ABSTRACT
The 5
'
untranslated leader (
[Omega]
sequence) of tobacco mosaic virus (TMV) genomic RNA was utilized as a
translational enhancer sequence in expression of the 17 kDa putative movement
protein (pr17) of potato leaf roll luteovirus (PLRV).
In vitro
translation of RNAs transcribed from appropriate chimeric constructs, as well
as their expression in transgenic potato plants, resulted in the expected wild-type pr17 protein, as well as in larger translational products recognized
by pr17-specific antisera. Mutational analyses revealed that the extra proteins
were translated by non-canonical initiation at AUU codons present in the wild-type
[Omega]
sequence. In the plant system translation initiated predominantly at the AUU
codon at positions 63-65 of the
[Omega]
sequence. Additional AUU codons in a different reading frame of the
[Omega]
sequence also showed the capacity for efficient translation initiation
in vitro
. These results extend the previously noted activity of the TMV 5
'
leader sequence in ribosome binding and translation enhancement in that the TMV
translation enhancer can mediate non-canonical translation initiation
in vitro
and
in vivo
.
Translational efficiencies of eukaryotic mRNAs are influenced by various
factors, such as primary (5'-cap) and secondary (hairpin) structures, the sequence context of
the start codon or upstream regulatory elements, such as enhancer sequences or
small upstream open reading frames (uORFs) (
1
-
7
). While for their specific interaction with ribosomes and for start codon
recognition prokaryotic mRNAs make use of the Shine-Dalgarno sequence (
8
), the lack of a corresponding sequence in eukaryotic mRNAs upstream of the
start codon has led to various models for pre-initiation complexes binding to the RNA 5'-end and then scanning along the mRNA for recognition of the
translational start codon(s) (
7
,
9
,
10
).
An additional facet of eukaryotic mRNA translation has come from the
identification of 5' untranslated sequences which largely enhance translation. Such regulatory translational enhancer sequences
have been primarily documented to exist in the 5' leader sequences of RNAs from plant and animal viruses, such as potato
virus X, rous sarcoma virus, brome mosaic virus and tobacco mosaic virus (TMV)
(
11
,
12
). In the case of the TMV translational enhancer ([Omega]) sequence (consisting of the 5'-terminal 68 nt) it has been proposed that the absence of extended
secondary structures in this region causes the increase in translational
efficiency (
13
). In fact, a detailed analysis of the TMV (strain U1) [Omega] sequence pointed to the importance of the primary structure by
identifying two elements, a direct repeat of 8 nt and a CAA-rich region, as being responsible for translation enhancement (
14
). In line with previous observations that the [Omega] sequence is capable of promoting binding of two ribosome molecules
(disome formation) when elongation is blocked in the presence of sparsomycin (
15
,
16
), it was proposed that the core regulatory elements of the [Omega] sequence allow specific binding of a protein factor(s) required for
efficient initiation (
14
). In the disome complex one of the two ribosomes occupies the AUG start codon
of the replicase gene and the second was postulated to bind further upstream in
the [Omega] sequence (TMV strain SPS) at an AUU codon in position 14 (AUU
14
). Translation initiation at this AUU codon was proposed to occur (
16
), but with appropriate chimeric constructs consisting of the TMV (strain U1) translational enhancer in-frame with the AUG start codon of a reporter gene putative initiation at
the corresponding AUU codon (AUU
15
) did not contribute to increased reporter gene activity (
14
).
Initiation at non-AUG codons was originally proposed from experiments using synthetic
oligonucleotides (
17
). Furthermore, usage of AUU as a translational start was postulated for human
mitochondrial mRNA (
18
), but the first evidence for involvement of AUU as a start codon was described
for the
Escherichia coli
gene encoding initiation factor IF3 (
19
). Since then further evidence for eukaryotic translation initiation at AUU and
other codons has accumulated for animal and plant systems (
10
,
20
-
22
). Here we show that the TMV translational enhancer sequence can promote
alternative translation initiation at AUU codons.
