ABSTRACT
We have isolated the poly(A) polymerase (PAP) encoding gene
pla1
[for
The 3'-ends of almost all eukaryotic mRNAs are generated by
endonucleolytic cleavage of a primary transcript and subsequent polyadenylation
of the upstream cleavage product [for review, see (
1
,
2
)]. Both steps are accomplished by a multiprotein complex. In the budding yeast
Saccharomyces cerevisiae
four 3'-processing activities have been separated: cleavage factor I (CFI), cleavage
factor II (CFII), polyadenylation factor I (PFI) and poly(A) polymerase (PAP).
Cleavage can be reconstituted
in vitro
with CFI and CFII, whereas polyadenylation requires PAP, PFI and CFI (
3
). Polyadenylation is specific in the presence of Mg
2+
and only RNAs containing specific 3'-processing signals can be polyadenylated. In the absence of the
specificity factors PFI and CFI, PAP can polyadenylate an RNA only very
inefficiently and without substrate specificity. This unspecific activity is
enhanced when Mg
2+
is replaced by Mn
2+
(
4
).
Poly(A) polymerase has been cloned from
S.cerevisiae
(
5
) and from various vertebrates including human (
6
), bovine (
7
,
8
) and
Xenopus
(
9
). The vertebrate PAPs are highly homologous to each other throughout their
entire sequence.
S.cerevisiae
PAP is encoded by the essential gene
PAP1
.
PAP1
-encoded protein and its bovine counterpart are 47% identical within the
first 395 amino acids. Mutational analysis of bovine PAP revealed a catalytic domain within the
conserved N-terminal part (
10
). The high conservation of this domain suggests the same function also in other
poly(A) polymerases. The C-terminal regions are not conserved on the primary sequence level.
Nevertheless, recent results show that both C-termini contain an RNA-binding domain. In both cases a nuclear localization signal (NLS)
was found to be essential also for RNA-binding (
10
,
11
).
In spite of these similarities, bovine PAP is not able to rescue a
PAP1
deletion mutant (
12
) and
S.cerevisiae
PAP can not replace the bovine enzyme in the mammalian
in vitro
polyadenylation system. The reason for this might be the inability of both polymerases to interact with the specificity factors of the respective other species.
S.cerevisiae
PAP has been shown to directly interact with the FIP1 protein, a subunit of PFI
(
13
). Bovine PAP forms a complex with the mammalian cleavage and polyadenylation specificity factor CPSF on the substrate RNA (
14
). Other differences between mammalian and
S.cerevisiae
polyadenylation are the poly(A) tail length and the signals on the RNA directing 3'-processing (
15
,
16
). In
S.cerevisiae
, poly(A) tails consist of 60-70 A residues whereas mammalian poly(A) tails reach a length of ~250 A residues.
In summary, many similarities, but also several differences, have been found
between pre-mRNA 3'-processing in mammals and in
S.cerevisiae
. It is not known whether or not
S.cerevisiae
is a typical representative of the lower eukaryotes with regard to 3'-processing. Some evidence for a strong conservation of 3'-end formation among lower eukaryotes comes from a
study dealing with the 3'-processing signals of
S.cerevisiae
and the distantly related fission yeast
Schizosaccharomyces pombe
. It was found that
in vitro
, the
S.pombe
URA4
transcript is correctly processed in a
S.cerevisiae
extract and that
S.pombe
can efficiently process the
S.cerevisiae
CYC1
transcript
in vivo
(
17
). This finding was surprising because
S.pombe
and
S.cerevisiae
are evolutionarily distant. In fact, several
S.pombe
genes show about the same degree of homology with their counterparts in
S.cerevisiae
and with those in higher eukaryotes (
18
). On the other hand, the fact that the 3'-end formation signals can be exchanged between the two highly
diverged yeast species does not mean that the
trans
-acting factors are conserved as well. For example, the conservation could
be limited to the RNA-binding sites of the specificity factors. The cloning of the
S.pombe
poly(A) polymerase appeared to be a reasonable starting point to find out whether the 3'-processing factors of
S.pombe
are more closely related to those of mammals or those of
S.cerevisiae
.
The amino acid sequences of two conserved peptides of
S.cerevisiae
PAP and bovine PAP were used to design two degenerate PCR primers: 5'-CCGTCGACGGIAARATHTTYAC-3' (R = A or G, Y = C or T, H = A or C or T) from the
peptide sequence GKIFT (in Fig.
