(Received for publication, November 15, 1996)
From the Department of Cell Research and Immunology,
George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv
69978, Israel and the § Laboratory of Mathematical Biology,
DCBDC, NCI, NIH, Frederick, Maryland 21702
It has become clear that a given cell type can
qualitatively and quantitatively affect the expression of the
platelet-derived growth factor B (PDGF2/c-sis) gene at
multiple levels. In a previous report, we showed that
PDGF2/c-sis 5-untranslated region has a translational
modulating activity during megakaryocytic differentiation of K562
cells. This study points to the mechanism used for this translational
modulation. The unusual mRNA leader, which imposes a major barrier
to conventional ribosomal scanning, was found to contain an internal
ribosomal entry site that becomes more potent in differentiating cells
and was termed differentiation-linked internal ribosomal entry site
(D-IRES). The D-IRES element defines a functional role for the
cumbersome 1022-nucleotide-long mRNA leader and accounts for its
uncommon, evolutionary conserved architecture. The
differentiation-linked enhancement of internal translation, which
provides an additional step to the fine tuning of
PDGF2/c-sis gene expression, might be employed by numerous
critical regulatory genes with unusual mRNA leaders and might have
widespread implications for cellular growth and development.
The mechanisms that control the expression of genes operate at a
variety of levels that include all the transcriptional and post-transcriptional stages. In a rapidly growing number of examples, gene expression is regulated also at the level of translation, usually
involving the translational initiation step. mRNA leaders that
mediate efficient translation initiation are short (50-70 nt),1 contain 5-m7G cap
structure, and lack stable secondary structures as well as upstream AUG
codons (reviewed in Ref. 1). Interestingly, genes encoding for
regulatory proteins involved in growth and development often do not
meet these criteria and produce long, structured, and upstream
AUG-burdened mRNA leaders. These features are incompatible with
efficient translation by the ribosomal scanning mechanism which is
considered valid for the translation of the vast majority of cellular
and viral mRNAs (2).
Platelet-derived growth factor (PDGF) is a potent mitogen of all cells
of mesenchymal origin and has a major role in wound healing as well as
in embryogenesis and development. PDGF consists of homo- or
heterodimers of two protein chains, PDGF-A and PDGF-B. The PDGF-B chain
is the product of the PDGF2/c-sis proto-oncogene, which has
a neoplastic transformation potential and is expressed in various types
of tumor tissues. Normally, the expression of PDGF2/c-sis is
tightly regulated due to multiple control mechanisms at the
transcriptional and post-transcriptional levels. PDGF2/c-sis mRNA has a striking architecture: it has a 723-nt coding region and
exceptionally long 5- and 3
-untranslated regions (UTRs) of 1022 and
1625 nt, respectively. The extraordinarily long 5
-UTR contains stable
secondary structures and three mini-open reading frames upstream of the
translation initiator AUG codon. As such, it imposes a significant
barrier to the linear ribosomal scanning from the 5
-m7G
cap structure toward the fourth AUG codon, thereby serving as a
potent translational inhibitor (reviewed in Ref. 3).
One of the major sites of PDGF synthesis is within bone marrow
megakaryocytes, the platelets progenitor cells. As shown in our earlier
work, the 5-UTR-mediated translational inhibition was relieved at a
certain time window during megakaryocytic differentiation (4). The
translation inhibitory effect of PDGF2/c-sis 5
-UTR was not
relieved by elevated levels of the cap-binding protein eIF4E (4), nor
by mutations of the upstream AUG codons (5, 6). These observations
prompted us to examine whether during the differentiation process
PDGF2/c-sis mRNA is translated by an alternative
cap-independent initiation mechanism.
A cap-independent mechanism for translation initiation has been
demonstrated for picornaviruses, which lack 5-m7G cap and
have long, structured, AUG-burdened mRNA leaders. An internal
ribosomal entry site (IRES) was found in members of the cardio-,
aphto-, entero-, rhino-, and hepatoviruses, as well as in pestiviruses.
