ABSTRACT
The compact organization of the
Saccharomyces cerevisiae
genome necessitates that non-coding regulatory sequences reside in close proximity to one another. Here
we show there is an intimate association between transcription terminators and
DNA replication origins. Four replication origins were analyzed in a reporter
gene assay that detects sequences that direct 3
'
end formation of mRNA transcripts. All four replication origins function as
orientation-independent transcription terminators in this system, producing truncated
polyadenylated mRNAs. Despite this close association, the
cis
-acting elements that confer replication origin function are genetically
separable from those required for transcription termination. Several models are
explored in an attempt to address how and why the signals specifying
transcription termination and replication initiation overlap.
One of the surprising results from the
Saccharomyces cerevisiae
genome sequencing project is that the density of genes on yeast chromosomes is
greater than was initially estimated (
1
). An implication of this finding is that many of the non-coding regulatory DNA sequences must be confined within relatively short
intergenic regions. One clear example of this arrangement is exemplified by the
HMRE locus. At HMRE there is an association between the components of a genetic
silencer and a replication origin (reviewed in
2
). Thus, it is possible that there are examples of other distinct regulatory
elements of the yeast genome which reside within the same DNA sequence. While
attempting to identify possible sites of protein-binding in replication origin sequences by computer search, we noted that
several origins contain matches to sequences that may be involved in
transcriptional termination. Relatively few reports associating the sequences required for transcriptional
termination with those required for DNA replication exist. One of the best
studied yeast origins, ARS1, contains a termination site for the TRP1 gene
transcript (
3
), but no conclusion regarding the generality of this coincidence has been made.
Interestingly, a reported transcription termination site from a cryptic RNA
polymerase II (Pol II) mRNA in the rDNA of yeast (
4
) maps to the region identified to contain the minimal rDNA replication origin
sequence (
5
). Consequently, the limited biological information regarding coincidental
localization of terminators and replication origins compelled us to investigate
this relationship further.
The mechanism of mRNA 3' end formation, that is, cleavage, polyadenylation and transcription
termination, is conserved in eukaryotic organisms ranging from the yeast
S.cerevisiae
to mammalian cells (for reviews see
6
-
8
). However, the sequences which direct cleavage and polyadenylation in yeast are
not as conserved as in metazoans. In general, sequences that direct 3' end formation are A-T-rich, which is not surprising given their lack of coding
capacity. However the overall A-T richness of the DNA is insufficient to account for termination (
9
). The lack of a specific termination consensus sequence in yeast suggests that
either: (i) the signals are redundant, such that there are several classes of
signals which may be recognized by the processing/transcription machinery, or
(ii) that other features, such as RNA secondary structure, may be important for
mRNA 3' end formation.
In an attempt to determine which
cis
-acting sequences are important for transcription termination, analysis has
focused on the sequences found at the 3' end of genes. As noted, ARSs (autonomously replicating sequences) are
also found in this region of the DNA. In their native chromosomal context, only
a subset of ARSs actually function in chromosomal replication (
10
). Of the 13 ARSs found on chromosome III, all are located in intergenic
regions. This pattern is repeated for chromosome VI, where the locations of
nine ARSs have been determined (
11
). Interestingly, the sequences of these ARSs do not preclude their presence in
coding DNA, as there are often open reading frames contained within ARSs (
12
). Like terminators, ARSs share some limited sequence conservation. A near or
perfect match to the 11 bp consensus sequence, WTTTAYRTTTW, is common to all
ARSs and origin sequences (
13
,
14
). An origin recognition complex (ORC) recognizes the ARS consensus and flanking
DNA and may be necessary, but not sufficient to direct ARS function (
15
-
20
). The ORC binding site extends 3' from the consensus and partially overlaps an A-T-rich sequence of ~100 bp immediately downstream from the ARS consensus
sequence. Although the ARS 3' flanking region is not well conserved with regard to primary sequence,
it is conserved with regard to function, as one ARS 3' region can substitute for another (
21
). In addition, this region has the common feature of being easily unwound and
the ease of unwinding has been correlated with ARS replication efficiency (
5
,
22
,
23
). The ARS 3' flanking region is sensitive to linker scanning mutations, but generally
insensitive to point mutations (
20
,
21
). Some ARS elements contain sequences that bind transcription factors that
function as replication enhancers, the most notable being the Abf1 protein (
24
,
25
). Interestingly other transcription factor binding sites, such as the GAL 4 or
RAP 1 sites, can substitute for the ABF1 site (
24
).
