ABSTRACT
Mammalian tRNA 3
'
processing endoribonuclease (3
'
tRNase) removes 3
'
extra nucleotides after the discriminator from tRNA precursors. Here I examined
how the length of a 3
'
trailer and the nucleotides on each side of the cleavage site affected 3
'
processing efficiency. I performed
in vitro
3
'
processing reactions of pre-tRNA
Arg
s with various 3
'
trailers or various discriminator nucleotides using 3
'
tRNase purified from mouse FM3A cells or pig liver. On the whole, the
efficiency of pre- tRNA
Arg
3
'
processing by mammalian 3
'
tRNase decreased as the 3
'
trailer became longer, except in the case of a 3
'
trailer composed of CC, CCA or CCA plus 1 or 2 nucleotides, which was not able
to be removed at all. The distribution of 3
'
trailer lengths deduced from mammalian nuclear tRNA genomic sequences reflects
this property of 3
'
tRNase. The cleavage efficiency of pre-tRNA
Arg
s varied depending on the 5
'
end nucleotide of a 3
'
trailer in the order G
~
A > U > C. This effect of the 5
'
end nucleotide was independent of the discriminator nucleotides. The
distribution of the 5
'
end nucleotides of mammalian pre-tRNA 3
'
trailers reflects this differential 3
'
processing efficiency.
Eukaryotic tRNAs transcribed as larger precursors must be processed through a
series of steps to yield functional mature molecules. These processing
reactions include removal of extra 5' and 3' sequences, the addition of CCA to the 3' terminus, nucleotide modifications at specific residues,
and in a subset of the tRNA gene transcripts, RNA splicing (
1
,
2
). Recently, the editing of tRNA has also been found in mitochondria of various
species (
3
).
The 5' processing event in eukaryotic cells that generates the correct 5' ends of mature tRNAs is carried out by an endoribonuclease
similar to the well-studied
Escherichia coli
RNase P (
4
). On the other hand, the 3' processing in eukaryotes is essentially different from prokaryotic 3' processing in which the 3' extra residues following the CCA sequence of a tRNA
precursor are removed (
1
). 3' extra sequences, after the discriminator nucleotides, of eukaryotic tRNA
precursors must be removed precisely prior to the addition of the 3' terminal CCA sequence, because eukaryotic tRNA genes do not encode the
CCA residues (
1
). A general 3' processing model for
E.coli
tRNA precursors is proposed, in which the mature 3' terminus is generated by the first endonucleolytic cleavage in the 3' trailer, followed by at least two exonucleolytic trimming steps (
5
). Six exoribonucleases, RNase II, RNase D, RNase BN, RNase T, RNase PH and
RNPase, implicated in the tRNA 3' processing have been identified in
E.coli
by both biochemical and genetic studies (
5
).
There seems to be no general model for eukaryotic 3' processing because the mode of processing nucleases varies in different
systems. It has been demonstrated that 3' processing is achieved by the action of an endoribonuclease in many
studies using
in vitro
3' processing systems with cell extracts from human (
6
), pig (
7
),
Xenopus laevis
(
8
),
Drosophila
(
9
,
10
) and wheat (
11
); mitochondrial extracts from human (
12
), rat (
13
) and yeast (
14
); and chloroplast extracts from spinach (
11
). On the other hand, both endonuclease and exonuclease activities that can act
on the 3' termini of artificial tRNA precursors have been shown in germinal
vesicle extracts from
X.laevis
(
15
). Extracts from yeast nuclei and the silk gland of
Bombyx mori
contain a 3' -> 5' exonuclease that removes the 3' trailer of precursor tRNA (
16
,
17
). Despite these studies on many species, only two reports have described the
purification of 3' processing endoribonucleases, which cleave precursor tRNAs after the
discriminator nucleotide. A 3' processing endonuclease that accurately processes the 3' terminus of human pre-tRNA
Met
has been purified from
X.laevis
ovaries (
8
). It appears to function as a single polypeptide of ~97 kDa. Recently, I have purified 3' tRNase from pig liver, which appears to function as a dimer of ~45 kDa protein(s) (
7
).
Major determinants for substrate recognition by 3' tRNase probably reside in the mature tRNA domain which forms the well-conserved L-shape, because neither the sequences nor the structures of 3' trailers are conserved. This is supported by the
following two reports. In an
in vitro
Drosophila
system, more than half the pre-tRNA
His-48
variants containing a single mutation in secondary or tertiary base-pairs were 3'-processed less efficiently than the wild type (
10
). In a two half-tRNA mammalian system, pre-tRNA
Arg
variants with base changes in the T stem-loop region were cleaved by 3' tRNase less efficiently than the wild type, and extensive
deletions of the T stem-loop and extra loop regions abolished the substrate activity (
18
).