Although there is no similarity in the primary sequences of the
different IRES elements, there are similarities in their secondary
structural features (reviewed in Refs. 7 and 8). Interestingly,
internal translation has been found also for the cellular genes of
mammalian Bip (9), FGF2 (10), IGF-II (11), eIF4G (12), and
Antennapedia of Drosophila melanogaster (13). In
this study we demonstrate that the PDGF2/c-sis 5
-UTR contains an IRES element that responds to changes in the cellular milieu and facilitates translation of the mRNA during
megakaryocytic differentiation. Computer analysis involving inspection
of compensatory base changes among divergent PDGF2/c-sis
sequences from human, cat, and mouse predicts structural features
within the 5
-UTR that are common to known IRES elements.
This study provides the first example for a differentiation-linked IRES (D-IRES) element, which might be one of many inducible IRES elements in extraordinary mRNA leaders of cellular genes encoding critical regulatory proteins.
Plasmid Constructions
pT7CAT-LUC obtained from Dr. Sarnow (9) was digested with
SalI, filled in, and ligated with 1-kb
NcoI-MluI filled in fragment of pPD1 (4), to
create pT7CPL. pT7CEL was constructed by three-part ligation of the
following fragments: (i) 0.3-kb
SalI(filled-in)-NcoI fragment from pT7CAT-LUC,
(ii) 0.5-kb ApaI(filled-in)-NcoI fragment of
pEMCV-CAT (14), and (iii) the backbone NcoI fragment of
pT7CAT-LUC. To construct pT7CHL, 0.7-kb SalI fragment was
generated from pHAV/7 template (15) by polymerase chain reaction using
oligonucleotide primers 5-GGGGGTCGACCTCACCGCCGT-3
and
5
-GGGGGTCGACCTAAATGCCCC-3
, followed by SalI digestion and
ligation into the SalI site of pT7CAT-LUC. The bi-cistronic
plasmids under the CMV promoter were generated using pBI-FC1 or pHP-FC1
from Dr. A.-C. Prats as backbones (10). The following cloning steps
were performed: 1.8-kb NotI-SacI fragment
harboring the LUC gene was generated by polymerase chain reaction using
pT7CAT-LUC (9) as template and oligonucleotide primers
5
-GGGGTCGACGCGGCCGCCCATGGAAGACGCCAAAAACATA-3
and
5
-CCCCGAGCTCCATTCATCAATTTGC-3
and was inserted into NotI
and SacI sites of pBluescript II SK+ (Stratagene) downstream
of the T7 promoter to create pBS-LUC. In addition, the 1.8-kb
SacI-NotI fragment harboring the LUC gene was
ligated with the 4.1-kb SacI-NotI fragment of
pBI-FC1 to create the bi-cistronic pCL plasmid. Intact or truncated
c-sis 5
-UTRs were inserted into pBS-LUC plasmid, upstream
to LUC. This was done by ligating pBS-LUC
SpeI-NcoI fragment with polymerase chain reaction
fragments which were generated by using psis4.0 (5) as template and
oligonucleotide primers homologous to specific c-sis 5
-UTR
sequences. The following oligonucleotides, which bear synthetic
SpeI or NcoI sites, were used:
5
-CCCCACTAGTGGCAACTTCTCCTCC-3
(JB7) and
5
-CCCCCCATGGCGACTCCGGGCCCGGCCC-3
(OS35) to amplify the intact 5
-UTR
(1-1022); 5
-CCCCACTAGTAACCGGAGCAGCCGCAGC-3
(OS64) and
5
-CCAACCATGGCTTTGCAACGGCAGC-3
(OS65) to amplify region 215-488; and
5
-GGGTACTAGTGCTGCCGTTGC-3
(OS66) and 5
-CTCACCCCCATGGCCCCGGC-3
(OS67) to amplify region 475-870. The ATG of the NcoI site
was designed to match exactly with the ATG of LUC. The
SpeI-SacI fragment containing the full-length
5
-UTR fused to LUC was ligated with the 4.1-kb
SpeI-SacI fragment of pBI-FC1 or pHP-FC1 to
create pCPL or pHCPL, respectively. The SpeI-SacI
fragments containing the truncated 5
-UTR fused to LUC were ligated
with the 4.1-kb SpeI-SacI fragment of pSCT Bi28
to create pC(215-488)L and pC(475-870)L.
Megakaryocytic Differentiation and Plasmid Expression
Infection-TransfectionMegakaryocytic differentiation was induced by treatment of exponentially growing K562 cells with TPA (Sigma) at concentration of 5 nM for 48 h. Infection-transfection was carried out in a six-well dish using 106 cells and 5 µg of supercoiled plasmid DNA/well. 30 min prior to transfection, control or TPA-treated cells were infected with recombinant vaccinia virus vTF7-3 at multiplicity of infection of 1 plaque-forming unit/cell, followed by liposome-mediated transfection, as described previously (4).