The potato leaf roll luteovirus (PLRV)
pr17
(ORF4) gene was amplified by PCR from clone pCPL1 (
23
). The primers were designed to give unique restriction sites for
Spe
I and
Xba
I at the 5'- and 3'-ends respectively. A plasmid previously constructed
for high level expression of the PLRV capsid protein CP (ORF3) controlled by
the [Omega] sequence and the 35S promoter of cauliflower mosaic virus (CaMV) (
24
) was cut with
Spe
I and
Xba
I to remove the ORF3 N-terminal sequence. Subsequently the amplified ORF4 fragment was cut with
Spe
I and
Xba
I and cloned into the linearized plasmid pRT17/NIV. A
Hin
dIII fragment isolated from pRT17/NIV was cloned into the binary vector pBIN19 (
25
). Plasmids containing the ORF4 expression cassette were designated p17/NIV.
P17/NIV was transformed into
Agrobacterium tumefaciens
strain LBA 4404 (
26
) and stable transformation of
Solanum tuberosum
var. Desirée was performed according to published procedures (
27
), with the resulting agrobacteria carrying plasmid p17/NIV. Western blot
analysis of regenerated plants was carried out as described in Tacke
et al.
(
28
).
A
Hin
dIII fragment comprising the [Omega] sequence and ORF4 was isolated from plasmid p17/NIV and cloned into
plasmid pSP64 (Promega) under the control of the SP6 promoter (pS17N). Mutations in the [Omega] sequence were carried out by PCR using synthetic oligonucleotides
(synthesized on a DNA/RNA synthesizer 392; Applied Biosystems, Darmstadt, Germany) and plasmid pS17N as the template. The
upstream primer was located 5' of the SP6 promoter (position 2105 of pSP64), comprising a unique
Ssp
I restriction site. This oligonucleotide was combined with a set of downstream
primers complementary to the [Omega] sequence and bearing different point mutations and a
Ksp
I restriction site. Amplified fragments covered the SP6 promoter and the mutated
[Omega] sequence. These PCR fragments were subsequently cut with
Ssp
I and
Ksp
I and cloned into pS17N. Plasmids containing mutated forms of the [Omega] sequence were sequenced on a DNA Sequencer 373A (Applied Biosystems).
All pSP64-based plasmids were linearized downstream of ORF4 with
Eco
RI prior to
in vitro
transcription with SP6 RNA polymerase in the presence of the cap analogue m
7
GpppG (
29
). RNAs were translated either in a wheat germ extract or rabbit reticulocyte
lysate (Amersham Buchler) in the presence of [
35
S]methionine under conditions recommended by the supplier.
In vitro
products were analysed on 12.5% SDS-polyacrylamide gels and detected by fluorography (
30
).
Potato leaf discs were transformed with construct p17/NIV (Fig.
1
A) and transgenic lines carrying two or more copies of the transgene were
recovered. Western blot analysis of extracts from all independent transformants
detected the wild-type pr17 and an additional immunoreactive protein (pr17/n) with an
apparent molecular weight of 24 kDa (Fig.
1
B, C1). This pr17/n protein was not detected in PLRV-infected plants (
28
) nor in transgenic plants expressing ORF4 without the [Omega] sequence (Fig.
1
A and B, C4). Recloning and sequencing of the transgenes from a potato line
containing two transgene copies revealed identical sequences for the transcribed and translated regions (data not shown). Together with the
fact that a single copy line established at later stages also showed the same
two immunoreactive proteins and that, moreover, transgenic lines expressing
ORF4 without translational enhancer did not show the larger immunoreactive
protein pr17/n, these data indicate that formation of pr17/n was possibly a
result of alternative translation initiation at a non-AUG codon in the [Omega] wild-type sequence, thereby giving rise to an N-terminally elongated protein.
Inspection of the transgene sequence revealed that two AUU codons (AUU
15
and AUU
63
) of the [Omega] sequence were in-frame with the ORF4 AUG start codon (Fig.
