ABSTRACT
Unmodified tRNA molecules are useful for many purposes in cell-free protein biosynthesis, but there is little information about how the
lack of tRNA post-transcriptional modifications affects the coding specificity for
synonymous codons. In the present study, we prepared an unmodified form of
Escherichia coli
{t R N A} sub 1 sup {S e r}
, which originally has the cmo
5
UGA anticodon (cmo
5
U = uridine 5-oxyacetic acid) and recognizes the UCU, UCA and UCG codons. The codon
specificity of the unmodified tRNA was tested in a cell-free protein synthesis directed by designed mRNAs under competition
conditions with the parent
{t R N A} sub 1 sup {S e r}. It was found that the unmodified tRNA with the UGA anticodon recognizes the UCA codon nearly as efficiently as the modified tRNA. The unmodified tRNA recognized the UCU codon with low, but
detectable efficiency, whereas no recognition of the UCC and UCG codons was
detected. Therefore, the absence of modifications makes this tRNA more specific
to the UCA codon by remarkably reducing the efficiencies of wobble reading of
other synonymous codons, without a significant decrease in the UCA reading
efficiency.
Cell-free protein biosynthesis is becoming more useful in applications for
biochemical and molecular biological studies (
1
-
3
). This method is free of many of the limitations associated with
in vivo
protein production systems. For example, engineered tRNA molecules can be added
to the cell-free protein synthesis system. Proteins with an unnatural amino acid
residue at a specific position can be synthesized by the use of an engineered
suppressor tRNA molecule charged with the unnatural amino acid through chemical
aminoacylation (
2
,
4
).
In general, however, it is not easy to purify and engineer large amounts of
naturally occurring tRNA molecules. In contrast,
in vitro
transcription with T7 RNA polymerase (
5
) is a powerful method for the preparation and engineering of tRNA molecules. By
this method, a large amount of homogeneous tRNA of the desired sequence can
easily be obtained. The tRNA molecules generated by this method are not
natural, as they lack post-transcriptional modifications, which are considered to have important
roles in codon reading (
6
,
7
). Nevertheless, unmodified tRNAs have been shown to insert natural (
8
-
10
) and unnatural (
11
,
12
) amino acids into polypeptide chains.
Comparison of the codon reading specificity between the unmodified and naturally
modified forms have been done for
Escherichia coli
tRNA
Phe
(
9
) and
Mycoplasma mycoides
tRNA
Gly
(
8
). These studies indicated that the modifications had minimal effects on the
specificity concerning the third codon bases. Note that the natural
E.coli
tRNA
Phe
and
M.mycoides
tRNA
Gly
have no modified nucleoside in the anticodon. In contrast, many other tRNAs
have a modified nucleoside at the first position of the anticodon (position 34)
(
13
), which significantly contributes to codon recognition (
6
). Therefore, for these anticodon-modified tRNAs, it is interesting to examine the codon reading
specificities of the unmodified forms.
Uridines substituted at position 5 by an oxygen atom (xo
5
U) have been identified at position 34 of many tRNAs that correspond to a four-codon box (a set of four codons with the first two bases in common, which
specify the same amino acid) (
14
-
16
). Such tRNAs recognize the codons ending with A, G and U. For example,
E.coli
{t R N A} sub 1 sup {S e r} has uridine 5-oxyacetic acid (cmo
5
U), and the cmo
5
UGA anticodon reads the UCU, UCA and UCG triplets on ribosomes (
17
,
18
). On the other hand, in mitochondria and in
Mycoplasmas
, a single tRNA species with an unmodified uridine at the first position of the
anticodon recognizes all four cognate codons (
19
-
23
). Thus, it has been argued that the xo
5
U-type modifications restrict the reading of codons ending with C (
24
). In contrast, it has also been proposed that the xo
5
U-type modifications are responsible for increasing the reading efficiency
for U, but not for C, to levels nearly as high as those for A and G (
16
).
In the present study, we investigated the activity and specificity of an
unmodified form of
E.coli
{t R N A} sub 1 sup {S e r} in the reading of codons UCN (N = U, C, A and G). Direct comparison between the
modified and unmodified molecules revealed the codon reading properties of the
unmodified tRNA molecules, and the roles of the modifications.
[
14
C]Serine (5.98 GBq/mmol) was purchased from New England Nuclear. [
3
H]Serine (1.07 TBq/mmol) was from Amersham, and was concentrated with a
centrifugal concentrator to 8.0 mCi/ml (295 GBq/ml).