Here I investigated interactions between 3' tRNase and 3' trailers plus discriminator nucleotides. To see what effect 3' trailer length and the nucleotides on both sides of the
cleavage site had on 3' processing efficiency, I performed
in vitro
3' processing reactions of pre-tRNA
Arg
s with various 3' trailers or various discriminator nucleotides using 3' tRNase purified from mouse FM3A cells or pig liver. From the
results, I discovered that both the length and the 5' end nucleotide of pre-tRNA 3' trailers affect the 3' processing efficiency.
Wild type human pre-tRNA
Arg
, which is 5'-processed and has a 19 nt 3' trailer (
7
), and its variants were synthesized
in vitro
with T7 RNA polymerase (Takara Shuzo) from synthetic double-stranded DNA templates containing the T7 promoter. The transcription
reactions were performed in the presence or absence of [[alpha]-
32
P]UTP (Amersham Japan) under the conditions specified by the manufacturer
(Takara Shuzo). The transcripts were purified by denaturing polyacrylamide gel
electrophoresis.
Mouse FM3A cells were cultured on a large scale (a total of 192 l) in ES medium
(Nissui) containing 3% fetal calf serum and harvested at a density of 5 * 10
5
cells/ml. Cytosolic S100 extracts were prepared according to Nashimoto (
19
). 3' tRNase was purified from the S100 extracts basically in the same way as
it was from pig liver (
7
). Briefly, after precipitation of the extracts with ammonium sulfate (50%
saturation) and subsequent 55oC treatment, the sample was fractionated by a series of column
chromatographies with Q Sepharose Fast Flow, Blue Sepharose (twice), Heparin
Sepharose and Mono Q (Pharmacia). The purified 3' tRNase was aliquoted and frozen at -80oC.
The pre-tRNA 3' processing reaction was performed at 37oC for the indicated time in a reaction mixture (10 [mu]l) containing 10 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 3.2 mM
spermidine, 0.1 pmol of pre-tRNA
Arg
and 0.2 U of 3' tRNase purified from mouse FM3A cells or 0.4 U of 3' tRNase purified from pig liver (
7
). One unit of the enzyme is defined as the amount which converts 50% of R-G19 (0.1 pmol) to the 3' processed product in 10 min under the above conditions (
7
). After incubation, the processing products were resolved on a 10%
polyacrylamide-8 M urea gel, and quantitated with a Bio-Image Analyzer BA100 (FUJIX). In some cases, the gel was also
autoradiographed.
The 73 nt processing product of a cold pre-tRNA
Arg
, R-G19, was 3'-end-labeled with [5'-
32
P]pCp by T4 RNA ligase (Takara Shuzo), and gel-purified (
20
). The
32
P-labeled product was subjected to chemical RNA sequencing reactions (
21
) and resolved on a 20% polyacrylamide-8 M urea gel to determine the 3' terminal sequence.
3' processing of pre-tRNAs by 3' tRNase was examined at various concentrations of substrate
to obtain kinetic parameters. A reaction mixture (6 [mu]l) contained 10 mM Tris-HCl (pH 7.5), 0.5 mM dithiothreitol, 3.2 mM spermidine, 0.017-2.5 [mu]M
32
P-labeled pre-tRNA, and pig 3' tRNase fraction (0.2 ng) after Mono Q column chromatography
(
7
). After incubation at 37oC for 1 min, the reaction products were resolved on a 10% polyacrylamide-8 M urea gel and then quantitated with a Molecular Imager (BioRad).
Values of
K
m
and
V
max
were obtained from Lineweaver-Burk plots.
Mammalian nuclear tRNA genomic sequences were obtained from the GenBank/EMBL
data base by using a computer program, Findget (N. Fujita, National Institute
of Genetics, Mishima). The locations of tRNA genes were identified from the
GenBank/EMBL features table. Pre-tRNAs terminate with a uridine stretch of various lengths, such as UU, UUU
or UUUU (
22
,
23
). Therefore, the length of a pre-tRNA 3' trailer was defined as the distance from the end of a tRNA gene to
the third thymidine (including this nucleoside) in the four or more thymidine
stretch. Transcription termination also occurs at anomalous signals, TTCTT,
GTCTT or ATCTT (
24
). In this case the 3' trailer length was defined as the distance from the end of a tRNA gene
to the fourth nucleotide of such a signal. Rat mitochondrial tRNA genomic
sequences were likewise obtained from the data base; and human, bovine and
mouse mitochondrial tRNA genomic sequences were from published sources (
25
-
27
).
To analyze the influence of the length of 3' trailers of pre-tRNAs on 3' processing efficiency, I synthesized human pre-tRNA
Arg
(R-G19) with a 19 nt 3' trailer (
7
) and its six derivatives (R-G15, R-G13, R-G10, R-G8, R-G3 and R-G1) with shorter 3' trailers using an
in vitro
T7 transcription system (Fig.