1
A). To assess the size of a protein that would initiate in the [Omega] sequence two constructs were synthesized by site-directed mutagenesis in which the pr17 start codon was converted to
GCG and the AUU codons AUU
15
and AUU
63
of the TMV [Omega] sequence were mutated to AUG
15
and AUG
63
respectively (constructs p17/NI and p17/NIII; Fig.
1
A). Both constructs were used for transformation of
S.tuberosum
and protein extracts from leaves of regenerated plants were subjected to
Western blot analysis (Fig.
1
B). Plants transformed with construct p17/NI expressed a protein larger than
pr17/n (Fig.
1
B, C2), whereas p17/NIII transgenic plants showed a protein corresponding in
size to pr17/n, as detected in p17/NIV transgenic plants (Fig.
1
B, C3). It appears that expression of p17/NI and p17/NIII in transgenic plants
(Fig.
1
B, C2 and C3) resulted in much higher protein levels as compared with p17
transgenic plants (Fig.
1
B, C4). As p17 and p17/NIII did not contain the [Omega] sequence, this observation was explained by the unfavourable context of
the pr17 initiator codon (GGAA
The 5' leader of construct p17/NIV consisted of the TMV [Omega] sequence and an additional 63 nt derived from the multiple
cloning site. Due to the cloning strategy part of this cloning site was
inversely repeated, allowing the formation of a stable stem-loop structure (Fig.
2
A). This stem-loop is located 3 nt downstream of codon AUU
63
and could have made a substantial contribution to the signal for translation
initiation. In order to investigate the possible effect of this stem-loop on translation efficiency a
Hin
dIII fragment released from construct p17/NIV and comprising the complete 5' leader and ORF4 (pr17) sequence was cloned under the control of the SP6
promoter into vector pSP64 (construct pS17N). Furthermore, the stem-loop was deleted to yield plasmid pS17D. RNAs from both constructs were
transcribed
in vitro
and translated in a rabbit reticulocyte lysate, as well as in a wheat germ
extract.
In vitro
expression of the chimeric [Omega]-ORF4 construct (constructs pS17N and pS17D) indicated that
translation initiation occurred at two non-AUG codons of the [Omega] sequence upstream of the ORF4 AUG start codon. In addition,
translation of pS17N15 and pS17N63 RNAs provided circumstantial evidence for
initiation at codons AUU
15
and AUU
63
respectively. Further analyses were directed at unequivocally identifying the
non-AUG initiator codon in the [Omega] sequence utilized
in planta
. The most likely non-AUG codon recognized by the plant ribosomal initiation complex was AUU
63
, as transgenic plants transformed with construct p17/NIII (AUG
63
) expressed a protein corresponding in size to pr17/n.
The AUU
63
codon of plasmid pS17D was mutated to AGG
63
in order to inhibit translation initiation at this codon (construct pS17D3;
Fig.
3
A). In fact,
in vitro
-translated RNA of plasmid pS17D3 did not result in a product corresponding
in size to pr17/n, demonstrating that
in vitro
translation initiated at AUU
63
of the [Omega] sequence (Fig.
3
B, C12). Based on the results of Gordon (
21
) and Peabody (
22
), AUU
63
was further mutated to ACG or CUG (constructs pS17D1 and pS17D2 respectively;
Fig.
3
A). These codons are known to permit translation initiation with high efficiency
in mammalian cells and plant protoplasts. Similar results were obtained with
the mutated AUU
63
codon, as
in vitro
translation of pS17D1 and pS17D2 RNAs in the wheat germ system allowed
expression of pr17/n by initiation at both ACG
63
and CUG
63
(Fig.
3
B).
The flanking sequences at the AUU
63
codon largely conformed to the consensus context for plant AUG initiation
codons (Fig.
4
A). To further analyse the influence of bases neighbouring AUU
63
several point mutations were introduced into this region (Fig.