The preparation of the S30 extract was as described (
25
), except that an MSK cell homogenizer (B. Braun) was used for lysis of the
cells. The glass beads were removed by centrifugation at 2300
g
immediately after lysis.
The mRNAs were designed and prepared on the basis of the method used in analyses
of codon assignment in
Mycoplasma capricolum
(
26
,
27
). A 156-bp fragment (capital letters in Fig.
1
) was chemically synthesized, amplified by PCR, and substituted for a 0.9 kb
Nde
I-
Hin
dIII fragment (the major part of the coding region of T7 gene
10
) of pGEMEX-1 (Promega). Then, the whole DNA fragment shown in Figure
1
was prepared by
Sma
I and
Bgl
I digestion, and substituted for a 0.3 kb
Pvu
II-
Bgl
I region of pUC118 (pART23GGC3TCT). The coding region has a single test codon,
TCT for Ser (shadowed in Fig.
1
), in the middle of 48 non-Ser codons; the amino terminal 11 amino acid residues are derived from
those of the gene
10
product except that the third codon, AGC, for Ser has been mutated to GGC for
Gly. Four variants in terms of the test codon were also prepared by PCR
mutagenesis of TCT to TCC, TCA, TCG and GAG (pART23GGC3TCC, pART23GGC3TCA,
pART23GGC3TCG and pART23GGC3GAG respectively).
The sequences of the unmodified tRNAs were constructed on the M13mp18 vector
(T7Ser1TGA, Fig.
2
). The single-stranded DNA of T7Ser1TGA was used as the template for a polymerase chain
reaction with the primers 5'-ACGACGTTGTAAAACGACGGCCAG-3' and 5'-ATTTCACACAGGAAACAGCTATGAC-3' both of which hybridize
outside the multi-cloning site. The product was digested with the restriction enzyme
Bgl
I, which produces the proper CCA end of the template, and was purified with a
standard polyacrylamide gel. This DNA was used for the transcription reaction
containing 20 mM GMP, in addition to the components in the mRNA preparation
described above. After transcription, the sample was purified with 15%
polyacrylamide gel containing 8 M urea.
The {t R N A} sub 1 sup {S e r} from
E.coli
A-19 (
29
) was prepared essentially as described (
18
). A modified nucleoside, 2'-
O
-methylcytidine, occurs at position 32 of tRNA
Ser
1
from
E.coli
B (
30
). In our sample, this position is only partially modified, or is nearly
completely unmodified, as judged from the facts that it was cleaved efficiently
between positions 32 and 33 with RNase A, and that the 3'-terminal nucleoside of the resulting 5'-oligonucleotide was an unmodified C (data not shown).
The DNA sequence containing the
serS
gene (
31
) was amplified by PCR from the chromosomal DNA of
E.coli
strain A-19, using the phosphorylated primers 5'-pTGACGTGCCGAATTCATTTGCGTAATG-3' and 5'-pCTCACGATTGAATTCCAGTAACAAA-3' (each with an
Eco
RI recognition site). The resulting DNA fragment was directly inserted into the
plasmid pBR322. The
Eco
RI fragment, which was considered to encompass the
serS
gene, was obtained from one clone and was transferred into plasmid pUC119
(pUCSRS).
Escherichia coli
MV1184 harboring pUCSRS was cultured overnight in 2* TY medium (4 l) containing 150 [mu]g/ml ampicillin. The cells were lysed and the lysate was separated on
a column (2.5 * 25 cm) of DE52 (Whatman) with a linear gradient of ammonium chloride concentration from 50 to 300
mM in buffer A [50 mM Tris-HCl, pH 7.9, 10 mM magnesium acetate, 5 mM 2-mercaptoethanol and 0.1 mM phenylmethylsulfonylfluoride (PMSF)].
The fractions with high activity were pooled. Solid ammonium sulfate was added
to 60% saturation, and the precipitate was dialyzed against buffer A without
PMSF. Glycerol was added to a final concentration of 50%. This crude SerRS (~2 mg protein/ml or 0.1-0.2 mg/ml SerRS) was stored at -20oC and was used for the preparation of Ser-tRNA.
We prepared [
14
C]Ser-tRNA(modified), [
14
C]Ser-tRNA(unmodified), [
3
H]Ser-tRNA(modified) and [
3
H]Ser-tRNA(unmodified) as follows.
