Eur. J. Biochem. 209, 151-156 (1992) 0FEBS 1992
Efficiency of the 5’-terminal sequence (52) of tobacco mosaic virus RNA for the initiation of eukaryotic gene translation in Eschevickia coli Ivan G. IVANOV*, Roumjana ALEXANDROVA ’, Bojan DRAGULEV’, Denis LECLERC’, Adriana SARAFFOVA ’, Vcra MAXIMOVA and Mounir G. ABOUHAIDAR2
’
Institute of Molecular Biology, Bulgarian Academy of Sciences, Sofia, Bulgaria
’ Department of Botany, University of Toronto, Canada (Received May 14/June 22, 1992) - EJB 92 0662
Recent studies have demonstrated that the 5’ leader (Q sequence) of tobacco mosaic virus RNA has a certain enhancing capacity for translation of mRNA in both prokaryotes and eukaryotes. In order to estimate the efficiency of 52 to initiate translation of mRNA in Escherichia coli, in comparison to the Shine-Dalgarno (S/D) sequence, we have inserted eight different eukaryotic genes into two types of E. coli expression vectors containing one constitutive promoter (Pl) but different translationinitiation sites (S/D or Q A 3 sequence, respectively). The efficiency of transcription and translation in vivo was evaluated for these vectors by measuring the yield of protein and both the level and stability of mRNA. We report that substitution of Q A 3 for S/D decreases the yield of expressed protein 41900-fold and the content of gene-specific mRNA is decreased by about sevenfold. However, in comparison with the S/D sequence, the level of protein expressed under the translational control of Q A 3 is less sensitive to changes in the 5‘ coding region. We also report that the 52 sequence contains a region of 10-12 nucleotides complementary to the small ribosomal subunit RNA (rRNA) of E. coli, Eikenella corrodens and Xenopus faevis, and to the rRNA of the (small ribosomal) subunit of Oryza sativa.
It has been shown that the 5’ untranslated region of the tobacco mosaic virus (TMV) RNA (52 sequence) enhances translation of mRNA in vivo and in vitro in both eukaryotes and prokaryotes [l -61. Gallie et al. [5]constructed a series of mutant sequence derivatives of the native Q sequence and studied their ability to enhance translation of two mRNA: pglucuronidase and chloramphenicol acetyltransferase in tobacco mesophyll protoplasts, Xenopus laevis oocytes and Escherichia roli. They found that deletion of the 25-base poly(CAA) sequence from the middle part of Q increased its enhancing effect in E. roli but not in eukaryotic cells. The efficiency of this sequence (designated QA3) to initiate translation in several types of bacteria has been studied, and the conclusion was drawn that Q is a sequence functionally equivalent to the Shine-Dalgarno (S/D) sequence [6]. Unlike the native S/D sequence having the consensus AAGGAGGT [7, 81, the 5‘ untranslated region of the TMV RNA is devoid of G and is therefore unable to interact with the 3’ end of 16s rRNA. This feature allows speculation that the 52 sequence initiates translation in bacteria by alternative mechanisms of binding of mRNA to the ribosomes. Several studies have shown that the efficiency of expression of a specific gene depends very much on the AUG-codon Correspondence to M. G. AbouHaidar, Department of Botany, University of Toronto, 25 Willcocks Street, Toronto, Ontario, Canada M5S 3B2 Fax: + 416978 5878. Abbreviafions. hIF, human interferon; hCT, human calcitonin; PMV, papaya mosaic virus; S/D, Shine-Dalgarno sequence; TMV, tobacco mosaic virus.
content of the mRNA [9, lo]. Due to the different codon bias in prokaryotes and eukaryotes, it is highly possible that certain domains in a eukaryotic mRNA might occasionally block (by base pairing) the S/D sequence added in front of the AUG codon as a translational initiator. An example of this mode of interaction is found in human calcitonin (hCT) mRNA containing a 3’-terminal sequence, homologous to that of the 3’ end of E. coli 16s rRNA [ll]. It is also possible that certain domains are homologous to the ribosome-binding site and could complete with the functional S/D sequence for the 3‘ end of the 16s rRNA. An example of such a sequence is the human a1 interferon (h1FaJ mRNA, containing four repeats of the hexanucleotide AGGAGG [12] and resembling the consensus of the S/D sequence. Both hCT and hIFcll genes are expressed at low efficiency in E. coli [ l l , 13, 141. The Q sequence may initiate translation of mRNA in E. coli by bypassing the routine interaction of S/D with 16s rRNA. In this report we compare the efficiency of Q with that of S/D to initiate translation of eukaryotic mRNA in E. coli. Eight different eukaryotic genes were inserted into expression vectors containing either S/D or Q A 3 sequence, and their expression was studied in E. coli LB392 cells. The role of the 10- 12 nucleotides of Q which are complementary to the 16s rRNA of E. coli will be discussed. MATERIALS AND METHODS
Materials
[p3’P]ATP (650 Ci/mM) was purchased from ICN (Canada) and Nai2’I in NaOH from Amersham International
152 hIFa,(L+)
5'
...
~ L ~ ~ K ~ ~ ~ ~ L G TTAAGCTTATGCTCAGCTGCAAGTCAAGCTGCTCTCTGGGCTGTGATCTCCCTGAGACCCACAGCCTGGATAAC-ACC
hIFa,(L)
5'
...
c D L P E T H s L D N R R T... TTAI\GCTTATGTGTGATCTCCCTGAGACCCACAGCCTGGAT~C~ACC
S\D
n
...
hIWCys')
5'
hlFy(Cys
5'.
)
H
L
...
P
E
~
T...
...
c Y c a o P...
TTAAGCTTATGTGTTACTGCCAGGACCCA..
..
M Q D P TTARGCTTKTGCAGGRCCCA.
..
...
M Q D P TTAAGCTTATGCAGGACCCA.
..
hlFy-hCT
5'
hCT#
_ ... __ M S T P N ___ 5 ' . .. &ASGAMWTAAGCTTATGTCCACACCCAAC.. . M
C G N L TTAAGCTTATGTGCGGTAATCTG
5' ...
PRVICP)
... D
...
M C 0 L P E T H S L D N R R TTAIGCTTATGTGTGATCTCCCTGAGACCCACAGCCTG~T~C-ACC
5'...
hIF*,(E,L)
~
Fig. 1. 5'-terminal region of eight eukaryotic genes designed for expression in E. coli. Shaded area, S/D sequence. Remainders of signal-peptide codons are underlined. Tandems of argine codons are underlined twice. The natural tandem arginine codons AGGAGG (which are rare in E. coh) are preserved in the genes hIFal (L') and hIFcl, (L-) and changed to CGCCGG in the hTFcll (B,L-) gene. S/D
R 5'
TCGAGTATTTTTACAACAATTACCTTACAATTACTATTTACAATTACA
3'
CATAAAAATGTTGTTAATGGAATGTTAATGATAAATGTTA
XhoI
2
t
4
t
6
3' 5'
HindIII
Fig. 2. Synthesis of 9 4 3 sequence. Six synthetic oligonucleotides used for construction of 9 4 3 are marked by arrows.
(England). Restriction endonucleases, T4 polynucleotide kinase and T4 DNA ligase were products of New England Biolabs. Synthetic hCT (RIA standard) was purchased from Peninsula Labs, USA. Growth media (bacto-tryptone, yeast extract and bacto-agar) were purchased from Difco (USA).
Genes and regulatory sequences The 5' end sequences of the different genes used are shown in Fig. 1. Three variants of the hIFa, were used in this study: hIFoll (L'), hIFcr, (L-) and hIFa, (B1L-). hIFal (L') is a genomic DNA sequence coding for 177 amino acid residues (166 belonging to the mature hIFal, 10 to the signal peptide, and one methionine). The isolation and modification of this gene is described elsewhere [14]. hIFal (L-) is derived from hIFcr, by deleting the signal peptide codons. hIFal (BlL-) is a derivative of hIFa, (L-) from which the first tandem of arginine codons AGGAGG is changed to CGTCGG. The hIFy (Cys') gene codes for 147 amino acid residues (three belonging to the signal peptide and one methionine) and its structure is described by others [15]. The hlFy (Cys ) gene is derived from the hIFy (Cys') by deleting the three signal-peptide codons (Cys-Tyr-Cys). The construction of the hybrid hIFy-hCT gene is described elsewhere [13]. The tetrameric hCT gene (hCT,) codes for four copies (32 amino acids each) of head-to-tail-linked hCT molecules [I 61. The papaya mosaic virus (PMV) coat-protein gene was isohted from a cDNA library [l?]. This gene codes for a 24kDa viral-capsid protein. The 5' ends of the structural genes used in this study are shown in Fig. 1. A distance of eight nucleotides between the start codon ATG and the S/D sequence is preserved in all constructs. This distance was found to be optimal for the expression in E. coli. The 0 4 3 DNA was constructed from six synthetic oligonucleotides, as shown in Fig. 2. The 5243 sequence is identical to that published by Gallie et al. [5], with the exception that two restriction sites, XhoI and HindlI1, have becn added for convenient insertion into the expression vectors (see below).