1
: amino acids 82-86 in
PAP1
) and 5'-CCGGATCCACIARCATNGCCCA-3' from the peptide sequence WAMLV (amino acids 236-240 in
PAP1
). The 50 [mu]l PCR mixture contained: 10 mM Tris-HCl (pH 8.0), 50 mM KCl, 2.5 mM MgCl
2
, 0.01% Triton X-100, 0.2 mM deoxynucleoside triphosphates, 250 pmol PCR primers, 4 ng genomic
S.pombe
DNA and 1 U
Taq
DNA polymerase (Perkin Elmer). Twenty-five cycles were carried out, with annealing at 48oC for 30 s and extension at 72oC for 2 min. A PCR product of the expected size (476 bp) was
reamplified and subcloned into the
Sal
I-
Bam
HI site of the Bluescript KS vector (Stratagene) and sequenced with Sequenase
(Amersham Life Science Inc.).
A random-primer labelled probe was synthesized from the subcloned PCR fragment. The
probe was used to screen a
S.pombe
cDNA plasmid library (vector: pFL61, see Table
1
). The cDNA library was a gift from Dr François Lacroute (Gif-sur-Yvette, France). 2.5 * 10
4
colonies were screened on nitrocellulose filters according to standard
procedures (
19
).
Table 1
Protein sequences were analyzed with GCG software programs. Sequence databases
were searched with the programs FASTA and TFASTA (
20
). For sequence alignments the program PILEUP was used. The program BESTFIT was
used to calculate the percentages of identity. The nucleotide sequence of
pla1
has been deposited in the EMBL database under the accession number X79705. The
email server at the EMBL was used for the secondary structure prediction (
21
,
22
).
Proteins were separated on 9% polyacrylamide gels (
23
) and blotted to nitrocellulose membranes. The blots were blocked in TN Tween
[10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.05% Tween 20] containing 5% non-fat milk. The same buffer was also used for the antibody
incubations. Peroxidase-conjugated swine anti-rabbit immunoglobulins were used as secondary antibodies. Chemiluminescence (ECL kit, Amersham) was used for detection. Affinity-purified antibodies directed against
S.cerevisiae
PAP were a gift from Pascal Preker.
Yeast media were prepared as described elsewhere (
24
). Yeast transformations were carried out following the lithium acetate procedure (
25
). The strain used in this study was
S.cerevisiae
JL17-3A (
Mat
[alpha],
ade2-1
his3-11,15
leu2-3,112
trp1-1
ura3
can1-100
pap1::LEU2
, pHCp50) (
26
).
In the plasmid shuffling experiment, JL17-3A was transformed either with pFLT-pla1, pHCBS3 (positive control, see Table
1
) or pFLT (negative control). Transformants were transferred to SD-Leu plates containing 5-fluoroorotic acid (5-FOA) to select for cells having lost pHCp50 (
27
). The strain carrying pFLT-pla1 instead of pHCp50 is called JL17PLA, the one carrying pHCBS3 is
called JL17PAP.
The open reading frame of
pla1
was amplified by PCR with two primers: 5'-CGGGATCCATGACTACCAAGCAATGG-3' and 5'-CGGGATCCTTATGCCGTTGAAACTTT-3'. The reaction was carried out with
Pfu
DNA polymerase (Stratagene) according to the manufacturer's instructions. The PCR product was subcloned into the
Bam
HI site of the histidine-tag expression vector pQE9 (Qiagen).
Escherichia coli
M15 was used to overexpress pQE9-pla1. The expression was induced by adding 0.4 mM isopropyl-[beta]-D-thiogalactopyranoside (IPTG) to a 500 ml LB
culture. The induced culture was grown overnight. After harvesting by
centrifugation, the cells were resuspended in 50 ml TK buffer [50 mM Tris-HCl (pH 8.0) , 100 mM KCl] and frozen at -80oC. The cells were thawed after addition of 100 ml lysis
buffer [50 mM Tris-HCl (pH 8.0), 100 mM KCl, 15% glycerol, 1.5 mM phenylmethylsulfonyl
fluoride (PMSF), 0.6 mg/ml lysozyme, 15 [mu]g/ml DNase I, 3 [mu]g/ml leupeptin, 3 [mu]g/ml pepstatin]. After sonication and centrifugation the supernatant
was filtered and batch absorbed to 2 ml of a 50% slurry of nickel-nitrilotriacetic acid agarose (Ni-NTA, Qiagen), which had been
equilibrated with GTK buffer [50 mM Tris-HCl (pH 8.0), 100 mM KCl, 10% glycerol]. After 30 min the suspension was poured into a
column. The column was washed with 10 ml GTK buffer containing 10 mM imidazole.