For the preparation of Ser-tRNA, tRNA (2 A
260
U/ml or 3.4 [mu]M) was incubated at 37oC for 15-25 min with 50 mM HEPES-KOH, pH 7.2, 16 mM MgCl
2
, 2 mM ATP, potassium salt, 10 mM KCl, 1/10 vol of the crude SerRS described
above and 16-36 [mu]M
3
H- or
14
C-labeled serine. The reaction was stopped by the addition of 1/4 vol of 1 M
potassium acetate, pH 4.5. The Ser-tRNA was extracted with phenol, and was precipitated with ethanol. The
precipitate dissolved in aminoacyl-tRNA buffer (10 mM acetic acid, 4 mM magnesium acetate, pH 4.5). An
aliquot was used to measure the radioactivity using ReadyCaptm (Beckman). It should be noted that Tris inhibits SerRS, although it has
been used in many published experiments.
The nonradioactive aminoacyl-tRNA mixture (cold aminoacyl-tRNA
Mix
) was prepared by incubating 2 mg/ml of the tRNA crude fraction from
E.coli
with a mixture of 0.55 mM threonine, alanine, glycine, tyrosine, methionine,
glutamine and lysine hydrochloride and 1.1 mM valine, in the presence of 1/10
vol of the S100 extract (
32
), in the same manner as in the serylation reaction described above. The cold
aminoacyl-tRNA
Mix
is not essential for the protein synthesis reaction, but it shortens the
initial lag period of the reaction.
The method utilized in the present study is a modification of that for the
relative codon-reading efficiency analysis (
33
) according to the translational error analysis method (
34
). For a given test mRNA, cell-free protein synthesis reactions were performed with [
14
C]Ser-tRNA(modified) and [
3
H]Ser-tRNA(unmodified) and also with [
14
C]Ser-tRNA(unmodified) and [
3
H]Ser-tRNA(modified). During translation of the mRNA, the test codon, UCU, UCC,
UCA, UCG or GAG, was recognized, in competition, by the Ser-tRNAs, for example, [
14
C]Ser-tRNA(modified), [
3
H]Ser-tRNA(unmodified) and/or other intrinsic, nonradioactive aminoacyl-tRNAs. We added rather large amounts of the radioactive Ser-tRNAs so that they could recognize codons in competition with
nonradioactive Ser-tRNAs present in the reaction. In order to prevent transfer of the
radioactive serine from the preaminoacylated Ser-tRNAs to other tRNAs, we used a large amount of nonradioactive serine in
the reaction (see below).
Four micrograms of cold aminoacyl-tRNA
Mix
, 1.9 * 10
4
c.p.m. (~60 pmol) of [
14
C]Ser-tRNA(modified or unmodified), 8.5 * 10
5
c.p.m. (~60 pmol) of [
3
H]Ser-tRNA(unmodified or modified) and 0.6 A
260
unit (~0.16 nmol) of one of the five test mRNAs were first mixed in 0.2 M
potassium acetate, pH 4.5, and precipitated with ethanol. The precipitate was
rinsed with ethanol, dried
in vacuo
, and dissolved in 20 [mu]l of the aminoacyl-tRNA buffer (RNA mixture). The protein synthesis reaction (240 [mu]l) was incubated at 37oC and contained 55 mM HEPES-KOH, pH 7.5, 1.7 mM DTT, 275 [mu]M GTP, 26 mM phosphoenolpyruvate, potassium salt
(Boehringer Mannheim), 1.2 mM ATP, 1.9% (w/v) polyethyleneglycol 8000 (Sigma),
34 [mu]g/ml folinic acid, calcium salt (Sigma), 6.9 mM ammonium acetate, 1 mM
spermidine, 7.5 mM magnesium acetate, 210 mM potassium glutamate (Sigma), 0.74
mM valine, 1.1 mM serine, 0.37 mM each of threonine, alanine, glycine,
tyrosine, methionine, glutamine and lysine hydrochloride, 1/6 vol of S30
extract, and the RNA mixture. The reaction was started by mixing the RNA
mixture with the other reaction components.
Aliquots (50 [mu]l) were transferred into 50 [mu]l of 0.2 N NaOH at 0, 4, 8 and 12 min, and were incubated at 37oC for 30-60 min. Aliquots (10 [mu]l) were transferred to 90 [mu]l of chilled 5% TCA at 2, 6 and 10 min. These
samples (100 [mu]l) were placed on the Whatman 3MM filter disks, and were washed three times
with ice-cold 5% TCA. The retained radioactivity on the filter disk was counted
using a standard liquid scintillator with an LSC-700 liquid scintillation counter (Aloka, Japan). The
14
C and
3
H radioactivities incorporated into the polypeptide chain were measured
separately in c.p.m. The amounts of the surviving, labeled Ser-tRNA at (2
n
+ 2) min (
n
= 0, 1, 2) also in c.p.m. were estimated as those of the alkali-sensitive, TCA-insoluble matter by subtracting the average of the amounts of the
alkali-resistant, TCA-insoluble matter at 2
n
and (2
n
+ 4) min from the amount of the total TCA-insoluble matter at (2
n
+ 2) min. Codon reading efficiencies of the unmodified tRNA relative to the
modified tRNA were calculated as described in detail in the Results and
Discussion. Note that the calculation does not require the values of the
counting efficiencies for
14
C and
3
H, because they were each canceled out upon comparison between modified and
unmodified tRNAs.