Electroporation100 µg of supercoiled plasmid DNA per a total of 107 exponentially growing K562 cells resuspended in 0.8 ml of RPMI 1640 without serum were used for each electroporation pool, using an electric pulse of 240 V, 1500 microfarads (Easy Ject+ electroporator, Equibio). Immediately following the electric pulse, the cells were transferred to RPMI 1640 medium supplemented with 50 units/ml of penicillin, 50 µg/ml of streptomycin, and 15% fetal calf serum. 24 h after electroporation the cells were diluted to a final concentration of 5 × 105 living cells (as determined by using trypan blue for counting) in RPMI 1640 supplemented with 10% fetal calf serum, with or without 5 nM TPA (Sigma), for 48 h.
CAT and Luciferase AssaysThe TPA-treated and control transfected cells were harvested simultaneously for CAT and luciferase assays, as well as for RNA analysis. For the enzymatic activity assays, the cell pellet was lysed by three freeze-thaw cycles in 0.1 M Tris, pH = 8.0. CAT activity in the cells lysates was determined by a phase extraction assay, which quantified butyrylated 3H-labeled chloramphenicol products by liquid scintillation counting, followed by xylene extraction (pCAT reporter gene system, Promega). LUC activity in the cell lysate was determined using TD-20e-Luminometer (Turner) following 15-s incubation of 1-5 µg of the protein extract with 470 µM luciferin (Sigma) and 270 µM coenzyme A (Sigma) in 20 mM K-Hepes, pH = 7.8, 1 mM EDTA, 4 mM MgAc, 1 mg/ml bovine serum albumin, and 530 µM ATP. Total protein concentration in each sample was determined by the Bradford assay (16).
RNA AnalysisTotal RNA was isolated from 2 × 107 cells using Tri Reagent procedure (Tri Reagent®, MRC, Inc.), which is based on the phenol and guanidine thiocyanate in a monophase solution. Poly(A)+ RNA was purified from 150 µg of total RNA using oligo(dT) magnetic beads (Dynatech, Inc.). Poly(A)+ RNA was subjected to Northern blot analysis according to standard procedures (17), using 32P-labeled cDNA probes specific for the CAT, PDGF2/c-sis, LUC, or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) genes. To quantify the RNA, the intensities of specific bands were determined using the phosphorimager (Fujix Bas 100, Fuji).
The leader of PDGF2/c-sis mRNA is highly
complex and imposes a major difficulty to linear ribosomal scanning.
However, as demonstrated previously, megakaryocytic differentiation of
K562 cells alleviates its translational inhibitory effect (4). We therefore wished to examine whether PDGF2/c-sis mRNA
leader harbors a weak IRES element which is activated upon
differentiation. Since internal translation activity is known to be
dependent on trans-acting factors (reviewed in Refs. 7 and
18), the performance of IRES elements is assumed to be affected by the
cellular availability of such factors. Hence, the first objective was
to determine whether the cellular environment created by the
TPA-induced differentiation is more supportive for internal
translation. Internal ribosomal entry is traditionally demonstrated
using bi-cistronic expression plasmids. In these plasmids, the second
cistron is not translated unless it is located immediately downstream
to an active IRES which is downstream to the stop codon of the first
cistron. Chloramphenicol acetyltransferase (CAT) and luciferase (LUC)
reporter genes were used as the first or second cistrons, respectively,
under control of the T7 promoter. The well characterized IRES elements
of encephalomyocarditis virus (EMCV) or hepatitis A virus (HAV) were
inserted between the two cistrons, downstream to the CAT translational
stop codon, to create plasmids pT7CEL or pT7CHL, respectively. In
addition, the 5-UTR of PDGF2/c-sis was inserted between the
two cistrons to create pT7CPL. Plasmid pT7CL, in which LUC is located
immediately downstream to the stop codon of CAT (Fig.
1A), served as the control plasmid. The
hybrid T7/vaccinia expression system was employed to express
bi-cistronic mRNA from T7 promoter in the cytoplasm of control or
TPA-treated K562 cells.