4
A). Single point mutations did not alter translation efficiency at AUU
63
(the total amount of protein synthesized from construct pS17A3 RNA and loaded
in lane C16 is lower as compared with total protein in the other lanes). Even
the replacement of a purine by a pyrimidine at the mutation-sensitive position -3 did not inhibit expression of pr17/n (construct pS17A6, Fig.
4
, C19). Only when the entire context of the AUU
63
codon was disrupted, as in pS17A7, expression of pr17/n was reduced (Fig.
4
, C20). On the other hand, adaptation of the flanking sequences according to the
consensus sequence did not increase translation initiation at AUU
63
as compared with the wild-type sequence (Fig.
4
B, C6, C14 and C15). These results indicate that the AUU
63
flanking sequences have only a minor effect on pr17/n translation efficiency
in vitro
.
As the flanking sequences exhibited little activity in modulating the efficiency
of translation initiation at codon AUU
63
, sequences located further upstream of AUU
63
(positions 44-58 of the [Omega] sequence) were examined for their influence on translation
initiation. An element composed of three AUU codons separated from each other
by one codon (`triple AUU block') is located 4 nt upstream of AUU
63
in a different reading frame (Fig.
5
A). Simultaneous mutation of all three AUU codons to ACU slightly increased
expression of pr17/n (Fig.
5
B, C21), whereas a point mutation of the central AUU codon to ACU had no effect
on translation initiation at AUU
63
(Fig.
5
B, C22). Thus the triple AUU block in the wild-type [Omega] sequence obviously decreased translation initiation at AUU
63
to some extent.
Potato plants transformed with PLRV ORF4 under the translational control of the TMV [Omega] sequence expressed two immunoreactive proteins, wild-type pr17 and mutant protein pr17/n. We were able to show that initiation at the internally located
translational start codons proceeded by leaky scanning of pre-initiation complexes and that a non-canonical translation mechanism was responsible for pr17/n formation
by alternative translation initiation at a non-AUG codon of the TMV translational enhancer.
In planta
and during
in vitro
translation in a cell-free plant system (wheat germ) initiation occured efficiently at the ORF4
AUG start codon, as well as some 25 codons upstream at AUU
63
of the [Omega] sequence. When AUU
63
was replaced by AUG
63
(construct p17/NIII) a protein corresponding in size to pr17/n was expressed in
transgenic plants, but mutation of AUU
63
to AGG
63
prevented expression of this N-terminally elongated pr17 (pr17/n).
In vitro
translation of ORF4 under the control of the [Omega] sequence resulted in expression of three proteins instead of the two
detected in transgenic plants. This was observed both in the reticulocyte
lysate and wheat germ extract: translation initiated additionally at AUU
15
, as is obvious from a mutant RNA in which AUU
15
had been replaced by AUG
15
. Differences in the expression patterns for both cell-free systems could probably reflect conditions of the
in vitro
translation systems which allow translation initiation at a non-AUG codon upstream of AUU
63
not recognized
in planta
. In addition, it is noteworthy that the animal and plant
in vitro
systems show different affinities for the two codons AUU
15
and AUU
63
. The fact that AUU
15
is predominantly used by the reticulocyte lysate for translation intiation does
not reflect preferences of the animal system for a different consensus context
of this AUU start codon, as the flanking sequences for AUU
15
and AUU
63
are identical. Although artefacts of the conditions of the
in vitro
translation cannot be excluded, animal-specific protein factors may be involved in mRNA interaction and specific
recognition of the first initiator codon, a phenomenon recently discussed in
detail for eukaryotic gene expression (
7
).
