 core ARE II sequence
To investigate the means by which these ARE elements cooperate in modulating PAI-2 mRNA stability, REMSA analyses were performed to determine the role of the ARE II and ARE IV sites in the binding of proteins to the ‘extended ARE’ Binding of cytoplas-mic proteins to the ‘extended ARE’ probe was first shown to be specific as determined by competition titration experiments ARE II was shown to play a sig-nificant role in this binding activity because a four nucleotide substitution introduced into the ARE II caused substantial decrease in binding activity By con-trast, mutagenesis of ARE IV had no noticeable effect
on the binding of proteins to the ‘extended ARE’ and had no additional suppression of protein binding activ-ity in the presence of the mutated ARE II Hence, ARE IV does not appear to modulate protein binding activity to the ‘extended ARE’ The means by which ARE IV cooperates with ARE II to destabilize mRNA still remains unknown The role of the ARE 1 site was not investigated in the present study and will be the subject of future research
Functional multidomain ARE structures have been observed in a variety of class I and II ARE elements, including those of c-fos, GM-CSF and IL-8, amongst others (Fig 9) [28,29,33], and appear to function via similar mechanisms Of greatest relevance to the PAI-2 ARE is the c-fos multidomain class I ARE, for which the structure and function has been characterized in detail; this ARE is composed of two structurally dis-tinct but functionally interdependent domains [33] (Fig 9) The c-fos ARE core sequence consists of three pentameric motifs embedded within a U-rich region and is independently capable of destabilizing a tran-script The c-fos ARE auxiliary domain II is a
20 nucleotide U-rich sequence that cannot
indepen-A
B
C
Fig 8 An atypical AU-rich sequence (ARE IV) abutting 3¢ to the
ARE II pentameric sequence can optimize the mRNA destabilizing
activity of the ARE II nonameric sequence (A) Rabbit-b-globin-PAI-2
3¢ UTR constructs prepared for the transient transfection of
HT1080-TET OFF cells The AU-rich sequence (ARE IV) abutting 3¢
to the ARE II pentameric motif was mutated to create plasmid
pTETGLO ARE IV-MUT ; a double ARE mutant that combined ARE
II-MUT and ARE IV was constructed to create plasmid
pTETGLO ARE II+IV-MUT (Fig 2) (B, C) HT1080-TET OFF cells were
transfected with the TET-responsive b-globin reporter plasmids
described in (A) and the b-globin mRNA decay curves were
quanti-fied by RPA as described in Fig 1 The experiments shown in
(C) were repeated three times and each point represents the
mean ± SE.
Trang 9dently destabilize the transcript; however, when
pres-ent in the appropriate context (i.e immediately
up- or downstream of the core domain I) [33], it
can stimulate the deadenylation rate and thereby
increase the decay rate of the transcript Moreover,
domain II of c-fos also serves the essential function
of buffering the effects of mutations occurring within
domain I [33]
The relative location of a functional auxiliary
domain, with respect to the core domain is flexible
because auxiliary domains have been identified either
5¢ or 3¢ to the core domains in class I and II AREs
[28,29,33] (Fig 9); moreover, placing the c-fos
auxil-iary domain either 5¢ or 3¢ to the core domain resulted
in a similar deadenylation and overall mRNA decay
rate [33] The PAI-2 ARE is unusual in that in
addition to an auxiliary domain (Fig 7, the atypical
AU-rich ARE IV; Fig 8) immediately 3¢ to the core,
the ARE I (AUUUA) element located 5¢ to the core
element (Figs 7 and 9) also behaves as a functional
auxiliary domain in the presence of a mutated ARE
IV (Fig 8) Whether the two PAI-2 auxiliary domains
are simultaneously active cannot be determined from
the data obtained in the present study, although it
does remain a plausible hypothesis However, we
suggest that, under normal physiological conditions,
the destabilizing activity of the core domain is
prefer-entially optimized by auxiliary domain I (Fig 7, the
atypical AU-rich ARE IV; Fig 9) based on the high
degree of homology in the equivalent sequences of
other species (Fig 7) Moreover, the addition of the
second auxiliary sequence, domain I (Fig 9), can
sup-port the destabilizing activity of the core domain in
the absence of domain IV (Fig 8)
In summary, under normal physiological conditions,
the PAI-2 mRNA transcript is unstable, which we now
attribute to the presence of a multidomain AU-rich
element within the 3¢ UTR (Fig 9) ARE-mediated
PAI-2 mRNA instability significantly contributes to
the low constitutive levels of PAI-2 protein; however, the ARE can also modulate PAI-2 mRNA stability during physiological conditions that require high levels
of PAI-2 gene expression and, subsequently, the contri-bution of post-transcriptional regulation to PAI-2 gene expression cannot be underestimated The present study has focused on the characterization and fine mapping of the functional destabilizing AU-rich region within the PAI-2 3¢ UTR under physiological condi-tions that result in an unstable PAI-2 mRNA tran-script We have shown that the PAI-2 ARE is a
74 nucleotide multidomain (Fig 9) class I element con-sisting of a destabilizing core nonameric element with activity that is supported by one of two auxiliary ele-ments Hence, in an attempt to severely inhibit constit-utive PAI-2 gene expression, nature has evolved a functional, mutation insensitive, multidomain ARE element We are currently determining the contribution and mechanism of this ARE, and the individual ARE domains, to PAI-2 mRNA stabilization and, subse-quently, PAI-2 gene expression
Experimental procedures
Plasmids and mutant construction The vector pTETBBB was provided by A B Shyu (Univer-sity of Texas Medical School, Houston, TX, USA) This plasmid contains the gene for b-globin under the control of
a tetracycline-regulated promoter that allows the transcrip-tion of this gene in the absence of tetracycline within an appropriate mammalian cell line (e.g HT1080-TET OFF) pTETBBB is referred to as pTETGLO throughout the present study
The PAI-2 3¢ UTR was amplified from plasmids pCMV-glo-PAI-2 3¢ UTR and pCMV-pCMV-glo-PAI-2 3¢ UTR-ARE MUT [48] with primers SJS133 and SJS134, and cloned into the BglII site in the b-globin 3¢ UTR in pTETGLO, to gener-ate pTETGLOPAI)2and pTETGLOARE II-MUT
,respectively
Fig 9 Multidomain structure of PAI-2 (accession no J02685), c-fos (accession no NM_005252), GM-CSF (accession no M11220) and IL-8 (accession no Y00787) AREs Sequences of the AREs are shown with the AUUUAs underlined, and the relative positions of the core and auxiliary domains are overlined.