Control or TPA-treated K562 cells were infected with vTF7-3, a
recombinant vaccinia virus that expresses the T7 RNA polymerase in the
cytoplasm, and transfected with one of the above bi-cistronic plasmids.
24 h after the infection-transfection experiment, CAT and LUC
enzymatic activities were measured. Fig. 1B illustrates the
LUC/CAT ratio obtained in the differentiated cells relative to the
control cells for each of the plasmids. As shown, the differentiation process generated a cellular milieu that facilitated the internal translation conferred by the viral IRES elements. The enhancement of
EMCV IRES activity was about 3-fold whereas that of HAV was less
prominent (Fig. 1B). Interestingly, differentiation-linked activation of the second cistron translation was also observed upon
insertion of PDGF2/c-sis 5-UTR (Fig. 1B,
pT7CPL). This phenomenon led us to speculate that the mRNA
leader of PDGF2/c-sis might contain an IRES element that
gains activity upon certain cellular changes.
From
studies on picornaviruses, IRES elements are known to have structural
features that are important for their role as mediators of internal
ribosomal entry (7, 18). The RNA structural analysis of
PDGF2/c-sis 5-UTR was performed by theoretical prediction of RNA folding and phylogenetic analysis. In the structural analysis, the thermodynamically favored helical stems folded in the 5
-UTR were
predicted by EFOLD, a method that computes all possible alternative RNA
structures based on the fluctuation of thermodynamic energy parameters
within the range of experimental errors for these parameters (19).
These computed helical stems were evaluated by a statistical simulation
SEGFOLD (20, 21) and verified by inspecting their conservation of
complementarity and compensatory base changes among divergent PDGF
sequences from human, cat, and mouse. Based on both the conserved and
thermodynamic favored helical stems within these divergent 5
-UTR
sequences, the possible RNA secondary structure was constructed. As
shown in Fig. 2A, the predicted common RNA
secondary structure is divided into six domains, A-F. Domain B, composed of the sequence from 191-685, was predicted to be
an extensive RNA secondary structure of sufficient stability to inhibit
the ribosome scanning process. Distinct Y-type stem-loop structures
were observed just upstream of the first, third and fourth AUG codons.
These conserved Y-type structures, denoted B5, D, and
F (Fig. 2A), have significance scores of
1.75,
1.87, and
2.57, respectively. Such significance scores suggest
structural role for these regions (19). Structure B5 (512-581) is just 1 nt upstream of the first AUG triplet. Structures D (784-860) is just
9 nt upstream of the third AUG triplet, and structure F (934-1005) is
about 17 nt upstream of the fourth AUG triplet. Structures D and F are
composed of GC-rich sequences (more than 84% GC). Polypyrimidine
sequences rich in U residues, that have been found to be important for
viral IRES activity (22), are located in the Y-type structure B5 and in
the stem-loop structure designated as B8 (Fig. 2B).
Two pseudoknot structures, denoted by the letter K in Fig.
2, were predicted. One (K1) is formed by interaction between
sequences in loop B5b of Y-type structure B5 with sequences in the loop of structure B7. The potential pseudoknot is present in the human, mouse, and cat structures, although the interacting sequences are not
phylogenetically conserved. The second pseudoknot (K2) is
formed by interaction of CCCG in the Y-type structure F loop (F1, Fig. 2B), with CGGG just downstream of it.
The evolutionary conservation of this interaction suggests that the
structure has a functional relevance.
A sequence complementary to the conserved 3-end of 18 S rRNA was found
at the 3
-end of the 5
-UTR, just upstream of the fourth AUG codon,
from which the PDGF2/c-sis open reading frame begins. The
region of complementarity was predicted to be single-stranded (G, Fig. 2, A and B). It is noteworthy
that picornavirus IRES elements share a conserved complementary
sequence to the 3
-end of human 18 S rRNA (23, 24).
All the above features are compatible with the notion that
cis- elements within the 5-UTR of PDGF2/c-sis
confer internal translation initiation. A Y-type structure, followed by
a pseudoknot interaction and a sequence complementary to 18 S rRNA,
immediately upstream to the start codon of the internal translation, is
a common structural feature observed in all IRES elements of
picornaviruses, hepatitis C virus, and pestiviruses RNAs (23-25). All
the Y-type structures predicted for human PDGF2/c-sis
mRNA leader were found to be conserved in the feline and murine
PDGF2 mRNA leaders. The evolutionary conservation of these distinct
structures suggests a role in the IRES-dependent
translation.