First, we prepared [
14
C]Ser-tRNA(modified), [
14
C]Ser-tRNA(unmodified), [
3
H]Ser-tRNA(modified) and [
3
H]Ser-tRNA(unmodified). The unmodified tRNA was serylated efficiently with the
crude SerRS. The recovery of Ser-tRNA was ~60% for the unmodified tRNA and ~90% for the modified tRNA. As
E.coli
SerRS recognizes characteristic secondary and tertiary structures of serine
tRNAs during aminoacylation (
35
), we concluded that the unmodified tRNA was properly folded into such
characteristic structures.
We then measured the incorporation of labeled serine into the polypeptide during
a protein synthesis reaction containing mRNA(UCA) in the presence of the [
14
C]Ser-tRNA(modified) and [
3
H]Ser-tRNA(unmodified). The recognition of codon UCA by the tRNAs is through
three Watson-Crick-type base pairs. As the synthesized polypeptide contained many
valine residues, it was easily precipitated in 5% TCA. As shown in Figure
3
, [
14
C]serine was efficiently incorporated from the [
14
C]Ser-tRNA(modified) into the acid insoluble fraction (Fig.
3
A, [squf]). As a control, we performed, in parallel, a reaction containing
mRNA(GAG), which does not contain any serine codons; the incorporation was
expectedly inefficient (Fig.
3
A and C, -).
The results indicated that the Watson-Crick type recognition of the A of the UCA codon was only slightly
affected by the lack of modifications in the present case of {t R N A} sub 1 sup {S e r}. Two other unmodified tRNA molecules have been compared with their parent tRNAs
for their codon reading activities (
8
,
9
). As compared with the modified tRNA
Phe
, the unmodified tRNA
Phe
was weaker (~70%) in the formation of a ternary complex with elongation factor Tu, but
was more efficient (~260%) in the dipeptide formation rate per ternary complex (
9
). An unmodified form of
M.mycoides
tRNA
Gly
is as active as the naturally modified form (
8
). All these results indicate that the unmodified tRNA is, in general, nearly as
efficient as the modified tRNA.
Actually, {t R N A} sub 1 sup {S e r} is different from tRNA
Phe
and tRNA
Gly
, in that it is modified at position 34, while the other tRNAs are not. However,
the modification does not change the arrangement of the hydrogen donor-acceptor pattern at positions 2, 3 and 4 of the pyrimidine ring. Thus, it
is not likely that it dramatically changes the efficiency of the formation of
the Watson-Crick base pair with A. The effects of the xo
5
U modification are being analyzed in our laboratories by using two tRNA
molecules that are different only in the modification at position 34.
The fact that the unmodified tRNA had a comparable codon-reading activity with the modified tRNA indicates that the tRNA was
properly recognized by EF-Tu-GTP and ribosome. The serylation reaction was also efficient as
described above. To be recognized both by EF-Tu and SerRS, the tRNA should be folded into the characteristic secondary
and tertiary structure of tRNA molecules (
35
,
36
). Thus, the unmodified tRNA that we prepared here appears to have a
conformation required for the proper recognition and discrimination of codons
on ribosomes.
The U-U base pair, in which the N3s of both U residues serve as the proton
donors, and the O4 of one and the O2 of the other accept the protons, was
considered in the wobble hypothesis, but it was concluded that this type of U-U base pair was unlikely to be formed on ribosomes, because it would cause
misreading of the genetic information (
37
). Afterwards, however, we indicated that the U-U base pair is stably formed when the ribose moiety of the uridine of the
anticodon takes the C2'-endo conformation (
16
). In addition to this Crick-type wobble U-U base pair, a water-bridged base pair was also proposed (
38
). In any case, the detectable, but weak, reading of the UCU codon by the
unmodified UGA anticodon may be ascribed to some unusual structural features of
the U-U base pair.
As for the modified tRNA, it was clearly shown that the cmo
5
U-U pair is much more efficiently formed than the U-U pair (`M-N' means hereafter that the first nucleoside of the anticodon
is M and the third nucleoside of the codon is N). This efficient base pairing
can be explained by the characteristic conformational properties of cmo
5
U, which prefers the C2'-endo form to the C3'-endo form (
16
), but not by a formation of a water-bridged base pair (
38
).