The recombinant protein was eluted with 5* 1 ml GTK buffer containing 100 mM imidazole. The pool of recombinant
protein was loaded on a 1 ml DEAE-Sepharose column. The column was washed with
2* 1 ml GTK buffer containing 150 mM KCl and with 2* 1 ml GTK buffer containing 500 mM KCl. Recombinant PLA1 protein
eluted in the 150 mM step.
Unspecific polyadenylation was done as described previously (
28
). Specific polyadenylation of the
in vitro
transcript pG4-CYC1 pre (
13
) was performed as described previously (
29
). Specific polyadenylation of the L3 precleaved RNA (
30
) was done as described (
14
). The reaction products were analyzed on denaturing polyacrylamide gels. Purified CPSF was a gift from Silvia Barabino and Andreas
Jenny. Recombinant bovine PAP and
S.cerevisiae
PAP were gifts from Georges Martin.
Yeast extracts were prepared as described previously (
31
) with the modifications described below. Cells were harvested at OD 2.5-4. Spheroplasts were homogenized in the presence of the protease
inhibitors PMSF (1 mM), pepstatin (0.7 [mu]g/ml) and leupeptin (0.4 [mu]g/ml). Potassium acetate was used instead of KCl. Whole cell extracts
were brought to 45% ammonium sulphate saturation. The same protease inhibitors (at the same concentrations) as described above were also used in the dialysis buffer.
Alternatively, yeast extracts for
in vitro
3'-processing were prepared by freezing the cells in liquid nitrogen and homogenizing them with a mortar and pestle as described previously (
32
). The ammonium sulphate precipitation and the dialysis were done as described
above.
In order to clone the poly(A) polymerase of
S.pombe
two degenerate primers were derived from the peptide sequences GKIFT and WAMLV,
which are conserved in
S.cerevisiae
PAP and bovine PAP. The primers were used for PCR on genomic
S.pombe
DNA. A band corresponding to the expected size was subcloned and sequenced. The
putative peptide sequence of the PCR product was found to be 70% identical to
S.cerevisiae
PAP and 58% identical to bovine PAP. The amplified DNA could therefore be used
as a probe for screening a
S.pombe
cDNA library. Three positive clones with a 2.1 kb insert and identical
restriction pattern were obtained. The insert of one of them (
pla1
) was subcloned into the
Not
I site of the Bluescript KS vector and sequenced on both strands. Sequencing of
pla1
revealed a putative ORF of 566 amino acids corresponding to a 64 kDa protein.
As there are stop codons in front of the start ATG we conclude that
pla1
is a full length clone. The
pla1
protein sequence contains a putative bipartite nuclear localization signal (
33
) at its C-terminus (amino acids 545-562).
Database searches with
pla1
yielded high scores with poly(A) polymerases but not with any other protein
sequence (data not shown) suggesting that
pla1
encodes a poly(A) polymerase. An alignment of the
pla1
amino acid sequence with that of
S.cerevisiae
PAP and bovine PAP reveals a high degree of identity, especially in the N-terminal part. In the first 400 amino acids,
pla1
is 62% identical to
S.cerevisiae
PAP and 47% identical to bovine poly(A) polymerase. The alignment in Figure
1
shows blocks of complete conservation in all three sequences in the N-terminal part but also stretches that are highly conserved only among the
two yeast enzymes, for example in residues 308-316 (the numbers refer to the
pla1
sequence). The region recently identified as the catalytic core of bovine PAP (
10
; see Figs
1
and
2
, residues 59-179) is quite well conserved in
pla1
and within this part the three aspartates involved in catalysis in the
mammalian enzyme (
10
) are conserved (see Fig.
1
). Although the C-termini are less conserved, there is significant homology among the two
yeast poly(A) polymerases (35% identity). One of the most striking common C-terminal features of
pla1
and
PAP1
is a predominantly acidic stretch (amino acids 528-538) which is followed by a predominantly basic region, containing the
putative NLS. The sequence LPDEVF(D/E) is conserved among
pla1
and
PAP1
but is absent in the C-terminus of the bovine poly(A) polymerase (Figs
1
and
2
). The C-terminal portions of
pla1
and bovine PAP are 23% identical. The overall identity of
pla1
and
S.cerevisiae
PAP is 55%, whereas
pla1
and bovine PAP are only 38% identical. From these data we conclude that
pla1
is more closely related to
S.cerevisiae
PAP than to its mammalian counterpart.
The ORF of
pla1
was expressed with a histidine-tag in
E.coli
and the recombinant protein was purified as described in Materials and Methods.
On a SDS-polyacrylamide gel the recombinant protein appears as a band of ~67 kDa which is in good agreement with the calculated molecular mass
(64 and 1.4 kDa for the histidine tag).