The relevance of the results shown in
Fig. 1B to a cellular setup in which transcription takes
place in the nucleus was verified by determining the behavior of the
bi-cistronic mRNA in the absence of vaccinia virus infection. To
that end, another set of bi-cistronic plasmids was constructed
employing the CMV promoter (Fig. 3A). In pCL
plasmid, the LUC reporter gene was located immediately downstream to
the translational stop codon of CAT, whereas in pCPL the entire
PDGF2/c-sis 5-UTR was located between the two cistrons. To
determine the ability of specific structural motifs to mediate the
TPA-linked IRES activity, two additional plasmids, pC(215-488)L and
pC(475-870)L, were constructed according to the structural model of
the 5
-UTR presented in Fig. 2A. In these plasmids, the
regions spanning nucleotides 215-488 or 475-870 of the 5
-UTR were
located between the two cistrons, respectively. Each of these plasmids
was introduced into K562 cells by electroporation, followed by
incubation of the cells in medium with or without TPA for 48 h.
Following TPA treatment, CAT and LUC enzymatic activities and mRNA
levels were detected. The integrity of the bi-cistronic mRNA was
verified by Northern blot analysis using 32P-labeled
cDNA probes specific for the CAT, LUC, or GAPDH genes. The
endogenous GAPDH mRNA level served as internal control to monitor
the plasmid-derived mRNA level present in each of the different
samples. As observed in Fig. 3B, the expected 3.8-kb bi-cistronic mRNA was detected by both CAT and LUC probes. The anticipated reductions in the bi-cistronic mRNAs length were
observed, as expected, from plasmids that contained the truncated
5
-UTR. An additional, unexpected, smaller mRNA picked up by the
CAT probe probably reflects cross-hybridization with cellular mRNA,
as seen previously by others (9). However, the LUC probe picked
exclusively the bi-cistronic mRNA, indicating its integrity. Longer
exposure of the blot after its hybridization with
32P-labeled LUC cDNA clearly demonstrated the absence
of smaller LUC mRNA molecules (Fig. 3B), implying that
LUC protein synthesis results from translation of the second cistron.
Unexpectedly, TPA treatment led to elevation in the steady-state level
of all the CMV promoter-driven mRNAs. However, in view of the
integrity of the bi-cistronic mRNA molecules, both CAT and LUC
proteins are believed to be translated from an intact bi-cistronic
mRNA. Thus, the LUC/CAT ratio deduced from LUC and CAT enzymatic
activities in the different samples reflects the relative translation
efficiencies of both cistrons, regardless of the absolute mRNA
level in the samples.
Effect of differentiation on
c-sis IRES activity. A, schematic drawing of the
transcription unit of pCL, pHCPL, pCPL, pC(215-488)L, and
pC(475-870)L plasmids. pHCPL contains a hairpin with stability of
G =
40 kcal mol
1 upstream to CAT
(see "Experimental Procedures"). Triangles, CMV promoter; dark boxes, the CAT coding sequence; open
boxes, intact or truncated 5
UTR of c-sis, as
indicated; light boxes, the LUC coding sequence.
B, Northern blot analysis. K562 cells were mock-transfected (lane 1) or transfected with pCMV-CAT (lanes 2 and 3), pCL (lanes 4 and 5), pCPL
(lanes 6 and 7), pHCPL (lanes 8 and
9), pC(215-488)L (lanes 10 and 11),
or pC(475-870)L (lanes 12 and 13), followed by
incubation in medium with or without TPA (see "Experimental Procedures"). 72 h after the electroporation,
poly(A)+ RNA from 150 µg of total RNA were used for
Northern hybridization analysis using sequentially
32P-labeled cDNA probes specific for CAT, LUC, or
GAPDH, as indicated. Following each hybridization, the blot was exposed for 15 h, except for the LUC blot, which was
also subjected to longer exposure for 4 days to rule out the presence
of shorter luciferase transcripts. C, K562 cells were
transfected with each of the indicated plasmids as detailed in
B. 72 h after the electroporation, CAT and LUC enzymatic activities were determined. The data represent the LUC/CAT ratio obtained in the TPA-treated cells relative to control cells. The
bars represent the average ± S.E. of three independent
transfection experiments.