H
I
I
Xhol
EcoRl
R
pBR322
~
bi
I
EcoRl
I
tet
Gene
H
H
I
1
Hindlll
1
Hin'dlll
~ 3Gene H i - - \ -
1
Xhol Hindlll
pBR322
tet
pBR322
I
Hindlll
Fig. 3. Expression cassettes substituted for the EcoRI/HindIlI fragment of the plasmid pBR322. PI is a strong synthetic constitutive promoter variant of the T5 phage early promoter [18]. S/D contains the consensus sequence AAGGAGGT, and the primary structure of 0243 is shown in Fig. 1 . Gene, one of the eighl genes described in the text; tet, tctracyclme-transacetylase gene of pBR322.
Plasmid constructions Expression plasmids for the eight structural genes, which are under the translational control of either S/D or D243 sequences, were constructed in pBR322. The 29-bp fragment between EcoRI and HindIII of pBR322 was removed and replaced with one of the two cassettes (PI S/D gene or PI 5243 gene) shown in Fig. 3.
Oligonucleotide synthesis, purification, phosphorylation and assembly Oligonucleotides for construction of 043 sequence and for hybridization were synthesized by the solid-phase, phosphoramidite method using automated gene assemblers (Pharmacia and Applied Biosystems). After unblocking with concentrated ammonia, the oligonucleotides were purified by HPLC on a Mono Q column (Pharmacia). Oligonucleotides for hybridization were phosphorylated using [ Y - ~ ~ P ] Aand T P polynucleotide kinase to a specific activity higher than lo7 cpm/yg and purified by DEAE-cellulose (mini-column) chromatography. Oligonucleotides used for synthesis of DA3 DNA were phosphorylated, purified from non-phosphorylated forms and ligated as described in [16].
Bacterial transformation and cultivation Competent E. roli LE 392 cells were prepared and transformed as recommended by Hanahan [19]. Bacterial cells, for
~
~
L
153 quantitative determination of recombinant protein and gene specific mRNA, were prepared as follows: 20 ml LuriaBertani medium (in 100-ml flasks) containing 50 pg/ml ampicillin was inoculated and the cellular suspension was shaken (300 rpm) at 37°C for 8 h. Each cell culture was divided into two and used for analysis of both protein and mRNA. Colony and dot hybridization The method of RNA colony hybridization [20] was used for selection of clones containing genes in the correct orientation and for the quantitative determination of specific mRNA in E. coli cells harbouring the expression plasmids pP1 S/D and pP,QA3. Total cellular RNA was isolated by standard procedures from 10 ml bacterial cells. RNA was treated with DNase I (RNase free), deproteinized and precipitated with ethanol. The RNA was dissolved in distilled water which had been treated with diethyl pyrocarbonate, and the concentration of RNA was determined by A260measurement. Samples of 40 pg RNA were evaporated to dryness in a Speed Vac centrifuge, the RNA was dissolved in 20 pl300 mM NaCl and 30 mM sodium citrate, pH 7.0, containing 10% formaldehyde, heated at 60°C for 15 min and spotted onto nitrocellulose filters. The filters were baked in a vacuum oven at 80°C for 1 h and prehybridized and hybridized as described elsewhere [20]. Filters were air-dried and exposed to X-ray films. After autoradiography filters were cut and the 32P radioactivity was measured by Cerenkov counting in a Beckman scintillation counter. Determination of in vivo stability of mRNA To study the in vivo stability of mRNA, RNA synthesis was blocked by rifampicin and nalidixic acid as described [21]. 20 ml Luria-Bertani medium (100-ml flask) containing 50 pg/ ml ampicillin were inoculated with an overnight culture (1 :50 by vol.) and cultivated with vigorous shaking at 37'C for 1 h. Nalidixic acid and rifampicin were added simultaneously to final concentrations of 0.5 p,g/ml and 60 pg/ml, respectively. Samples of bacterial cells with ASq0= 2.0 were harvested at the following time intervals: 0 (zero time control: sample is taken immediately upon addition of nalidixic acid and rifampicin), 10, 20, 30, 40, 50, 60, 70, 80, 120, 180, 240, 300, 500 and 600 s. The samples were frozen at - 70 'C, thawed on ice and spun down at 4°C. Cells were resuspended in ice-cold solution containing 50 mM sodium acetate, pH 5.0, 1 mM EDTA and 0.5% SDS. Total cellular RNA was isolated, spotted onto nitrocellulose filters and hybridized with j2P-labelled oligonucleotides specific for the corresponding mRNA. The radioactivity of each spot was mcasurcd by Cerenkov counting and plotted versus the time (is). The half-life of mRNA was defined as the time corresponding to a 50% decrease in radioactivity of each spot, compared with the level of the zero-time control. Determination of plasmid copy number Plasmid copy number was measured for all the expression vectors used, as described in [22]. Determination of level of protein expression The yield of hIFcll and hIFy was measured by the antiviral activity (protective effect on Wish cells against the cytopathic
action of the vesicular stomatitis virus) of crude bacterial lysates. Cells with Asg0= 5 were sonicated in 1 ml 10 M Tris, pI1 7.4, and 1 mM EDTA and centrifuged in an Eppendorf centrifuge at 4°C for 5min. The clear lysates were serially diluted twofold and added to Wish cells grown in 96-well plates. After 4 h at 37"C, the cells were infected with vesicular stomatitis virus (multiplicity of infection lo), and after 24 h of incubation at 37°C the cells were stained with crystal violet. The titer was quantitatively monitored using a micro-ELISA reader (LKB) and the antiviral titer was expressed as the reciprocal of the dilution inhibiting 50% of the cytopathic effect. The yield of hCT was quantitated by RIA. This analysis was based on the usc of synthctic hCT as a standard and highly specific rabbit antiserum for precipitation of the antigen/antibody complex. RIA was carried out as follows. 100 p1 synthetic hCT in 50 mM phosphate buffer, pH 7.8, or E. coli lysate (prepared as above) was mixed with 200 pl antiserum (working dilution 1 :3000) and incubated at 4°C for 12 h. '251-labelled standard synthetic hCT (1000015000 cpm in 100 pl) was added and the mixture was incubated for 6 h at 4°C. The antibody-bound radioactivity was precipitated with 100 p1 anti-(rabbit Ig) serum (dilution 1 : 30) in 50 mM phosphate, pH 7.8. After 4 h incubation at 4 'C, the antigen/antibody complex was pelleted at 2500 x g for 20 min. For every sample tested, the zero and non-specific binding wcre determined. The level of expression of the PMV coat-protein gene was determined by Western blotting. Bacteria containing the PMV coat-protein gene were grown overnight in 20 ml LuriaBertani medium containing 1% dextrose at 37 "C. Bacteria at A260= 1 were sedimented and resuspended by sonication in 0.1 M Na2C03/0.1M dithiothreitol. The suspension was adjusted to 1.7% SDS and 10% sucrose, boiled and electrophoresed in SDSiPACJE [23]. Following electrophoresis, the proteins were transferred onto nitrocellulose membranes using a Transblot (Bio-Rad), according to suppliers' instructions. Rabbit polyclonal antisera were used to identify the PMVi coat protein according to Bailly and Coleman [24].