Comparison of the components of ARSs with those of transcriptional terminators
reveals many similarities. ARSs and transcriptional terminators share the
features of A-T richness, poor sequence conservation and intergenic location. ARSs and
transcription terminators may utilize these features to specify function with
limited regard to primary sequence identity. The usage of such features may
account for the difficulty in readily identifying the critical
cis
-acting components of replication origins and transcriptional terminators.
It has been postulated that there is a requirement for ARSs to be located in
transcriptionally silent regions of the chromosome (for review see
12
). In support of this hypothesis, several groups have demonstrated that
transcription from a strong promoter into ARS1 impairs its function (
3
,
26
,
27
). Interestingly, when ARS1 is downstream of the GAL 1 promoter, transcripts end
within the ARS, suggesting a dual role in directing both transcription
termination and plasmid replication (
3
). However, the mechanism of termination and how the termination signal relates
to sequences known to be important for replication are unknown. In order to
examine the generality of the relationship between sequences directing
replication and those signaling transcription termination, we tested the
ability of four ARSs to stop transcription using an
in vivo
termination assay. In the present study we demonstrate that termination
activity is associated with each of the ARSs tested.
Culture media components were obtained from Difco. Amino acids 5-bromo-4-chloro-3-indolyl-[beta]-D-galactopyranoside (X gal)
and o-nitrophenyl-[beta]-D-galactosidase (ONPG), as well as most other
chemicals, were from Sigma. Restriction endonucleases, T3 and T7 RNA
polymerases,
Taq
DNA polymerase, AMV reverse transcriptase, RNasin and calf intestinal alkaline
phosphotase (CIAP) were obtained from Promega and used according to
manufacturer's specifications. T4 DNA ligase was obtained from New England
Biolabs. Sequencing was carried out using the Sequenase Kit (US Biochemicals)
according to the manufacturer's instructions. Radiolabelled nucleotides were
obtained from Dupont/NEN.
The reporter plasmid, pHZ18[Delta]2, was constructed as described previously (
28
,
29
). pL101 was constructed as
described in (
30
). Each of the plasmids described below were generated by cloning
Xho
I-digested, gel purified, PCR products into the
Sal
I site of pHZ18[Delta]2. pL601 and pL602 were derived by amplifying a 368 bp H4ARS fragment
from the parent plasmid, YCpMM3, using the primers shown in Table
1
. YCpMM3 is the same as the plasmid YRp14/CEN4/H4ARS described previously (
31
).
The plasmids pL603 and pL604 were created by the same strategy except the
parental plasmid was pVHA74X36. pVHA74X36 was obtained by ligating the L74 and
R36
Bam
HI linker-deletion derivatives (
32
) and subcloning the resulting H4ARS derivative. The 74X36 plasmid was then
subcloned as an
Eco
RI-
Hin
dIII fragment into pVHA (
14
). This 74X36 derivative of the H4ARS contains an internal linker deletion that
removes a 33 bp region that includes the entire ARS consensus and substitutes
in a 10 bp
Bam
HI linker sequence for a net deletion of 23 bp. 74X36 is ars
-
in high frequency transformation assays in YIp5 and pVHA (C. Miller and D.
Kowalski, unpublished results). The plasmids pL701 and pL702 were derived by
inserting a 293 bp fragment obtained by amplification of plasmid YCpMM1 using
the primers shown in Table
1
. YCpMM1 is a derivative of YRp14/CEN4 with a 1.4 kb TRP1ARS1
Eco
RI fragment insertion (C. Miller, unpublished). The plasmids pL703 and pL704
were derived by inserting a 32 bp (5'-GTCGAGAATAATCGTTAAACGAAACTCGACAT-3') oligonucleotide into the test vector. This
oligonucleotide contains a high affinity Abf1p binding site, 5'-AATAATCATGTTAAACGAAA-3', from the
SPT2
gene promoter (
33
). The plasmids pL705 and pL706 are identical to pL703 and pL704 respectively,
with the exception that the oligonucleotide inserted has a double point
mutation in the Abf1p binding site (5'-AATAATCATGTTAAA
GC
AAA-3').
The plasmids pL801 and pL802 were derived by insertion of a 250 bp PCR fragment
containing the ARS305 sequence obtained after amplification of the parent
plasmid p305BP (
23
) with the primers shown in Table
1
. The plasmids pL901 and pL902 were derived by inserting a 110 bp PCR fragment
containing the rDNA ARS, obtained by amplification of the parent plasmid pVHAr8
(
5
) with the primers shown in Table
1
.