Fig. 3C shows the LUC/CAT ratio obtained in the
differentiated cells relative to control cells for each of the
plasmids. The pCL-derived LUC activity, which represents second cistron
translation from an IRES-less mRNA, probably reflects some level of
read-through, leaky scanning, re-initiation, or translation of
undetectable degradation products of the bi-cistronic mRNA.
However, such pCL-derived second cistron translation was hardly
affected by TPA treatment (Fig. 3C, pCL). Insertion of
region spanning nucleotides 215-488, which harbor structure B3 of the
5-UTR (Fig. 2A) upstream of LUC, decreased LUC translation
level about 2-fold compared with pCL (not shown), but did not change
the relative LUC/CAT upon differentiation (Fig. 3C,
pC(215-488)L). Insertion of the full-length (1022 nt) or a 395-nt
segment spanning nucleotides 475-870, which harbor structures B4-D of
the 5
-UTR (Fig. 2A), also decreased the second cistron
translation by 5- or 2-fold, respectively, compared with pCL (not
shown). However, these two segments were able to confer
differentiation-linked LUC translation enhancement (Fig. 3C,
pCPL and pC(475-870)L). In summary, the LUC/CAT ratio was not affected by the TPA treatment unless the full-length or a
specific region of the 5
-UTR was inserted between the cistrons. These
specific segments conferred about 2-fold enhancement in translation
efficiency of the second cistron over the first cistron due to the TPA
treatment. These observations suggest the presence of a
differentiation-linked internal ribosomal binding site within the
5
-UTR of PDGF2/c-sis.
To verify
the ability of the 5-UTR to confer internal translation, we tested its
ability to confer 5
-end-independent translation to the second cistron.
For that purpose, a hairpin structure with stability of
G=
40 kcal mol
1 was inserted upstream of
the first cistron of pCPL to create plasmid pHCPL (Fig. 3A).
Both pCPL and pHCPL plasmids contain the entire PDGF2/c-sis
5
-UTR between the CAT and LUC cistrons. K562 cells were transfected
with each of the plasmids, followed by 48-h incubation in control or
TPA containing medium prior to detection of CAT and LUC enzymatic
activities. The CAT or LUC expression levels obtained from the 5
hairpin-containing plasmid pHCPL were compared with those from plasmid
pCPL which lacks the 5
hairpin. As shown in Fig.
4A, translation of the first cistron represented by CAT activity was inhibited about 10-fold, whereas the
second cistron translation represented by LUC activity was inhibited
only about 3-fold by the 5
hairpin.
The lower CAT and LUC levels obtained from pHCPL plasmid compared with
pCPL might result in part from lower transfection efficiencies of
pHCPL. However, variation in transfection efficiencies cannot account
for the difference between the relative expression of the two reporter
genes. The relative CAT expression was always 3-fold lower than the
relative LUC expression (Fig. 4A). Thus, the 5 hairpin
inhibitory effect on the first cistron was 3-4-fold stronger than on
the second cistron. This absolute dissimilarity of the 5
hairpin
effect on translation of the two cistrons points to the ability of
PDGF2/c-sis 5
-UTR to confer 5
-end-independent translation
to the second cistron. As shown in Fig. 4A,
PDGF2/c-sis conferred the 5
-end-independent translation in
both the control (
TPA) as well as in the differentiated (+TPA) K562
cells. This observation indicates that the 5
-UTR harbors an IRES
element that is active in both cellular conditions. However, the IRES potency varies with changes in the cellular milieu, as shown in Fig.
4B illustrating the LUC/CAT ratio in the differentiated
cells relative to the control cells for each of the plasmids. The
relative LUC/CAT ratio of the IRES-less plasmid was hardly changed by
TPA treatment, while the effectiveness of the IRES was enhanced about 2-fold by the TPA treatment regardless of the 5
-end, as demonstrated for both pCPL and pHCPL. This experiment further signifies the differentiation-linked IRES activity of PDGF2/c-sis
5
-UTR.
Structural elements that determine the efficiency of translation
initiation include the 5-m7G cap structure, the context
surrounding the initiator codon, 5
-UTR length, stability of secondary
structures in the 5
UTR, and presence or absence of open reading frames
upstream of the initiator codon (uORFs) (1). A hairpin structure with
predicted stability of
G=
50 kcal mol
1 in
the 5
-UTR was found to significantly inhibit protein synthesis (26).