To our surprise, the unmodified tRNA only slightly reads the UCG codon; the G at
the third position of the codon cannot be recognized efficiently by the
unmodified U at the first position of the anticodon. This is probably because
the U of the putative U-G base pair cannot stack well onto its 3' neighbor (39). Such a loss of stacking may also be the case for
cmo
5
U, at the first position of the anticodon. On the other hand, this modified
uridine has another peculiar feature, in that the C2'-endo form is unusually stable as compared with the C3'-endo form. In this context, our model building study
has revealed that the formation of the cmo
5
U-G pair is possible, without any steric hindrance, with both the C2'-endo and C3'-endo conformations of the cmo
5
U (16). Therefore, the present results point out the possibility that the cmo
5
U(C2'-endo) -G pair is very stable.
As in the ribosome binding experiment (
18
), the naturally-modified {t R N A} sub 1 sup {S e r} exhibited no detectable recognition of the UCC codon in the present model
mRNA translation. Furthermore, the unmodified form of {t R N A} sub 1 sup {S e r} was found to have very little, if any, activity to read the UCC codon in the
present system. The UCC codon was properly read by {t R N A} sub 2 sup {S e r} with the GGA anticodon (data not shown). Therefore, it is concluded that
neither the unmodified U nor the cmo
5
U recognizes C. On the other hand, it has been proposed that an unmodified U at
the first position of the anticodon can form a base pair with C, through a
water bridge, as efficiently as those with A, G and U, and that the xo
5
U modification selectively abolishes this recognition of C without affecting the
recognition of the other three bases (
24
). It is obvious that this model is incorrect, at least for
E.coli
{t R N A} sub 1 sup {S e r}
.
In contrast, in systems such as mitochondria and
Mycoplasmas,
most tRNA species with an unmodified U at the first position of the anticodon
can recognize all four of the cognate codons terminating in A, G, U and C to
the same extent (
27
,
40
). However, it has been found recently that the
M.capricolum
threonine tRNA with the UGU anticodon is an exception: the efficiency of U-C recognition is much weaker than those of the other three, probably
because of the t
6
A modification at position 37 (
27
). Accordingly, some structural feature outside the wobble position of tRNA is
responsible for the efficient reading of C by U. In fact, the
M.mycoides
tRNA
Gly
with the UCC anticodon, but not an unmodified version of the
E.coli
tRNA
Gly
with the same anticodon, can read the GGC codon as efficiently as the other
three Gly codons on the
E.coli
ribosome (
40
). Furthermore, a single mutation of U to C, at position 32 of the unmodified
E.coli
tRNA
Gly
, was sufficient to switch its reading property to that of
M.mycoides
tRNA
Gly
(
41
), and vice versa (
42
). Thus, in the absence of particular structural features, the unmodified U at
position 34 of tRNA
Gly
cannot recognize the C at the third position of the codon. In contrast, the
unmodified form of
E.coli
{t R N A} sub 1 sup {S e r}
has C32, like
M.mycoides
tRNA
Gly
, but it did not recognize the UCC codon. This difference between
E.coli
{t R N A} sub 1 sup {S e r}
and
M.mycoides
tRNA
Gly
in the recognition of C at the third position of codon may be ascribed to the
difference in the stability of the base pairs between the first and second
positions of the codon and the second and third positions of the anticodon,
respectively: two G-C pairs for tRNA
Gly
, and one A-U pair and one G-C pair for {t R N A} sub 1 sup {S e r} (
43
,
44
).
In summary, the unmodified form of {t R N A} sub 1 sup {S e r} primarily reads the UCA codon, weakly reads the UCU codon, and hardly reads the UCC and UCG codons. This means that the specificity is higher for the
unmodified form than the modified form. This property is advantageous for use
in a cell-free translation system, particularly for the introduction of unnatural
amino acids into specific positions of proteins. An unnatural amino acid can be
introduced specifically into the positions encoded by UCA codons with the
unmodified tRNA, but not with the modified tRNA.
The higher specificity of the unmodified {t R N A} sub 1 sup {S e r} is mainly due to the lack of the wobble-extending, xo
5
U-type modification at position 34. This type of modified nucleoside occurs
in many other tRNAs, such as a tRNA
Val
and a tRNA
Ala
from
E.coli
(
13
). The unmodified forms of such tRNAs may be more specific than the modified
molecules, and would be useful for cell-free protein synthesis.
REFERENCES
Return