Further analyses of the [Omega] sequence focused on elements contributing to translation initiation at
AUU
63
. Optimal initiation of protein biosynthesis depends on the sequence context for
the start codon (
1
-
4
), which is different in plant and animal consensus sequences. However, in both
systems positions -3 and +4, with reference to the +1 adenosine of the AUG start codon,
require purine residues for efficient translation initiation (
31
). According to Cavener and Ray (
32
) the flanking sequences of mono- and dicotyledonous plants differ substantially. As the experiments
described here were carried out in a wheat germ system we cannot exclude that
in
S.tuberosum
the point mutations in the flanking sequences would exert a more prominent
effect on translational efficiency. The data presented here on the AUU flanking
sequences confirm their importance for optimal protein initiation, but mutation
of the entire consensus sequence did not completely inhibit translation
initiation. While mutation of a purine to a pyrimidine residue at position -3 did not apparently alter initiation efficiency at AUU
63
, the triple AUU block preceding this codon reduces its efficiency in
initiation. As was shown by site-directed mutagenesis, the triple AUU block may itself interact with the
scanning complex, forming initiation complexes and thereby competing with AUU
63
. In fact, leaky scanning is obviously the mechanism by which recognition of
start codons occurs in the TMV [Omega] sequence. When AUU
63
was mutated to AUG
63
expression of pr17 at the AUG of ORF4 was barely detectable, indicating that
the canonical AUG start codon at position 63 was now almost exclusively used
for formation of initiation complexes.
The translation initiation at AUU codons described here is a novel feature of
the [Omega] sequence, in addition to its function as a translational enhancer.
Previously a detailed analysis of the [Omega] sequence had identified two motifs necessary for translation
enhancement, a (CAA)
n
region and a direct repeat of 8 nt (
12
). Both the AUU
15
and AUU
63
codons are part of this direct repeat ACA
Alternative translation initiation at AUU
63
of the TMV [Omega] sequence, as well as leaky scanning and initiation at the canonical ORF4
AUG start codon from the identical mRNA, resulted in expression of two
proteins. Such bifunctional mRNAs are known for a number of other viruses and
eucaryotic mRNAs (
6
,
20
) and may lead to N-terminally altered proteins with modified functions, as for example in the
expression of two N-terminally different serine-threonine protein kinases encoded by the mouse
pim-1
oncogene or the translation of three proteins (C', C and Y) from Sendai virus RNA by exploiting an ACG and two different
AUG codons (
35
,
36
). Our results indicate that translation initiation at AUU
63
of the [Omega] sequence takes place with high efficiency
in planta
, with both pr17/n and wild-type pr17 accumulating to similar levels in transgenic plants. The
question remains whether the TMV [Omega] sequence directs expression of two N-terminally different proteins from TMV RNA. In TMV RNA the [Omega] sequence is followed by the polymerase gene and initiation
at AUU
63
of TMV strain U1 would extend the viral polymerase by only two amino acid
residues. The [Omega] sequences of other TMV strains (U2, L and Dahlemense) are slightly
different and the AUU codons corresponding in position to AUU
63
of TMV U1 are not in-frame with the polymerase gene. Hence, eventual expression of an N-terminally modified polymerase protein would not be a conserved
feature of different TMV strains.
As an alternative to the production of an N-terminally modified viral replicase the small uORFs starting at AUU codons
of the [Omega] sequence could represent a regulatory mechanism for TMV gene expression,
as they would decrease the number of ribosomes initiating at the AUG start
codons of the polymerase gene, either by direct competition through the
formation of initiation complexes or as a consequence of poor re-initiation of eukaryotic ribosomes subsequent to termination at the uORF
stop codons. The efficiency of uORF translation may be regulated during TMV
replication by interaction of this sequence with protein factors of the host
cell, like eIF-2, which is involved in initiation site recognition and stabilization of
tRNA-mRNA interactions (
37
). However, virus encoded proteins, like CaMV
trans
-activator protein (
38
), may also function in modulation of translational efficiency and it remains to
be determined whether TMV proteins make use of this mechanism for regulation of
TMV gene expression during the late stages of replication, when genomic TMV RNA
is preferentially assembled into progeny virus particles.
The technical assistance of Alice Kaufmann and Dieter Becker is gratefully
acknowledged. This work was in part supported by the Deutsche
Forschungsgemeinschaft through grant Ro 300/63 to WR.
REFERENCES
Return