In addition, uORFs, which in some cases have a regulatory role (27),
were found to decrease the frequency of initiation of the major ORF.
Thus, it is not surprising that the extraordinarily long
PDGF2/c-sis 5
-UTR, which contains stable secondary
structures and three uORFs acts as a strong translational inhibitor
(4-6, 28). Mutations of the uORFs did not relieve the inhibitory
effect (5-6, 28). This observation, together with the highly stable secondary structures (
G =
270 kcal
mol
1) within the PDGF2/c-sis 5
-UTR, raised a
question about the ability of ribosomes to linearly scan along this
mRNA leader from its 5
-end according to the conventional scanning
model (29).
As demonstrated in this study, when the PDGF2/c-sis 5-UTR
was placed upstream of the second cistron in a bi-cistronic vector, it
conferred a differentiation-linked enhancement of translation with
preference for the second cistron (Figs. 3C and
4B). However, regardless of the cellular conditions, the 5
hairpin inhibitory effect on the first cistron was 3-4-fold stronger
than on the second cistron, indicating the inherent ability of
PDGF2/c-sis 5
-UTR to confer internal translation to the
second cistron even before differentiation (Fig. 4A). These
observations indicate the presence of an IRES element within
PDGF2/c-sis 5
-UTR which has low or high activity depending
on the cellular milieu. As shown in Figs. 1B, 3C,
and 4B, the PDGF2/c-sis IRES potency was enhanced
about 2-fold by TPA treatment. These data do not match our previous
experiments with monocistronic vectors showing 8-fold 5
-UTR-mediated
translational enhancement upon TPA treatment (4). We strongly believe
that the stronger effect observed using monocistronic vectors reflects
the behavior of the authentic PDGF2/c-sis mRNA. Sequences flanking the 5
-UTR in the bi-cistronic vector may
dramatically affect its folding and consequently its IRES activity.
This discrepancy is analogous to the discrepancies reported previously
for other IRES elements studied in the context of mono- and
bi-cistronic vectors (11, 30-31). In addition, since luciferase
protein is very unstable compared with CAT (32), LUC activity reflects the level of newly synthesized LUC molecules, whereas CAT activity represents the total accumulated CAT enzyme. Thus, the relative LUC/CAT
ratio, which in this study was used to measure IRES potency, can detect
only the differentiation-linked enhancement phenomenon and not its
actual intensity.
The evolutionary conservation of the stable secondary structures
predicted for the PDGF2/c-sis mRNA leader, and the
conserved complementarity to 18 S rRNA immediately upstream of the
major ORF (Fig. 2), strongly suggest a role in mediating internal
translation. The D-IRES activity was conferred by the full-length (1022 nt) 5-UTR or by a 395 nt segment spanning nucleotides 475-870 of the
5
-UTR. In contrast, the 273-nt segment spanning nucleotides 215-488
of the 5
-UTR did not confer D-IRES activity and thereby served as a
negative control in our experiments (Fig. 3). Interestingly, this
non-IRES segment, harboring structure B3 (Fig. 2A), is
composed of highly stable stem-loop structures and is located upstream of the D-IRES-conferring segment. Hence, we suggest that the role of
structure B3 is to inhibit PDGF2/c-sis mRNA translation
in case it is present at the wrong tissue or time, by imposing a barrier to ribosomal scanning. As shown previously, the 5
-UTR-mediated translation inhibition is relieved during megakaryocytic
differentiation of K562 cells (4). The present study demonstrates that
translational inhibition relief is achieved by enhanced activity of an
IRES element that resides downstream of structure B3, the scanning barrier.
It can be argued that expression of the major ORF reflects translation
of undetectable shorter transcripts that are transcribed from a cryptic
promoter and therefore lack most of the inhibitory 5-UTR. Our data
rule out this possibility for PDGF2/c-sis, since we used
both cytoplasmic and nuclear expression systems. The cytoplasmic system
employed the bacteriophage T7 promoter in the cytoplasm of vaccinia
virus-infected cells, in which the host protein synthesis is shut off
by the virus. The fact that the D-IRES phenomenon was observed in that
cytoplasmic expression system (Fig. 1B) proves that a
cryptic promoter activity within the PDGF2/c-sis 5
-UTR is
not responsible for the differentiation-linked enhanced expression effect. We cannot rule out possible translation of degradation products
that contain only the second cistron, but such a phenomenon does not
negate our interpretation, since the small amount of potential
degradation products is probably similar in all the samples, including
the control plasmid pCL (Fig. 3B).