RESULTS AXD DISCUSSION The untranslated 5' leader sequence of TMV RNA is described in the literature as a universal enhancer of translation of foreign gene transcripts in vivo and in vitro for both prokaryotes and eukaryotes [l -61. In all these studies, only prokaryotes genes (chloramphenicol acetyltransferase, pglucuronidase and neomycin phosphotransferase) have been used as a model for quantitative estimation of translation enhancement. The mRNA of these genes are translated efficiently in both prokaryotic and eukaryotic cells and cell-free systems. Low levels of expression of these genes (due to selfinitiation of translation) were found, even in the absence of specific translation initiators. However, when the 52 sequence is added in front of these genes, translation activity is enhanced. We have investigated the efficiency of 52 sequence to initiate translation of eukaryotic mRNA in E. coli cells and checked the functional equivalence [6] of L? sequence by comparing its efficiency with that of the consensus S/D sequence. Eight different eukaryotic genes were inserted into two types of expression vectors (pPIS/D and ~P~S2.43) and used in this study. As shown in Fig. 3, both vectors differ only by the nature of the translation-initiation site. In the plasmid
154 Table 1. Effect of Q43 on the expression of eukaryotic proteins in E. cnli LE392. Yields of hIFcll and hIFy were calculated from the spccific antiviral activities of the protein? (10 IJ/ng for hIFcll and 50 U/ng for the hTFy). Yields of hCT and hIFy-hCT were determined by RIA. For all genes used, the removal of S/D or QA3 led to undetectable amounts of protein (data not shown). Yield of PMV coat protein is presented as arbitrary units based on the band intensity
in immunoblots. Gene
Yield of recombinant protein SiD
5243
mgll
hlFcr, (L ' j hIFml (L-) hlb'iccl (BIL-)
hIFy (Cys* j hIFy (Cys-)
hIFy-hCT hCT, PMV coat protein
2.70 37.00 22.00 10.00 285.00 0.11 2.55 1.00
Decrease
-fold
0.61 3.4Y
3.60 0.12 0.15 < 0.01 0.05 0.3
4.4 10.6 6.1 83.9 1900.0 > 10 51 .o
3.3
pPIS/D, a strong synthetic ribosome-binding site (containing the consensus AAGGAGGT sequence) was used as a translation initiator, whereas in pP,QA3, the translation initiator was a synthetic derivative of R (Qd3) known to be much more effective in E. coli than the native Q sequence [5]. The coding sequences used in our study were either natural [(hIFal (L+)], isolated from a gene library, semi-synthetic [(hlFa, (L-), hlFal(BILp) and PMV coat-protein gene] or fully synthetic genes [hlFy(Cys+),hIFy(CysC), hIFy-hCT and hCT,]. They were suitable for the present study for the following reasons. Their mRNA were unable to initiate translation in E. coli cells in the absence of translation-initiation sequences (unpublished data). The efficiency of expression of these genes varied [from 30-50% of the total cellular protein for hIFy(Cys-) to negligible amounts for hIFy-hCT and PMV coat protein], and they could be classified as follows: highly expressable [hIFy(Cys-), hIFy(Cys+), hCTt, hIFal(L-) and hIFr,(BILp)]; expressable with moderate efficiency [hIFa, (L')]; expressable with low efficiency [hIFy-hCT and PMV coat-protein gene]. The same design of the 5' ends of these genes ensured an identical nucleotide context in the corresponding mRNA between the + 1 nucleotide and the start codon AUG. The presence of three groups of related genes allows comparison of the significance of codon use downstream of AUG for the efficiency of expression in both types of vectors pPIS/D and pP152d3. The sensitive methods used for quantitative analysis of the yield of protein make it possible to detect small differences in the level of gene expression and to derive quantitative values for the enhancing or attenuation effect of Q in comparison to the S/D sequence. Level of protein expression in E. coli LE 392 cells transformed with expression plasmids pPIS/D and pP1QA3 The level of gene expression (as protein) was determined for all eight genes under the translational control of the S/D (consensus) or 5243 sequence, and the results are presented in Table 1. The yield of recombinant protein obtained is lower for the expression plasmids containing 5243 sequence. These data show that the level of expression of the eight genes varies over broad range for both series of expression vectors (PIS/D
gene and P1Q43 gene), although their 5' ends are all identical (Fig. 1). However, the variation in yield for different genes is less significant (e.g. 360 times between hIFy-hCT and hIFal(BIL-) for the constructs containing QA3 sequence compared to those with S/D sequence (Table 1).The initiation of translation by S/D, unlike that of QA3, is more sensitive to changes in the 5' coding region of the gene. For example, the deletion of only three codons (Cys-Tyr-Cys) of the signal peptide of the hIFy gene resulted in a %-fold increase in the yield of protein under the translation control of S/D sequence, and only 1.2-fold increase in the presence of 5243 sequence (Fig. 1, Table 1). Consequently, a decrease of 1900-fold resulted in the level of expression of hlFy(Cys-) gene between the S/D and 5243 as translation initiators (Table 1). Furthermore, the level and half-life of mRNA is also decreased in the presence of QA3 sequence (see below). All eight genes used in this study gave no detectable protein when the expression vectors did not contain translation initiators (data not shown). Our data (Table 1) clearly showed that independently of the lower yield obtained from the vectors pP1QA3, the TMV leader sequence is a real initiator of translation of mRNA in E. coli. These results confirm the conclusions of others [l - 61. The lower level of protein expressed from vectors containing 9 4 3 sequence may be explained by lower plasmid copy number, lower level of mRNA, shorter half-life of mRNA or lower translation-initiation efficiency. Plasmid copy number was determined for both series of expression plasmids pPIS/ D and pP1QA3 and the results are presented in Table 2. Substitution of 5243 for S/D sequence did not change detectably the plasmid copy number values. Surprisingly, the expression of the PMV coat-protein gene lowered the plasmid copy number by about 2.5-fold. It has been shown that below pH 7.5,PMV coat protein binds non-specifically to any RNA in vitro [25]. We can speculate that the effect of PMV coat protein on plasmid replication may be due to a non-specific interaction with the primer RNA (RNA I). Level and half-life of mRNA transcribed in vivo from the expression plasmids pPIS/D and pP15243 The P1 promoter in plasmids pPIS/D and pP10d3 initiates transcription of dicistronic mRNA (due to the absence of a transcription terminator after the eukaryotic gene). The 5' end of these mRNA are specific for the inserted gene, and the 3' termini represent transcripts of the tetracycline-transacetylase gene of pBR322 which is common for all the expression vectors. This allows the determination of mRNA contents for all the constructions by comparing the hybridization of equal amounts of total cellular RNA with the same hybridization probe specific for the tetracycline-transacetylase gene. The results presented in Table 2 show that the substitution of 5243 for S/D in the expression vcctor dccrcascd the level of mRNA in the cell for all the genes studied. The amounts of mRNA (32P radioactivity) in cells correlated with the plasmid copy number (e.g. the mRNA of PMV coat protein compared to other genes is about three times less; Table 2). The effect of QA3 in this case was not as dramatic as for the protein (Table 1) and did not exceed a 7.1-fold decrease in the content of mRNA. The half-life of mRNA transcribed from plasmids pPIS/ D and pPlQA3 was studied after double blocking of transcription by rifampicin and nalidixic acid (for details, see Materials and Methods). The half-life of mRNA produced from plasmid pP152A3 varied over 60-70 s, whereas that of the mRNA
155 Table 2. Plasmid copy number, stability and content of mRNA in E. coli LE 392 expressing eukaryotic genes under the control of PI S/D or PI 043. Decrease in mRNA was assessed using a common tetracyclinetransacctylase mRNA, 32P-labelledspecific probe. Data represent the ratio between the levels of mKNA containing S/D and 5243. Gene
Plasmid copy number S/D
5243
Half-life of mRNAQd3 S/D
23 24 21 20 21 23 22 8
22 23 22 21 21 20 21 9
70 70 70 80 90 70 70 70
1341 1360 CGCUAGUAAUCGUGGAUCAG AUCAUUt@CAUU
5‘
2.
5’ 3’
1341 1360 UCGCUAGUAAUCGCAGGUCA AUCAUUAACAUUC
3’ 5’
3.
5’ 3’
511 530 GGCCCUGUAAUUGGAAUGAG CAUUAAWUU$C
3’ 5’
4.
5’ 3’
140 163 GGAUAACCGUAGUAAUUCUAGAGC UAUCAUUAAGAUU
-fold 65 70 70 65 60 65 60 70
3’ 5’
5
Decrease in mRNA
843
S hIFal (L’) hIFEl (L-) hIFa, (B,L-) hIFy (Cys’) hIFy (Cys-) hIFy-hCT hCT, PMV coat proteiii
3’
1.
2.8 1.5 2.4 3.2 7.1 1.9 3.5 1.4
containing the S/D sequence was 70-90 s (Table 2). For most of the genes studied, the destabilizing effect of 5243 was negligible ( 5 - 10 s only) with the exception of the mRNA f o r hIFy(Cys ’) and hIFy(Cys-), where the destabilking effect is in the order 15 - 30 s (Table 2). In general, the lower level of mRNA in E. coli cells containing expression plasmids pP104 3 cannot be explained by destabilization (increased RNase sensitivity) of mRNA caused by 52243. The results presented in this study clearly demonstrate that the substitution of 5243 for S/D in the expression vector decreases both the yield of recombinant protein and the content of mRNA. The lower level of mRNA cannot be explained either by the reduced copy number of expression plasmids or by the decreased stability (half-life) of mRNA. Consequently, we conclude that the presence of 5243 in the plasmid has an attenuating effect on transcription. Furthermore, our results show that there is no direct correlation between the extent of decrease in the level of mRNA and the yield of protein (compare the data in Tables 1 and 2). From the result that protein yield is more sensitive to the substitution of Q43 for the S/D sequence than the mKNA we that Q243is less efficient as a translation-initiation signal in vivo than the S/D sequence. This can explain the discrepancy between the shorter half-life of mRNA transcribed from the expression vectors pPl52A3-hIFy(Cys+) and pPlf2243-hIFy(Cys-) (Table 2). 0 sequence contains a domain complementary to rRNA of prokaryotes and eukaryotes
The mechanism of initiation of translation of mRNA in E. coli by the TMV i2 sequence is still unknown [6]. The 52 sequence (and its derivative 5143) does not contain any G residues, and cannot therefore interact with the 3‘ end of the small ribosomal subunit rRNA. To find an alternative site of interaction of i2 and Q A 3 sequences with 16s rRNA of E. coli a computer search for complementarity was carried out. As seen in Fig. 4, a 12-nucleotide sequence (with one mismatch) was found to be complementary to 16s rRNA at positions 1344- 1355. The corresponding complementary domain in 52 comprises the hexanucleotide AAUUAC, occurring three times in both the native 52 and 0243 sequences [S]. This repeat has been shown to be important for the efficient initiation of translation of mRNA by Q in both prokaryotes and
3’
Fig. 4. Complementarity between TMV s2 sequence (3’ - 5’) and (5’3’): (1) E. coli 16s rRNA [26]; (2) E. corrodens 16s rRNA (271; (3) X . laevzs 185 rRNA [28]; (4) Oryza sativa 17s rRNA [29]. Mismatched bases are shaded.