1
A and Table
1
). The tRNA 3' processing reaction was performed with a mouse 3' tRNase fraction after Mono Q column chromatography. A protein of ~45 kDa was detected in this fraction on an SDS-polyacrylamide gel (Fig.
2
). The size of this mouse enzyme was the same as that of the pig enzyme, though
the pig enzyme fraction contained two 45 kDa proteins (
7
). Figure
3
A shows cleavage of those pre-tRNAs. On the whole, the cleavage efficiency increased as the 3' trailers became shorter (Table
1
and Fig.
4
A). Kinetic parameters were also determined using 3' tRNase purified from pig liver. The values of
V
max
/
K
m
increased with the decrease in length of the 3' trailers (Table
1
).
To examine the biological significance of the dependence of the 3' processing efficiency on the 3' trailer length, 80 different mammalian nuclear tRNA genomic
sequences were obtained from the GenBank/EMBL data base, and their 3' trailer lengths were deduced from the position of transcription
terminators of RNA polymerase III. A histogram of the 3' trailer lengths is shown in Figure
4
B. The peak is at 10-11 nt, and 72.5% of the tRNA genes have a 3' trailer with a length from 8 to 15 nt. The length distribution
patterns below and above 7 nt agree well with the graphic patterns of the
cleavage efficiency of the R-CCA and R-G series, respectively (Fig.
4
B). The implications of these apparent correlations are discussed below.
The above experimental results indicated that 3' processing efficiency is affected by the 5' 3 nt of a 3' trailer. To analyze this sequence effect, I synthesized
another five derivatives of the pre-tRNA
Arg
(R-CCA19): R-CCG, R-CUA, R- UCA, R-ACA19 and R-GCA (the 5' 3 nt of the 3' trailer shown after the
hyphen). They each have a 19 nt 3' trailer with another nucleotide substituted for one of the 5' 3 nt CCA. Although the 3' processing efficiency of R-CCG and R-CUA was as low as that of R-CCA19, the other three pre-tRNAs were cleaved in the order R-ACA19 > R-GCA > R-UCA more than
twice as well as R-CCA19 (Table 2). R-G19 with a 19 nt 3' trailer beginning with GUG was also cleaved three times as
well as R-CCA19. These results show that the 5' 3 nt, especially the 5' end nucleotide, of a 3' trailer are critical for determining tRNA 3' processing efficiency, and that cytidine at
the 5' end reduces the efficiency.
Table 2
I further analyzed the effect of 5' nucleotides on 3' processing efficiency using three additional pre-tRNAs: R-ACA3, R-U1 and R-A1, which have 3' trailers consisting of ACA, U and
A, respectively. Among the three tested pre-tRNAs with 3 nt trailers, the cleavage of R-ACA3 and R-G3 was very efficient, while the cleavage of R-CCA3 was not detected (Table
2
). With regard to the four pre-tRNAs with 1 nt trailers, U, A and G were removed very efficiently, but R-C1 was hardly cleaved (Fig.
3
B and Table
2
). These 3' processing analyses using various pre-tRNA
Arg
s with systematically mutated 3' trailers led me to conclude that, for pre-tRNAs with 3' trailers of the same length, the 5' end nucleotide of the 3' trailer is the key factor determining
cleavage efficiency.
I demonstrated above that the 5' terminal nucleotide of a 3' trailer greatly affects the tRNA 3' processing reaction. This conclusion raised a question as
to whether a nucleotide 5' of the 3' processing site, the discriminator, also affects the cleavage
efficiency. To answer this question, I tested 16 species of pre-tRNA
Arg
s, each of which contains one of the dinucleotides on both sides of the 3' processing site, for cleavage using 3' tRNase purified from mouse FM3A cells or pig liver. The enhancing
effect of the 5' end nucleotide of a 3' trailer on cleavage efficiency was in the order G ~ A > U > C (Figs
5
and
6
and Table
3
). On the other hand, although there were differences in 3' processing efficiency among the four pre-tRNAs with the same 5' end base of the 3' trailer, there seemed to be no general rule
governing the effect of the discriminator nucleotide on 3' trailer cleavage (Figs
5
and
6
).
Table
4
shows the distribution of the 5' end nucleotides of pre- tRNA 3' trailers deduced from mammalian nuclear and mitochondrial
tRNA genomic sequences. The nuclear distribution differs considerably from the
mitochondrial one: each purine occurs more frequently than each pyrimidine in
the nuclear distribution, and A occurs prominently in the mitochondrial one
(Table
4
). The nuclear and mitochondrial frequencies were plotted against the values of
V
max
/
K
m
of pre-tRNAs with a 3' trailer beginning with the corresponding nucleotide (Fig.