Trang 10Mutant variants of the PAI-2 3¢ UTR were generated via
overlap extension PCR mutagenesis [57] using SJS133 and
SJS134 as the external primers and the constructs
pTET-GLOPAI)2 and pTETGLOARE II-MUT as the templates
Mutation of the ARE I-AUUUA and ARE III-AUUUA
sequences used primers SJS172 and SJS173, and SJS174 and
SJS175, respectively Mutation of the atypical AU-rich
sequence used primers SJS259 and SJS260, and the creation
of the ARE II⁄ ARE IV double mutant used primers SJS261
and SJS262 The mutagenesis of ARE I and III introduced
HindIII restriction sites and so the creation of the
pTET-GLO3¢ UTRDARE involved digesting construct
pTET-GLOARE I+III-MUT mutant with HindIII to remove the
108 bp ARE, gel purifying the larger fragment and
self-liga-tion The PAI-2 ‘extended ARE’ was amplified from plasmid
pTETGLOPAI)2 using primers SJS137 and SJS138, and
cloned into the BglII site in the b-globin 3¢ UTR in
pTET-GLO, to generate pTETGLOEXT.ARE The sequences of the
primers used in the present study are listed in Table 1
Cell culture and transfection
HT1080-TET OFF cells (Clontech) were maintained in
DMEM supplemented with 10% fetal bovine serum and
100 lgÆmL)1G418 (Life Technologies, Inc Carlsbad, CA,
USA) Cells were maintained at 37C in the presence of
5% CO2 Transient transfections were performed via the
Fugene (Roche, Basel, Switzerland) method according to
the manufacturer’s instructions A typical mRNA decay
experiment involved seeding five 35 mm plates with 5.0· 105cells and incubating overnight The next day, each plate was transfected with a total of 1 lg of plasmid DNA and incubated at 37C for 5 h These cells were then washed once with NaCl⁄ Pi, trypsinized, combined and equally seeded into five 35 mm to ensure equal transfection efficiency within samples and the plates were returned to the incubator for further incubation
In vitro transcription and RNase protection assay and northern hybridization
A cDNA library prepared with 1 lg of total RNA extracted from an HT1080 TET-off cell line transiently transfected with pTETBBB was used to generate the rabbit b-globin and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) riboprobes Accordingly, a 295 bp b-globin frag-ment that spans the first intron was amplified using primers SJS167 and SJS170 and a 155 bp GAPDH fragment was amplified using primers ALS030 and SJS209; these frag-ments were then cloned into the pGEM-T Easy vector (Promega, Madison, WI, USA) generating pTEasy-Globin and pTEasy-GAPDH For in vitro transcription, 500 ng of SpeI linearized pTEasy-Globin and SacII linearized pTEasy-GAPDH were incubated for 1 h in the presence of
50 lCi [a-32P]UTP (PerkinElmer Life and Analytical Sci-ences, Inc., Waltham, MA, USA), 10 lm UTP, 0.5 mm ATP, 0.5 mm CTP, 0.5 mm GTP, 40 U of RNase Inhibitor (Promega Corporation, Madison, WI, USA) and either
Table 1 PCR and overlap PCR mutagenesis primers The name, nucleotide sequence, orientation and GenBank nucleotide reference (where available) are provided The introduced mutations are underlined,the restriction enzyme sites are italicized, and lower case indicates the T7 promoter sequence PAI-2 cDNA (accession no M18082), GADPH cDNA (accession no M33197), pTETBBB plasmid sequence from Profes-sor A B Shyu (University of Texas Medical School, Houston, TX, USA) nt, nucleotide.