eukaryotes (tobacco protoplasts and X . laevis oocytes) [5, 61. A search for complementarity was carried out with other rRNA. Fig. 4 shows that the same nucleotide sequence was identified at the same location in the 16s RNA of Eikenella corrodens. In both organisms (E. coli and E. corrodens), this sequence was about 150 nucleotides from the 3‘ end of 16s rRNA. The natural site of interaction of the ribosomes with S/D sequence has been shown to be at the 3‘ end of 16s rRNA [7]. According to the three-dimensional model of the E. coli 30s ribosomal subunit [30], the 16s rRNA region complementary to the TMV Q sequence (positions 1344- 1355) is situated in a domain close to that where the S/D sequence interacts with the 30s ribosomal subunit. These two domains form a groove where the mRNA is presumably interacting with the ribosomal subunit. This result strengthens the role of Q as an alternative binding site for mRNA. Q sequence was also found to be complementary to a region near the 5’ ends of the 18s rRNA of the small ribosomal subunit of X . fuevis [l -41 and to the 17s rRNA of Orjiza sativa. Since the site of interaction ofmRNA with eukaryotic ribosomes is not known, our data suggest that the mRNA-binding site may be localized in the first third of the eukaryotic small ribosomal subunit rRNA (at least for mRNA containing Q-like translation initiators). This work was partly funded by grants from the Natural Sciences and Engineering Research Council Canada and Ontario Ministry of Food and ~ ~to M. G , AbouHaidar ~ i and by a~ grant 2681~ 06.05.86 from the Ministry of Industry and Technology, Bulgaria, to I. Ivanov.
REFERENCES 1. Sleat, D. E., Gallie, D. R., Jefferson, R. A., Bevan, M. W., Turner, P. C. & Wilson, T. M. A. (1987) Characterization of the 5’-
leader scquence of tobacco mosaic virus RNA as a general enhancer of translation in vitro, Gene (Amst.) 66, 217-225. 2. Gallie, D. R., Sleat, D. E., Watts, J. W., Turner, P. C. & Wilson, T. M. A. (1987) The 5’-leader sequence of tobacco mosaic virus RNA enhances the expression or forcign gene transcripts in vitro and in vivo, Nucleic Acids Res. 15, 3257 - 3272. 3. Gallie; D. R., Walbot, V. & Hershey, J. W. (1988) The ribosomal fraction mediates the translation enhancement associated with the 5’-leader of tobacco mosaic virus, Nudeic Acids Res. 17, 8675 - 8695. 4. Sleat, D. E., Hull, R., Turner, P. C. &Wilson, T. M. A. (1988) Studies on the mechanism of translational enhancement by the 5’4eader sequence of tobacco mosaic virus RNA, Eur. J . Biochem. 175, 75 - 86.
l
156 5. Gallie, D. R., Sleat, D. E., Watts, J. W., Turner, P. C. &Wilson; T. M. A. (1988) Mutational analysis of the tobacco mosaic virus 5’-leader for altered ability to enhance translation, Nucleic Acids Res. 16, 883 - 893. 6. Gallie, D. R. & Kado, C. I. (1989) A translational enhancer derived from tobacco mosaic virus is functionally equivalent to a Shine-Dalgarno sequence, Proc. Nut1 Acad. Sci. USA 86, 129- 132. 7. Shine, J. & Dalgdrno. L. (1975) Dctermination of cistron specificity in bacterial ribosomcs, Nature 254, 34-38. 8. Gold, L., Pribnow, D., Schneider, T., Shinedling, S., Singer, B. & Stormo, G. (1981) Translational initiation in prokaryotes, Annu. Rev. Microbiol. 35, 365 -403. 9. Koxak, M. (1983) Comparison of initiation of protein synthesis in procaryotes, eucaryotes and organelles, Microbiul. Rev. 47, 1-45. 10. Kozak, M. (1989) The scanning model for translation: an update, J . Cell Biol. 108, 229 - 241. 11. Bachvarov, D.. Wishart, P., Usheva, A., Ivanov, I. & Jay, E. (1990) Expression in E. coli of a human Va18-calcitonin gene by fusion to the chloramphenicol acetyl transferase (CAY) gene, Biol. Zentralbl. 109, 113- 117. 12. Nagarta, S., Mantei, N. & Weissman, C. (1980) The slructure of one of the eight or more distinct chromosomal genes for human interferon a, Nature 287, 401 --408. 13. Ivanov, I.. Gigova, L. & Jay, E. (1987) Chemical synthesis and exprcssion in E. coli of a human Vals-calcitonin gene by fusion to a synthetic human interferon-y gene, FEBS Lett. 210, 5660. 14. Ivanov, I., Markova, N., Bachvarov, D., Alexciev, K., Sarafova, A., Maximova, V., Tsaneva, I. & Markov, G. (1989) Constitutive expression of a native human interferon-a, gene in E. coli, int. J . Biochem. 21.983-985. 15. Jay, E., Rommcns, J., Cloncy, L., McKnight, D., Lutze-Wallace, C., Wishart, P., Lin, W., Asundi, V., Dowood, M. & Jay, F. (1984) High-level of expression of a chemically synthesized gene for human interferon-y using a prokaryotic expression vector, Proc. Natl Acad. Sci. USA 81. 2290 - 2294. 16. Ivanov, I . (1990) Shotgun concatenation of synthetic genes: construction of concatemeric human calcitonin genes, Anal. Biochern. 189,213-216. 17. Sit, T. L., AbouHaidar, M. G. & Holy, S. (1989) Nucleotide sequence of papaya mosaic virus RNA, J . tien. Virol. 70, 2325 -2331. 18. Rommens, J., MacKnight, D., Pomeroy-Claney, L. & Jay, E. (1 983) Gene expression: chemical synthesis and molecular
19. 20. 21. 22. 23. 24.
25.
26.
27.
28. 29. 30.
cloning of a bacteriophage T5 (P25) early promoter, Nucluic Acids Res. 11, 5921- 5940. Hanahan, D. (1983) Studies on transformation of E. coli with plasmids, J. Mol. Biol. 166, 557-580. Ivanov, I. & Gigova, L. (1986) RNA colony hybridization method, Gene (Amst.) 46,287 - 290. Silengo, L., Nicolaev, N., Schlessinger, D. & Imamoto, 1. (1974) Stabilizalion of mRNA with polar effects in an E. coli mutant, Mol. tien. tienet. 134, 1 - 19. Ivanov, I. & Bachvarov, D. (1987) Determination of plasmid copy number by the boiling method, Anal. Biochem. 165, 137-141. Laemmli, U. K. (1970) Cleavage of structural proteins during the assembly of the hcad of bactcriophages T, Nature 227, 680685. Bailly, J. & Coleman, J. R. (1988) Effect of C 0 2 concentration on protcin biosynthesis and carbonic anhydrase expression in Chlamidornonus reinhardtii, Plant Physiol. (Bethesda) 87,833 840. AbouHaidar, M. G . & Erickson, J. W. (1985) Structure and in vitro assembly of papaya mosaic virus, in Molecular plant virology (Davies, J., ed.) pp. 85-121, CRC Press Inc., Boca Raton, Florida. Schnare, M. N. & Gray, M. W. (1982) 3’-terminal sequence of wheat mitochondria1 18s ribosomal RNA: furlher evidence of a eubacterial evolutionary origin, Nucleic Acids Res. 10,3921 3932. Dewhirst; F. E., Paster, €3. J. & Bright, P. L. (1989) Chromobacteriim, Eikenella, Klingella, Neisseriu, Simonesiellu and Vitreoscilla species comprise a major branch of the /Igroup Proteobacteria by 16s ribosomal ribonucleic acid sequence comparison : transfer of Eikenella and Simonsiella l o the family Neisseriaceae (emend.), Int. J . Syst. Bacteriol. 39, 258 - 266. Maden, B. E. H. (1986) Idcntification of the locations of the methyl groups in 18s rihosomal RNA from Xenupus laevis and man, J . Mol. Biol. 189, 681 -699. Takaiwa, F.. Oono, K. & Suqiura. M. (1984) The complete nucleotide sequence of a rice 17s rRXA gene, Nucleic Acid.7 Res. 12, 5441 - 5448. Brimacombe, R., Atmadja, J., Stiege, W. & Schbler. D. (1988) A detailed model of the three-dimensional structure of Escherichia coli 16s ribosomal RNA in situ in the 30s subunit, J. Mol. Biol. 199,115--136.