It has become clear that a given cell type can qualitatively and
quantitatively affect the expression of the PDGF A and B chains at the
levels of transcription, RNA processing, translation, and
post-translation modifications (3). The presence of an IRES element
within PDGF2/c-sis 5-UTR defines a functional role to the
cumbersome mRNA leader and justifies its conserved unusual architecture. The evolutionary conservation of the secondary structures hints that the IRES activity is mediated by evolutionary conserved RNA-binding proteins. We speculate that the availability of such trans-acting factors differs in response to specific signals
according to the cell's need. The differentiation-linked enhancement
of PDGF2/c-sis IRES activity might involve
post-translational modifications of existing proteins or synthesis of
new proteins with RNA binding activity. Apparently, the profile of
proteins that bind to PDGF2/c-sis 5
-UTR dramatically
changes due to differentiation of K562
cells.2 It has become evident that viral
IRES elements require cellular trans-acting RNA binding
proteins for their function (reviewed in Ref. 7). Thus, some of the
differentiation-induced trans-acting factors might also
mediate the IRES activity of certain viruses. The enhanced IRES
activity of EMCV in the TPA-treated K562 cells (Fig. 1) supports this
notion and underscores the importance of cell type to virulence
potential. In addition, the variable potency of different known IRES
elements might result from variable binding affinities to common
factors and/or different requirements for additional tissue-specific
factors. In contrast to viral IRES elements, cellular IRES elements are
expected to be much less potent. Indeed, we found that
PDGF2/c-sis IRES is about 10-fold less active in TPA-treated
K562 cells than EMCV IRES (not shown).
PDGF belongs to a subclass of genes that encode for regulatory proteins
with a major role in cell proliferation. Among these are several
proto-oncogenes as well as growth factors, cytokines, receptors, and
transcription factors (2). A typical property of these genes is their
tight expression regulation to guarantee correct protein level at the
appropriate time and location. An associated feature is their complex
architecture, that is, cis-regulatory elements that mediate
expression level in response to signals. The stringent regulation is a
result of various mechanisms that simultaneously control expression at
multiple levels. One of the cis-elements shared by the above
subclass of genes is their extraordinarily long, structured and
AUG-burdened mRNA leader, which provides a built-in blockade
against efficient ribosomal scanning. Although leaky scanning and
reinitiation are the mechanisms responsible for some translation
modulation phenomena (reviewed in Ref. 27), a conditional internal
ribosomal entry may also provide a widespread mechanism for translation
regulation. First, it gives certain mRNA molecules cap-independent
translation ability in response to viral infection or stress
conditions, as was shown for immunoglobulin heavy-chain binding protein
(BiP) IRES (9) and recently proposed for eIF4G IRES (12). Second, it
can enforce an alternative translational start site, resulting in
translation of different proteins from the same mRNA molecule,
as mediated by the IRES element of FGF2 (10). Third, it may provide a
developmentally regulated protein synthesis like that proposed for the
homeotic gene Antennapedia of D. melanogaster and
for human IFG-II leader 1 IRES elements (13, 11). The present study
provides evidence for a differentiation-linked IRES element residing
within a cumbersome mRNA leader. In the absence of appropriate
trans-acting factors, it may provide a double safety control
mechanism, since it prevents efficient translation of the mRNA in
cases of uncontrolled transcription. Upon differentiation, the IRES
becomes more active, leading to efficient protein synthesis. The
translational enhancement has an additive effect with other levels of
control to achieve significant expression enhancement in the
appropriate time window during differentiation. This mechanism, which
provides an additional step to the fine tuning of
PDGF2/c-sis gene expression, might be employed by numerous
critical regulatory genes with unusual 5-UTRs and might have
widespread implications for cellular growth and development.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) M19719[GenBank]
We thank P. Sarnow for pT7CAT-LUC plasmid and A. C. Prats for pBI-FC1 and pHP-FC1 plasmids. We also thank E. Guetta, Z. Stein, and G. Rotman for helpful comments.
After completion of the study, we have noticed that TPA from Calbiochem is more potent than that of Sigma with regards to the effects measured in this research.