In an unspecific polyadenylation assay a specific activity of 8 * 10
5
U/mg was measured. The activity of PLA1 protein is in the same range as that of
S.cerevisiae
PAP (1.3 * 10
6
U/mg).
By complementation of the
S.cerevisiae
PAP1
deletion mutant JL17-3A we addressed the question whether
pla1
can functionally substitute for
PAP1
. In JL17-3A the chromosomal copy of
PAP1
is replaced by
LEU2
(
26
). The strain is rescued by pHCp50 which contains
PAP1
and
URA3
. JL17-3A was transformed with pFLT-pla1. Transformants were transferred on a SD-LEU plate containing 5-FOA. 5-FOA kills cells that contain a wild-type copy of
URA3
(
27
). Therefore, survival depends on the loss of pHCp50 and on the ability of
pla1
to take over the essential function of
PAP1
. Viable colonies were obtained in the plasmid shuffling with pFLT-pla1. These cells grew slightly slower than those rescued by pHCBS3 (160
min doubling time versus 95 min). To exclude the possibility that the growing
colonies had maintained pHCp50 (for example by mutating the
URA3
marker), we prepared an extract from the
S.cerevisiae
strain expressing the
S.pombe
poly(A) polymerase (JL17PLA) and probed it on a western blot with a polyclonal
antiserum directed against
S.cerevisiae
PAP. Figure
3
shows that the extract does not contain
S.cerevisiae
PAP (lanes 6-8) whereas PAP can easily be detected in a control extract from JL17PAP
(lanes 3-5). A significant amount (30 ng) of recombinant PLA1 protein (lane 2) is
only visible as a very faint band due to a crossreaction with anti-PAP antibodies. It is thus not surprising that PLA1 protein could not be
detected in the extract from JL17PLA cells.
In order to test whether PLA1 protein can also have a specific poly(A)
polymerase activity
in vitro
we made an extract from JL17PLA and tested it in a specific polyadenylation
assay. The substrate for this reaction was the CYC1 precleaved RNA which ends
at its natural polyadenylation site and can be polyadenylated by a
S.cerevisiae
wild-type extract (
13
). Figure
4
shows a polyadenylation assay, in which extracts from JL17PAP and JL17PLA were
titrated. The extract from JL17PLA (lanes 6-9) produces essentially the same reaction products as the wild-type (lanes 2-5). The polyadenylation reaction seems to be slightly less
efficient with the extract from JL17PLA than with the wild-type extract, but in both cases the poly(A) tails reach approximately the
same length (compare lanes 5 and 9). The long poly(A) tails in lanes 2-4 are probably due to inefficient poly(A) length control in reactions
with low amounts of wild-type extract. This effect is less pronounced for JL17PLA (lanes 6-8). In the reactions with the extract from JL17PLA a band of
intermediate length can be seen. This band represents oligoadenylated precursor RNA, as
confirmed by binding to poly(U)-Sepharose (results not shown). The intermediate
band may be caused by decreased stability of the polyadenylation complex in the
presence of PLA1, leading to a premature release of the polyadenylation
product. In any case, we conclude that PLA1 can replace the
S.cerevisiae
poly(A) polymerase in a specific polyadenylation assay.
In this paper we report the cloning of the
S.pombe
gene
pla1
which is highly homologous to the
S.cerevisiae
poly(A) polymerase gene
PAP1
and is capable of complementing a
S.cerevisiae PAP1
deletion mutant.
Recombinant PLA1 protein is active in unspecific polyadenylation. Thus,
pla1
encodes a poly(A) polymerase. Sequence comparisons with other poly(A)
polymerases show that
pla1
is more closely related to
PAP1
(62% identity within the first 400 amino acids) than to bovine PAP (47%
identity in the N-terminal part). The two yeast poly(A) polymerases differ approximately
equally from their bovine homologue on the primary sequence level. The same
tendencies can also be seen in an alignment of the C-termini. Whereas it was not possible so far to find a significant sequence
homology between the C-terminus of
PAP1
and any other poly(A) polymerase, the C-termini of
pla1
and
PAP1
are 35% identical. This identity is significant because alignments of
randomized versions of the two C-termini always yielded identities below 25%.
A high similarity of the two yeast PAPs is not only observed on the sequence
level but also with regard to the ability of
pla1
to rescue the
S.cerevisiae
PAP1
deletion mutant. An extract prepared from JL17PLA efficiently polyadenylates the CYC1 precleaved RNA and the products are comparable to those obtained with a wild-type
S.cerevisiae
extract. Extracts from mutants affected in essential polyadenylation factors
are inactive in specific polyadenylation
in vitro
(
13
,
29
), although they contain a functional PAP. The reason for this might be the very
inefficient manner by which poly(A) polymerases elongate RNA substrates in the
absence of specificity factors (
28
). Therefore, in order to produce normal poly(A) tails, PLA1 protein has to
interact with the
S.cerevisiae
polyadenylation specificity factors (PFI, CFI or yet unidentified additional
components). This suggests that similar factors exist in
S.pombe
and argues for a conservation of the polyadenylation machinery in the two yeast
species.