7
). The nuclear frequency correlates well with the 3' processing efficiency, while the mitochondrial frequency does not.
The length distribution of nuclear pre-tRNA 3' trailers correlates well with the 3' processing efficiency of the R-G series pre-tRNAs for 3' trailers with a length of >7 nt (Fig.
4
B). This may reflect a demand for efficient tRNA synthesis in cells. However,
the correlation does not hold with regard to smaller 3' trailers and the distribution agrees rather with the cleavage efficiency
of the R-CCA series pre-tRNAs (Fig.
4
B). RNA polymerase III usually terminates transcription within the four
thymidine stretch terminator, generating a transcript with a two to four
uridine stretch corresponding to the terminator at its 3' terminus. Thus, it is possible that the polymerase produces pre-tRNAs with trailers that are several nucleotides long. Among them 3' trailers beginning with C, CC or CCA should be hard for 3' tRNase to remove. In particular, pre-tRNAs with the 3' trailer CCAUU, which probably cannot be
removed by 3' tRNase, may be deleterious to cells. The low frequencies of short 3' trailers may reflect a demand for the prevention of deleterious
tRNA synthesis in cells.
In eukaryotic cells, the CCA terminal residues must be added to the 3' end of pre-tRNAs from which a 3' trailer was removed, to generate functional tRNAs, since
eukaryotic tRNA genes do not encode the CCA sequence. This reaction is
catalyzed by tRNA nucleotidyltransferase, which has the activity to incorporate
not only CMP and AMP residues, but also UMP incorrectly, into tRNA molecules (
29
). Under certain conditions, this enzyme produces anomalous tRNAs with 3' terminal residues other than the CCA (
29
). The CCA end of tRNA is critical for both aminoacylation (
30
,
31
) and binding to large rRNA in the peptidyl transferase center of the ribosome (
32
) in
E.coli
and probably also in eukaryotes. Therefore, some error-correcting mechanism may exist to remove non-CCA residues from those anomalous tRNAs. 3' tRNase might be one of the enzymes involved in such a
mechanism since this endoribonuclease can remove any 3' trailer except CC, CCA and CCA plus 1 or 2 nt from pre-tRNAs.
In this study I demonstrated that 3' tRNase recognizes not only the L-shaped mature tRNA domain but also the 3' trailer to determine cleavage efficiency. 3' tRNase clearly discriminates the nucleotide C at the
5' termini of a 3' trailer from the others (Table
2
and Figs
5
and
6
). Besides that, the very short 3' trailers CC, CCA and CCA plus 1 or 2 additional 3' nt can be distinguished by this enzyme from the other 3' trailers (Table
1
). 3' tRNase may have a binding domain for CCA residues. The CCA-binding domain may be composed of two cytidine binding sites and an
adenosine binding site. The cleavage efficiency of pre-tRNA
Arg
s varies depending on the 5' end nucleotide of a 3' trailer in the order G ~ A > U > C (Figs
5
and
6
and Table
3
). The differential 3' processing efficiency might reflect the affinity of the 5' end nucleotide of a 3' trailer to the first cytidine binding site in the order C
> U > A ~ G. It may as well be possible that CCA serves as a negative determinant
and that the substrate binding site of 3' tRNase preferentially accommodates other sequences. 3' trailers beginning with CCA residues can be removed by 3' tRNase if they are longer than 5 nt (Table
1
). This suggests that mammalian 3' tRNase may also grip a certain number of distal nucleotides of a 3' trailer.
I thank K. Maeda and N. Teramachi for excellent technical assistance and S.
Takano for assistance in preparing the figures.
Pre-tRNA
Sequence
a
Cleavage (%)
b
5' <=> 3'
R-CCA19
***CUCG CCAUAAGCAGGGUCGUUUU
10.1 +- 1.5
R-CCG
***CUCG CCGUAAGCAGGGUCGUUUU
8.9 +- 1.2
R-CUA
c
***CUCG CUAUAAGCAGGGUCGUUUU
11.8 +- 0.8
R-UCA
***CUCG UCAUAAGCAGGGUCGUUUU
21.2 +- 1.5
R-ACA19
***CUCG ACAUAAGCAGGGUCGUUUU
37.9 +- 2.9
R-GCA
***CUCG GCAUAAGCAGGGUCGUUUU
22.6 +- 2.7
R-CCA3
***CUCG CCA
ND
R-ACA3
d
***CUCG ACA
81.9 +- 4.8
R-G3
***CUCG GUG
73.7 +- 2.5
R-C1
***CUCG C
7.5 +- 1.4
R-U1
***CUCG U
99.0 +- 0.5
R-A1
***CUCG A
99.8 +- 0.1
R-G1
***CUCG G
99.2 +- 0.6
REFERENCES
Return