Efficiency of the 5’-terminal sequence (52) of tobacco mosaic virus RNA for the initiation of eukaryotic gene translation in Eschevickia coli Ivan G. IVANOV*, Roumjana ALEXANDROVA ’, Bojan DRAGULEV’, Denis LECLERC’, Adriana SARAFFOVA ’, Vcra MAXIMOVA and Mounir G. ABOUHAIDAR2
’
Institute of Molecular Biology, Bulgarian Academy of Sciences, Sofia, Bulgaria
’ Department of Botany, University of Toronto, Canada (Received May 14/June 22, 1992) - EJB 92 0662
Recent studies have demonstrated that the 5’ leader (Q sequence) of tobacco mosaic virus RNA has a certain enhancing capacity for translation of mRNA in both prokaryotes and eukaryotes. In order to estimate the efficiency of 52 to initiate translation of mRNA in Escherichia coli, in comparison to the Shine-Dalgarno (S/D) sequence, we have inserted eight different eukaryotic genes into two types of E. coli expression vectors containing one constitutive promoter (Pl) but different translationinitiation sites (S/D or Q A 3 sequence, respectively). The efficiency of transcription and translation in vivo was evaluated for these vectors by measuring the yield of protein and both the level and stability of mRNA. We report that substitution of Q A 3 for S/D decreases the yield of expressed protein 41900-fold and the content of gene-specific mRNA is decreased by about sevenfold. However, in comparison with the S/D sequence, the level of protein expressed under the translational control of Q A 3 is less sensitive to changes in the 5‘ coding region. We also report that the 52 sequence contains a region of 10-12 nucleotides complementary to the small ribosomal subunit RNA (rRNA) of E. coli, Eikenella corrodens and Xenopus faevis, and to the rRNA of the (small ribosomal) subunit of Oryza sativa.
It has been shown that the 5’ untranslated region of the tobacco mosaic virus (TMV) RNA (52 sequence) enhances translation of mRNA in vivo and in vitro in both eukaryotes and prokaryotes [l -61. Gallie et al. [5]constructed a series of mutant sequence derivatives of the native Q sequence and studied their ability to enhance translation of two mRNA: pglucuronidase and chloramphenicol acetyltransferase in tobacco mesophyll protoplasts, Xenopus laevis oocytes and Escherichia roli. They found that deletion of the 25-base poly(CAA) sequence from the middle part of Q increased its enhancing effect in E. roli but not in eukaryotic cells. The efficiency of this sequence (designated QA3) to initiate translation in several types of bacteria has been studied, and the conclusion was drawn that Q is a sequence functionally equivalent to the Shine-Dalgarno (S/D) sequence [6]. Unlike the native S/D sequence having the consensus AAGGAGGT [7, 81, the 5‘ untranslated region of the TMV RNA is devoid of G and is therefore unable to interact with the 3’ end of 16s rRNA. This feature allows speculation that the 52 sequence initiates translation in bacteria by alternative mechanisms of binding of mRNA to the ribosomes. Several studies have shown that the efficiency of expression of a specific gene depends very much on the AUG-codon Correspondence to M. G. AbouHaidar, Department of Botany, University of Toronto, 25 Willcocks Street, Toronto, Ontario, Canada M5S 3B2 Fax: + 416978 5878. Abbreviafions. hIF, human interferon; hCT, human calcitonin; PMV, papaya mosaic virus; S/D, Shine-Dalgarno sequence; TMV, tobacco mosaic virus.
content of the mRNA [9, lo]. Due to the different codon bias in prokaryotes and eukaryotes, it is highly possible that certain domains in a eukaryotic mRNA might occasionally block (by base pairing) the S/D sequence added in front of the AUG codon as a translational initiator. An example of this mode of interaction is found in human calcitonin (hCT) mRNA containing a 3’-terminal sequence, homologous to that of the 3’ end of E. coli 16s rRNA [ll]. It is also possible that certain domains are homologous to the ribosome-binding site and could complete with the functional S/D sequence for the 3‘ end of the 16s rRNA. An example of such a sequence is the human a1 interferon (h1FaJ mRNA, containing four repeats of the hexanucleotide AGGAGG [12] and resembling the consensus of the S/D sequence. Both hCT and hIFcll genes are expressed at low efficiency in E. coli [ l l , 13, 141. The Q sequence may initiate translation of mRNA in E. coli by bypassing the routine interaction of S/D with 16s rRNA. In this report we compare the efficiency of Q with that of S/D to initiate translation of eukaryotic mRNA in E. coli. Eight different eukaryotic genes were inserted into expression vectors containing either S/D or Q A 3 sequence, and their expression was studied in E. coli LB392 cells. The role of the 10- 12 nucleotides of Q which are complementary to the 16s rRNA of E. coli will be discussed. MATERIALS AND METHODS
Materials
[p3’P]ATP (650 Ci/mM) was purchased from ICN (Canada) and Nai2’I in NaOH from Amersham International
152 hIFa,(L+)
5'
...
~ L ~ ~ K ~ ~ ~ ~ L G TTAAGCTTATGCTCAGCTGCAAGTCAAGCTGCTCTCTGGGCTGTGATCTCCCTGAGACCCACAGCCTGGATAAC-ACC
hIFa,(L)
5'
...
c D L P E T H s L D N R R T... TTAI\GCTTATGTGTGATCTCCCTGAGACCCACAGCCTGGAT~C~ACC
S\D
n
...
hIWCys')
5'
hlFy(Cys
5'.
)
H
L
...
P
E
~
T...
...
c Y c a o P...
TTAAGCTTATGTGTTACTGCCAGGACCCA..
..
M Q D P TTARGCTTKTGCAGGRCCCA.
..
...
M Q D P TTAAGCTTATGCAGGACCCA.
..
hlFy-hCT
5'
hCT#
_ ... __ M S T P N ___ 5 ' . .. &ASGAMWTAAGCTTATGTCCACACCCAAC.. . M
C G N L TTAAGCTTATGTGCGGTAATCTG
5' ...
PRVICP)
... D
...
M C 0 L P E T H S L D N R R TTAIGCTTATGTGTGATCTCCCTGAGACCCACAGCCTG~T~C-ACC
5'...
hIF*,(E,L)
~
Fig. 1. 5'-terminal region of eight eukaryotic genes designed for expression in E. coli. Shaded area, S/D sequence. Remainders of signal-peptide codons are underlined. Tandems of argine codons are underlined twice. The natural tandem arginine codons AGGAGG (which are rare in E. coh) are preserved in the genes hIFal (L') and hIFcl, (L-) and changed to CGCCGG in the hTFcll (B,L-) gene. S/D
R 5'
TCGAGTATTTTTACAACAATTACCTTACAATTACTATTTACAATTACA
3'
CATAAAAATGTTGTTAATGGAATGTTAATGATAAATGTTA
XhoI
2
t
4
t
6
3' 5'
HindIII
Fig. 2. Synthesis of 9 4 3 sequence. Six synthetic oligonucleotides used for construction of 9 4 3 are marked by arrows.
(England). Restriction endonucleases, T4 polynucleotide kinase and T4 DNA ligase were products of New England Biolabs. Synthetic hCT (RIA standard) was purchased from Peninsula Labs, USA. Growth media (bacto-tryptone, yeast extract and bacto-agar) were purchased from Difco (USA).
Genes and regulatory sequences The 5' end sequences of the different genes used are shown in Fig. 1. Three variants of the hIFa, were used in this study: hIFoll (L'), hIFcr, (L-) and hIFa, (B1L-). hIFal (L') is a genomic DNA sequence coding for 177 amino acid residues (166 belonging to the mature hIFal, 10 to the signal peptide, and one methionine). The isolation and modification of this gene is described elsewhere [14]. hIFal (L-) is derived from hIFcr, by deleting the signal peptide codons. hIFal (BlL-) is a derivative of hIFa, (L-) from which the first tandem of arginine codons AGGAGG is changed to CGTCGG. The hIFy (Cys') gene codes for 147 amino acid residues (three belonging to the signal peptide and one methionine) and its structure is described by others [15]. The hlFy (Cys ) gene is derived from the hIFy (Cys') by deleting the three signal-peptide codons (Cys-Tyr-Cys). The construction of the hybrid hIFy-hCT gene is described elsewhere [13]. The tetrameric hCT gene (hCT,) codes for four copies (32 amino acids each) of head-to-tail-linked hCT molecules [I 61. The papaya mosaic virus (PMV) coat-protein gene was isohted from a cDNA library [l?]. This gene codes for a 24kDa viral-capsid protein. The 5' ends of the structural genes used in this study are shown in Fig. 1. A distance of eight nucleotides between the start codon ATG and the S/D sequence is preserved in all constructs. This distance was found to be optimal for the expression in E. coli. The 0 4 3 DNA was constructed from six synthetic oligonucleotides, as shown in Fig. 2. The 5243 sequence is identical to that published by Gallie et al. [5], with the exception that two restriction sites, XhoI and HindlI1, have becn added for convenient insertion into the expression vectors (see below).