In contrast to
pla1
bovine PAP can not rescue a
S.cerevisiae
PAP1
null allele (
12
) and none of the yeast polymerases is able to replace bovine PAP
in vitro
in the mammalian polyadenylation system. Consequently, there must be an
important difference between the two yeast polymerases and bovine PAP. Because
the
S.cerevisiae
poly(A) polymerase and its bovine homologue differ mostly in their C-terminal parts, the C-termini have been suggested to contain domains involved in species
specific functions such as interactions with specificity factors (
5
,
8
). One striking feature common to both yeast PAPs is the sequence LPDEVF(D/E)
which is located upstream of the putative NLS. This sequence is absent in the
mammalian poly(A) polymerase, which suggests that LPDEVF(D/E) may have a
function that is specific for yeast PAPs. Interestingly, a recent study showed
that a truncated version of
PAP1
ending at amino acid 525 could still rescue a
PAP1
deletion mutant whereas deletion of additional 12 amino acids, including
LPDEVFD, was lethal (
11
). These authors also showed that a predominantly basic region overlapping the
NLS is involved in RNA-binding. A similar basic region is also present in
pla1
and is aligned with the RNA-binding domain of
PAP1
in Figure
1
(amino acids 537-547 in
pla1
). The NLS1 of bovine PAP (amino acids 489-507) was also found to be essential for RNA-binding (
10
). This indicates that the involvement of a NLS in RNA-binding is a general feature of poly(A) polymerases.
It is surprising that the 525 amino acid truncation of
PAP1
is able to complement the null allele, although it is severely affected in RNA-binding (
11
). A possible explanation for this is the interaction of PAP with the
specificity factors, which might still allow the mutant polymerase to establish
the contact to its RNA substrate. This would require that also the truncated
PAP forms a stable complex with the specificity factors. Further deletion of 12
amino acids (position 514-525 in
PAP1
) may destroy the residual activity of the enzyme by affecting the complex
stability. It is tempting to speculate that the conserved LPDEVFD sequence
motif could be involved in the formation of the polyadenylation multiprotein
complex.
The region comprising the first 18 amino acids of
PAP1
has been implicated in the interaction with polyadenylation specificity factors (
11
). Because PLA1 must be able to interact with the
S.cerevisiae
polyadenylation specificity factors, a strong conservation of the extreme N-termini would be expected. Thus, it is surprising that the extreme N-terminus is not completely conserved among the two yeast
polymerases. In fact
pla1
and bovine PAP show about the same degree of identity to
PAP1
within this stretch. On the other hand, the similarity of this region is
clearly higher than random. This may serve as an argument for a functional
conservation of this part. However, all N-terminal truncations of bovine PAP tested so far were found to be inactive
even in unspecific polyadenylation (
10
). It is therefore still unclear whether or not the involvement of the extreme N-terminus in specific protein-protein interactions is a general characteristic of poly(A)
polymerases. This and other questions concerning the functions of specific
domains of poly(A) polymerases can be answered by further mutational analysis
and by the determination of the crystal structure of these enzymes.
We are grateful to Dr François Lacroute for providing the
S.pombe
cDNA library. We thank Georges Martin for discussions on multiple sequence
alignments and for help with secondary structure predictions. We also thank
Andreas Jenny, Pascal Preker and all other members of our laboratory for
helpful and constructive suggestions. L.M-S. was supported by a `Human Capital and Mobility ' postdoctoral
fellowship from the EU. This work was supported by the Kantons of Basel and a
grant from the Schweizerischer Nationalfonds.
Name
Description
pFL61
2[mu]-based vector with a
URA3
marker (34)
pFL61-pla1
pFL61 containing
pla1
in its
Not
I site under the control of the PGK promoter
pFLT-pla1
Derivative of pFL61-pla1 in which the
Bgl
II
URA3
cassette has been replaced by the
Bgl
II
TRP1
cassette
pHCp50
Contains
URA3
and the genomic
Hin
dIII fragment of
PAP1
; rescues a chromosomal
PAP1
disruption (26)
pHCBS3
Contains
TRP1
and the genomic
Hin
dIII fragment of
PAP1
; rescues a chromosomal
PAP1
disruption (26)
REFERENCES
Return