H
I
I
Xhol
EcoRl
R
pBR322
~
bi
I
EcoRl
I
tet
Gene
H
H
I
1
Hindlll
1
Hin'dlll
~ 3Gene H i - - \ -
1
Xhol Hindlll
pBR322
tet
pBR322
I
Hindlll
Fig. 3. Expression cassettes substituted for the EcoRI/HindIlI fragment of the plasmid pBR322. PI is a strong synthetic constitutive promoter variant of the T5 phage early promoter [18]. S/D contains the consensus sequence AAGGAGGT, and the primary structure of 0243 is shown in Fig. 1 . Gene, one of the eighl genes described in the text; tet, tctracyclme-transacetylase gene of pBR322.
Plasmid constructions Expression plasmids for the eight structural genes, which are under the translational control of either S/D or D243 sequences, were constructed in pBR322. The 29-bp fragment between EcoRI and HindIII of pBR322 was removed and replaced with one of the two cassettes (PI S/D gene or PI 5243 gene) shown in Fig. 3.
Oligonucleotide synthesis, purification, phosphorylation and assembly Oligonucleotides for construction of 043 sequence and for hybridization were synthesized by the solid-phase, phosphoramidite method using automated gene assemblers (Pharmacia and Applied Biosystems). After unblocking with concentrated ammonia, the oligonucleotides were purified by HPLC on a Mono Q column (Pharmacia). Oligonucleotides for hybridization were phosphorylated using [ Y - ~ ~ P ] Aand T P polynucleotide kinase to a specific activity higher than lo7 cpm/yg and purified by DEAE-cellulose (mini-column) chromatography. Oligonucleotides used for synthesis of DA3 DNA were phosphorylated, purified from non-phosphorylated forms and ligated as described in [16].
Bacterial transformation and cultivation Competent E. roli LE 392 cells were prepared and transformed as recommended by Hanahan [19]. Bacterial cells, for
~
~
L
153 quantitative determination of recombinant protein and gene specific mRNA, were prepared as follows: 20 ml LuriaBertani medium (in 100-ml flasks) containing 50 pg/ml ampicillin was inoculated and the cellular suspension was shaken (300 rpm) at 37°C for 8 h. Each cell culture was divided into two and used for analysis of both protein and mRNA. Colony and dot hybridization The method of RNA colony hybridization [20] was used for selection of clones containing genes in the correct orientation and for the quantitative determination of specific mRNA in E. coli cells harbouring the expression plasmids pP1 S/D and pP,QA3. Total cellular RNA was isolated by standard procedures from 10 ml bacterial cells. RNA was treated with DNase I (RNase free), deproteinized and precipitated with ethanol. The RNA was dissolved in distilled water which had been treated with diethyl pyrocarbonate, and the concentration of RNA was determined by A260measurement. Samples of 40 pg RNA were evaporated to dryness in a Speed Vac centrifuge, the RNA was dissolved in 20 pl300 mM NaCl and 30 mM sodium citrate, pH 7.0, containing 10% formaldehyde, heated at 60°C for 15 min and spotted onto nitrocellulose filters. The filters were baked in a vacuum oven at 80°C for 1 h and prehybridized and hybridized as described elsewhere [20]. Filters were air-dried and exposed to X-ray films. After autoradiography filters were cut and the 32P radioactivity was measured by Cerenkov counting in a Beckman scintillation counter. Determination of in vivo stability of mRNA To study the in vivo stability of mRNA, RNA synthesis was blocked by rifampicin and nalidixic acid as described [21]. 20 ml Luria-Bertani medium (100-ml flask) containing 50 pg/ ml ampicillin were inoculated with an overnight culture (1 :50 by vol.) and cultivated with vigorous shaking at 37'C for 1 h. Nalidixic acid and rifampicin were added simultaneously to final concentrations of 0.5 p,g/ml and 60 pg/ml, respectively. Samples of bacterial cells with ASq0= 2.0 were harvested at the following time intervals: 0 (zero time control: sample is taken immediately upon addition of nalidixic acid and rifampicin), 10, 20, 30, 40, 50, 60, 70, 80, 120, 180, 240, 300, 500 and 600 s. The samples were frozen at - 70 'C, thawed on ice and spun down at 4°C. Cells were resuspended in ice-cold solution containing 50 mM sodium acetate, pH 5.0, 1 mM EDTA and 0.5% SDS. Total cellular RNA was isolated, spotted onto nitrocellulose filters and hybridized with j2P-labelled oligonucleotides specific for the corresponding mRNA. The radioactivity of each spot was mcasurcd by Cerenkov counting and plotted versus the time (is). The half-life of mRNA was defined as the time corresponding to a 50% decrease in radioactivity of each spot, compared with the level of the zero-time control. Determination of plasmid copy number Plasmid copy number was measured for all the expression vectors used, as described in [22]. Determination of level of protein expression The yield of hIFcll and hIFy was measured by the antiviral activity (protective effect on Wish cells against the cytopathic
action of the vesicular stomatitis virus) of crude bacterial lysates. Cells with Asg0= 5 were sonicated in 1 ml 10 M Tris, pI1 7.4, and 1 mM EDTA and centrifuged in an Eppendorf centrifuge at 4°C for 5min. The clear lysates were serially diluted twofold and added to Wish cells grown in 96-well plates. After 4 h at 37"C, the cells were infected with vesicular stomatitis virus (multiplicity of infection lo), and after 24 h of incubation at 37°C the cells were stained with crystal violet. The titer was quantitatively monitored using a micro-ELISA reader (LKB) and the antiviral titer was expressed as the reciprocal of the dilution inhibiting 50% of the cytopathic effect. The yield of hCT was quantitated by RIA. This analysis was based on the usc of synthctic hCT as a standard and highly specific rabbit antiserum for precipitation of the antigen/antibody complex. RIA was carried out as follows. 100 p1 synthetic hCT in 50 mM phosphate buffer, pH 7.8, or E. coli lysate (prepared as above) was mixed with 200 pl antiserum (working dilution 1 :3000) and incubated at 4°C for 12 h. '251-labelled standard synthetic hCT (1000015000 cpm in 100 pl) was added and the mixture was incubated for 6 h at 4°C. The antibody-bound radioactivity was precipitated with 100 p1 anti-(rabbit Ig) serum (dilution 1 : 30) in 50 mM phosphate, pH 7.8. After 4 h incubation at 4 'C, the antigen/antibody complex was pelleted at 2500 x g for 20 min. For every sample tested, the zero and non-specific binding wcre determined. The level of expression of the PMV coat-protein gene was determined by Western blotting. Bacteria containing the PMV coat-protein gene were grown overnight in 20 ml LuriaBertani medium containing 1% dextrose at 37 "C. Bacteria at A260= 1 were sedimented and resuspended by sonication in 0.1 M Na2C03/0.1M dithiothreitol. The suspension was adjusted to 1.7% SDS and 10% sucrose, boiled and electrophoresed in SDSiPACJE [23]. Following electrophoresis, the proteins were transferred onto nitrocellulose membranes using a Transblot (Bio-Rad), according to suppliers' instructions. Rabbit polyclonal antisera were used to identify the PMVi coat protein according to Bailly and Coleman [24].
RESULTS AXD DISCUSSION The untranslated 5' leader sequence of TMV RNA is described in the literature as a universal enhancer of translation of foreign gene transcripts in vivo and in vitro for both prokaryotes and eukaryotes [l -61. In all these studies, only prokaryotes genes (chloramphenicol acetyltransferase, pglucuronidase and neomycin phosphotransferase) have been used as a model for quantitative estimation of translation enhancement. The mRNA of these genes are translated efficiently in both prokaryotic and eukaryotic cells and cell-free systems. Low levels of expression of these genes (due to selfinitiation of translation) were found, even in the absence of specific translation initiators. However, when the 52 sequence is added in front of these genes, translation activity is enhanced. We have investigated the efficiency of 52 sequence to initiate translation of eukaryotic mRNA in E. coli cells and checked the functional equivalence [6] of L? sequence by comparing its efficiency with that of the consensus S/D sequence. Eight different eukaryotic genes were inserted into two types of expression vectors (pPIS/D and ~P~S2.43) and used in this study. As shown in Fig. 3, both vectors differ only by the nature of the translation-initiation site. In the plasmid
154 Table 1. Effect of Q43 on the expression of eukaryotic proteins in E. cnli LE392. Yields of hIFcll and hIFy were calculated from the spccific antiviral activities of the protein? (10 IJ/ng for hIFcll and 50 U/ng for the hTFy). Yields of hCT and hIFy-hCT were determined by RIA. For all genes used, the removal of S/D or QA3 led to undetectable amounts of protein (data not shown). Yield of PMV coat protein is presented as arbitrary units based on the band intensity
in immunoblots. Gene
Yield of recombinant protein SiD
5243
mgll
hlFcr, (L ' j hIFml (L-) hlb'iccl (BIL-)
hIFy (Cys* j hIFy (Cys-)
hIFy-hCT hCT, PMV coat protein
2.70 37.00 22.00 10.00 285.00 0.11 2.55 1.00
Decrease
-fold
0.61 3.4Y
3.60 0.12 0.15 < 0.01 0.05 0.3
4.4 10.6 6.1 83.9 1900.0 > 10 51 .o
3.3
pPIS/D, a strong synthetic ribosome-binding site (containing the consensus AAGGAGGT sequence) was used as a translation initiator, whereas in pP,QA3, the translation initiator was a synthetic derivative of R (Qd3) known to be much more effective in E. coli than the native Q sequence [5]. The coding sequences used in our study were either natural [(hIFal (L+)], isolated from a gene library, semi-synthetic [(hlFa, (L-), hlFal(BILp) and PMV coat-protein gene] or fully synthetic genes [hlFy(Cys+),hIFy(CysC), hIFy-hCT and hCT,]. They were suitable for the present study for the following reasons. Their mRNA were unable to initiate translation in E. coli cells in the absence of translation-initiation sequences (unpublished data). The efficiency of expression of these genes varied [from 30-50% of the total cellular protein for hIFy(Cys-) to negligible amounts for hIFy-hCT and PMV coat protein], and they could be classified as follows: highly expressable [hIFy(Cys-), hIFy(Cys+), hCTt, hIFal(L-) and hIFr,(BILp)]; expressable with moderate efficiency [hIFa, (L')]; expressable with low efficiency [hIFy-hCT and PMV coat-protein gene]. The same design of the 5' ends of these genes ensured an identical nucleotide context in the corresponding mRNA between the + 1 nucleotide and the start codon AUG. The presence of three groups of related genes allows comparison of the significance of codon use downstream of AUG for the efficiency of expression in both types of vectors pPIS/D and pP152d3. The sensitive methods used for quantitative analysis of the yield of protein make it possible to detect small differences in the level of gene expression and to derive quantitative values for the enhancing or attenuation effect of Q in comparison to the S/D sequence. Level of protein expression in E. coli LE 392 cells transformed with expression plasmids pPIS/D and pP1QA3 The level of gene expression (as protein) was determined for all eight genes under the translational control of the S/D (consensus) or 5243 sequence, and the results are presented in Table 1. The yield of recombinant protein obtained is lower for the expression plasmids containing 5243 sequence. These data show that the level of expression of the eight genes varies over broad range for both series of expression vectors (PIS/D
gene and P1Q43 gene), although their 5' ends are all identical (Fig. 1). However, the variation in yield for different genes is less significant (e.g. 360 times between hIFy-hCT and hIFal(BIL-) for the constructs containing QA3 sequence compared to those with S/D sequence (Table 1).The initiation of translation by S/D, unlike that of QA3, is more sensitive to changes in the 5' coding region of the gene. For example, the deletion of only three codons (Cys-Tyr-Cys) of the signal peptide of the hIFy gene resulted in a %-fold increase in the yield of protein under the translation control of S/D sequence, and only 1.2-fold increase in the presence of 5243 sequence (Fig. 1, Table 1). Consequently, a decrease of 1900-fold resulted in the level of expression of hlFy(Cys-) gene between the S/D and 5243 as translation initiators (Table 1). Furthermore, the level and half-life of mRNA is also decreased in the presence of QA3 sequence (see below). All eight genes used in this study gave no detectable protein when the expression vectors did not contain translation initiators (data not shown). Our data (Table 1) clearly showed that independently of the lower yield obtained from the vectors pP1QA3, the TMV leader sequence is a real initiator of translation of mRNA in E. coli. These results confirm the conclusions of others [l - 61. The lower level of protein expressed from vectors containing 9 4 3 sequence may be explained by lower plasmid copy number, lower level of mRNA, shorter half-life of mRNA or lower translation-initiation efficiency. Plasmid copy number was determined for both series of expression plasmids pPIS/ D and pP1QA3 and the results are presented in Table 2. Substitution of 5243 for S/D sequence did not change detectably the plasmid copy number values. Surprisingly, the expression of the PMV coat-protein gene lowered the plasmid copy number by about 2.5-fold. It has been shown that below pH 7.5,PMV coat protein binds non-specifically to any RNA in vitro [25]. We can speculate that the effect of PMV coat protein on plasmid replication may be due to a non-specific interaction with the primer RNA (RNA I). Level and half-life of mRNA transcribed in vivo from the expression plasmids pPIS/D and pP15243 The P1 promoter in plasmids pPIS/D and pP10d3 initiates transcription of dicistronic mRNA (due to the absence of a transcription terminator after the eukaryotic gene). The 5' end of these mRNA are specific for the inserted gene, and the 3' termini represent transcripts of the tetracycline-transacetylase gene of pBR322 which is common for all the expression vectors. This allows the determination of mRNA contents for all the constructions by comparing the hybridization of equal amounts of total cellular RNA with the same hybridization probe specific for the tetracycline-transacetylase gene. The results presented in Table 2 show that the substitution of 5243 for S/D in the expression vcctor dccrcascd the level of mRNA in the cell for all the genes studied. The amounts of mRNA (32P radioactivity) in cells correlated with the plasmid copy number (e.g. the mRNA of PMV coat protein compared to other genes is about three times less; Table 2). The effect of QA3 in this case was not as dramatic as for the protein (Table 1) and did not exceed a 7.1-fold decrease in the content of mRNA. The half-life of mRNA transcribed from plasmids pPIS/ D and pPlQA3 was studied after double blocking of transcription by rifampicin and nalidixic acid (for details, see Materials and Methods). The half-life of mRNA produced from plasmid pP152A3 varied over 60-70 s, whereas that of the mRNA
155 Table 2. Plasmid copy number, stability and content of mRNA in E. coli LE 392 expressing eukaryotic genes under the control of PI S/D or PI 043. Decrease in mRNA was assessed using a common tetracyclinetransacctylase mRNA, 32P-labelledspecific probe. Data represent the ratio between the levels of mKNA containing S/D and 5243. Gene
Plasmid copy number S/D
5243
Half-life of mRNAQd3 S/D
23 24 21 20 21 23 22 8
22 23 22 21 21 20 21 9
70 70 70 80 90 70 70 70
1341 1360 CGCUAGUAAUCGUGGAUCAG AUCAUUt@CAUU
5‘
2.
5’ 3’
1341 1360 UCGCUAGUAAUCGCAGGUCA AUCAUUAACAUUC
3’ 5’
3.
5’ 3’
511 530 GGCCCUGUAAUUGGAAUGAG CAUUAAWUU$C
3’ 5’
4.
5’ 3’
140 163 GGAUAACCGUAGUAAUUCUAGAGC UAUCAUUAAGAUU
-fold 65 70 70 65 60 65 60 70
3’ 5’
5
Decrease in mRNA
843
S hIFal (L’) hIFEl (L-) hIFa, (B,L-) hIFy (Cys’) hIFy (Cys-) hIFy-hCT hCT, PMV coat proteiii
3’
1.
2.8 1.5 2.4 3.2 7.1 1.9 3.5 1.4
containing the S/D sequence was 70-90 s (Table 2). For most of the genes studied, the destabilizing effect of 5243 was negligible ( 5 - 10 s only) with the exception of the mRNA f o r hIFy(Cys ’) and hIFy(Cys-), where the destabilking effect is in the order 15 - 30 s (Table 2). In general, the lower level of mRNA in E. coli cells containing expression plasmids pP104 3 cannot be explained by destabilization (increased RNase sensitivity) of mRNA caused by 52243. The results presented in this study clearly demonstrate that the substitution of 5243 for S/D in the expression vector decreases both the yield of recombinant protein and the content of mRNA. The lower level of mRNA cannot be explained either by the reduced copy number of expression plasmids or by the decreased stability (half-life) of mRNA. Consequently, we conclude that the presence of 5243 in the plasmid has an attenuating effect on transcription. Furthermore, our results show that there is no direct correlation between the extent of decrease in the level of mRNA and the yield of protein (compare the data in Tables 1 and 2). From the result that protein yield is more sensitive to the substitution of Q43 for the S/D sequence than the mKNA we that Q243is less efficient as a translation-initiation signal in vivo than the S/D sequence. This can explain the discrepancy between the shorter half-life of mRNA transcribed from the expression vectors pPl52A3-hIFy(Cys+) and pPlf2243-hIFy(Cys-) (Table 2). 0 sequence contains a domain complementary to rRNA of prokaryotes and eukaryotes
The mechanism of initiation of translation of mRNA in E. coli by the TMV i2 sequence is still unknown [6]. The 52 sequence (and its derivative 5143) does not contain any G residues, and cannot therefore interact with the 3‘ end of the small ribosomal subunit rRNA. To find an alternative site of interaction of i2 and Q A 3 sequences with 16s rRNA of E. coli a computer search for complementarity was carried out. As seen in Fig. 4, a 12-nucleotide sequence (with one mismatch) was found to be complementary to 16s rRNA at positions 1344- 1355. The corresponding complementary domain in 52 comprises the hexanucleotide AAUUAC, occurring three times in both the native 52 and 0243 sequences [S]. This repeat has been shown to be important for the efficient initiation of translation of mRNA by Q in both prokaryotes and
3’
Fig. 4. Complementarity between TMV s2 sequence (3’ - 5’) and (5’3’): (1) E. coli 16s rRNA [26]; (2) E. corrodens 16s rRNA (271; (3) X . laevzs 185 rRNA [28]; (4) Oryza sativa 17s rRNA [29]. Mismatched bases are shaded.
eukaryotes (tobacco protoplasts and X . laevis oocytes) [5, 61. A search for complementarity was carried out with other rRNA. Fig. 4 shows that the same nucleotide sequence was identified at the same location in the 16s RNA of Eikenella corrodens. In both organisms (E. coli and E. corrodens), this sequence was about 150 nucleotides from the 3‘ end of 16s rRNA. The natural site of interaction of the ribosomes with S/D sequence has been shown to be at the 3‘ end of 16s rRNA [7]. According to the three-dimensional model of the E. coli 30s ribosomal subunit [30], the 16s rRNA region complementary to the TMV Q sequence (positions 1344- 1355) is situated in a domain close to that where the S/D sequence interacts with the 30s ribosomal subunit. These two domains form a groove where the mRNA is presumably interacting with the ribosomal subunit. This result strengthens the role of Q as an alternative binding site for mRNA. Q sequence was also found to be complementary to a region near the 5’ ends of the 18s rRNA of the small ribosomal subunit of X . fuevis [l -41 and to the 17s rRNA of Orjiza sativa. Since the site of interaction ofmRNA with eukaryotic ribosomes is not known, our data suggest that the mRNA-binding site may be localized in the first third of the eukaryotic small ribosomal subunit rRNA (at least for mRNA containing Q-like translation initiators). This work was partly funded by grants from the Natural Sciences and Engineering Research Council Canada and Ontario Ministry of Food and ~ ~to M. G , AbouHaidar ~ i and by a~ grant 2681~ 06.05.86 from the Ministry of Industry and Technology, Bulgaria, to I. Ivanov.
REFERENCES 1. Sleat, D. E., Gallie, D. R., Jefferson, R. A., Bevan, M. W., Turner, P. C. & Wilson, T. M. A. (1987) Characterization of the 5’-
leader scquence of tobacco mosaic virus RNA as a general enhancer of translation in vitro, Gene (Amst.) 66, 217-225. 2. Gallie, D. R., Sleat, D. E., Watts, J. W., Turner, P. C. & Wilson, T. M. A. (1987) The 5’-leader sequence of tobacco mosaic virus RNA enhances the expression or forcign gene transcripts in vitro and in vivo, Nucleic Acids Res. 15, 3257 - 3272. 3. Gallie; D. R., Walbot, V. & Hershey, J. W. (1988) The ribosomal fraction mediates the translation enhancement associated with the 5’-leader of tobacco mosaic virus, Nudeic Acids Res. 17, 8675 - 8695. 4. Sleat, D. E., Hull, R., Turner, P. C. &Wilson, T. M. A. (1988) Studies on the mechanism of translational enhancement by the 5’4eader sequence of tobacco mosaic virus RNA, Eur. J . Biochem. 175, 75 - 86.
l
156 5. Gallie, D. R., Sleat, D. E., Watts, J. W., Turner, P. C. &Wilson; T. M. A. (1988) Mutational analysis of the tobacco mosaic virus 5’-leader for altered ability to enhance translation, Nucleic Acids Res. 16, 883 - 893. 6. Gallie, D. R. & Kado, C. I. (1989) A translational enhancer derived from tobacco mosaic virus is functionally equivalent to a Shine-Dalgarno sequence, Proc. Nut1 Acad. Sci. USA 86, 129- 132. 7. Shine, J. & Dalgdrno. L. (1975) Dctermination of cistron specificity in bacterial ribosomcs, Nature 254, 34-38. 8. Gold, L., Pribnow, D., Schneider, T., Shinedling, S., Singer, B. & Stormo, G. (1981) Translational initiation in prokaryotes, Annu. Rev. Microbiol. 35, 365 -403. 9. Koxak, M. (1983) Comparison of initiation of protein synthesis in procaryotes, eucaryotes and organelles, Microbiul. Rev. 47, 1-45. 10. Kozak, M. (1989) The scanning model for translation: an update, J . Cell Biol. 108, 229 - 241. 11. Bachvarov, D.. Wishart, P., Usheva, A., Ivanov, I. & Jay, E. (1990) Expression in E. coli of a human Va18-calcitonin gene by fusion to the chloramphenicol acetyl transferase (CAY) gene, Biol. Zentralbl. 109, 113- 117. 12. Nagarta, S., Mantei, N. & Weissman, C. (1980) The slructure of one of the eight or more distinct chromosomal genes for human interferon a, Nature 287, 401 --408. 13. Ivanov, I.. Gigova, L. & Jay, E. (1987) Chemical synthesis and exprcssion in E. coli of a human Vals-calcitonin gene by fusion to a synthetic human interferon-y gene, FEBS Lett. 210, 5660. 14. Ivanov, I., Markova, N., Bachvarov, D., Alexciev, K., Sarafova, A., Maximova, V., Tsaneva, I. & Markov, G. (1989) Constitutive expression of a native human interferon-a, gene in E. coli, int. J . Biochem. 21.983-985. 15. Jay, E., Rommcns, J., Cloncy, L., McKnight, D., Lutze-Wallace, C., Wishart, P., Lin, W., Asundi, V., Dowood, M. & Jay, F. (1984) High-level of expression of a chemically synthesized gene for human interferon-y using a prokaryotic expression vector, Proc. Natl Acad. Sci. USA 81. 2290 - 2294. 16. Ivanov, I . (1990) Shotgun concatenation of synthetic genes: construction of concatemeric human calcitonin genes, Anal. Biochern. 189,213-216. 17. Sit, T. L., AbouHaidar, M. G. & Holy, S. (1989) Nucleotide sequence of papaya mosaic virus RNA, J . tien. Virol. 70, 2325 -2331. 18. Rommens, J., MacKnight, D., Pomeroy-Claney, L. & Jay, E. (1 983) Gene expression: chemical synthesis and molecular
19. 20. 21. 22. 23. 24.
25.
26.
27.
28. 29. 30.
cloning of a bacteriophage T5 (P25) early promoter, Nucluic Acids Res. 11, 5921- 5940. Hanahan, D. (1983) Studies on transformation of E. coli with plasmids, J. Mol. Biol. 166, 557-580. Ivanov, I. & Gigova, L. (1986) RNA colony hybridization method, Gene (Amst.) 46,287 - 290. Silengo, L., Nicolaev, N., Schlessinger, D. & Imamoto, 1. (1974) Stabilizalion of mRNA with polar effects in an E. coli mutant, Mol. tien. tienet. 134, 1 - 19. Ivanov, I. & Bachvarov, D. (1987) Determination of plasmid copy number by the boiling method, Anal. Biochem. 165, 137-141. Laemmli, U. K. (1970) Cleavage of structural proteins during the assembly of the hcad of bactcriophages T, Nature 227, 680685. Bailly, J. & Coleman, J. R. (1988) Effect of C 0 2 concentration on protcin biosynthesis and carbonic anhydrase expression in Chlamidornonus reinhardtii, Plant Physiol. (Bethesda) 87,833 840. AbouHaidar, M. G . & Erickson, J. W. (1985) Structure and in vitro assembly of papaya mosaic virus, in Molecular plant virology (Davies, J., ed.) pp. 85-121, CRC Press Inc., Boca Raton, Florida. Schnare, M. N. & Gray, M. W. (1982) 3’-terminal sequence of wheat mitochondria1 18s ribosomal RNA: furlher evidence of a eubacterial evolutionary origin, Nucleic Acids Res. 10,3921 3932. Dewhirst; F. E., Paster, €3. J. & Bright, P. L. (1989) Chromobacteriim, Eikenella, Klingella, Neisseriu, Simonesiellu and Vitreoscilla species comprise a major branch of the /Igroup Proteobacteria by 16s ribosomal ribonucleic acid sequence comparison : transfer of Eikenella and Simonsiella l o the family Neisseriaceae (emend.), Int. J . Syst. Bacteriol. 39, 258 - 266. Maden, B. E. H. (1986) Idcntification of the locations of the methyl groups in 18s rihosomal RNA from Xenupus laevis and man, J . Mol. Biol. 189, 681 -699. Takaiwa, F.. Oono, K. & Suqiura. M. (1984) The complete nucleotide sequence of a rice 17s rRXA gene, Nucleic Acid.7 Res. 12, 5441 - 5448. Brimacombe, R., Atmadja, J., Stiege, W. & Schbler. D. (1988) A detailed model of the three-dimensional structure of Escherichia coli 16s ribosomal RNA in situ in the 30s subunit, J. Mol. Biol. 199,115--136.