The yeast integrating plasmid YIplac128 (
34
) was used to generate the plasmids YIpHZ, YIp701 and YIp801. The parent
plasmids pHZ18[Delta]2, pL701 and pL801 were digested with
Eco
RI and the resulting 6 kb fragment (approximately) was gel purified. This
fragment was ligated to YIplac128 that had also been digested with
Eco
RI to create the YIp derivatives.
The plasmids used to synthesize the RNA probes were derived by cloning the same
PCR fragments described above into the plasmid pT7T319U (Pharmacia). The
orientations were determined by DNA sequencing and the sense and anti-sense RNA probes were synthesized using either T7 or T3 RNA polymerase and
[
32
P]UTP, according to the manufacturer's instructions.
Total RNA was prepared by a modification of a previously described method using
glass beads and hot phenol (
35
). Concentrations were determined spectrophotometrically. Poly(A) RNA was
isolated on oligo (dT) cellulose (Collaborative Biomedical Products). RNA was
separated by electrophoresis on 1.5% agarose-formaldehyde gels and Northern analyses were carried out as described (
35
). Blots were exposed to autoradiographic film with an intensifying screen and
also to a phosphorimaging plate and processed on a Fuji BAS1000 PhosphorImager.
The 3' ends of mRNAs were mapped according to a modification of the methods of
Russo
et al
. (
36
). Briefly, 2 [mu]g of total RNA were used in a reverse transcription reaction containing 50
pmol RT primer (Table
1
), 1 mM each of the dNTPs, 10 U AMV reverse transcriptase, 1* AMV buffer (supplied by manufacturer) and 40 U of RNAsin in a final
reaction volume of 50 [mu]l. The reaction mixture was incubated at 24oC for 5 min and then 1 h at 37oC. For the PCR reaction, 2 [mu]l of the RT products were added to a PCR reaction mixture
containing 20 pmol of each PCR primer (Table
1
), 3 mM MgCl
2
and 2.5 U
Taq
DNA polymerase. A second round of PCR was performed using primers specific to
the individual ARSs (listed in Table
1
). The PCR products were gel purified and digested with
Xho
I and
Kpn
I. DNA fragments were ligated to the vector pT7T319U and sequenced.
Cells were grown on complete media plates without uracil and with galactose.
Plates were overlaid with a 4% molten agar solution containing 0.1% SDS and
0.02% X-gal, and monitored for the development of blue color.
Yeast cultures were grown in complete medium lacking uracil with galactose as a
carbon source to an OD
600
of 1.0. Cells were harvested from 5 ml cultures by centrifugation and crude
extracts were assayed for [beta]-galactosidase activity using ONPG as the substrate (
37
). Protein concentrations were determined by a Bradford assay (BioRad). At least
two colonies from each strain were assayed in duplicate. The results of the
trials are presented as the mean +- standard deviation of [beta]-galactosidase specific activity in nmol/min/mg protein.
In order to assay for sequences that signal transcription termination, a
reporter plasmid was used which allowed us to correlate [beta]-galactosidase activity as a measure of 3' end formation
in vivo
(
30
). The salient features of this plasmid, pHZ18[Delta]2, are shown schematically in Figure
1
. The key component of the vector is a fusion gene which consists of part of the
ribosomal protein (rp) 51A gene fused in frame to the lacZ gene. The fusion
gene is controlled by the GAL upstream activating sequence (UAS). When cells
are grown on galactose the fusion gene is transcribed, the pre-mRNA is spliced and the mRNA translated to produce [beta]-galactosidase.
Table 3
To determine those characteristics of the ARS that cause transcription
termination, we examined the termination phenotype of a mutant version of the
H4ARS [pL603 (+) and pL604 (-)]. The 74X36 mutation consists of
Bam
HI linker insertion/deletion that alters a 33 bp region that includes the entire
ACS, which is responsible for the binding of the large protein complex required
for replication (ORC). Deletion of this sequence completely abolishes ARS
function (C. Miller and D. Kowalski, unpublished). One possible mechanism for
ARSs that act as terminators is that upon ORC binding, a roadblock to
transcription is established, such that the transcription complex cannot
proceed through an ARS sequence, or its movement is somehow impaired. This type
of termination mechanism may be mediated by a specific DNA-protein interaction that has been shown to be operative in the adenovirus
major late promoter transcription unit where the CCAAT binding interaction is
important for termination (
45
). As ORC is thought to be bound tightly to the DNA throughout most, if not all
of the cell cycle, a roadblock mechanism is viable (
46
). However, we found that the mutated H4ARS sequence is fully functional with
regard to transcription termination, despite its inability to function as a
replication origin (Fig.
4
). As shown in Table
3
, there is no difference in [beta]-galactosidase activity when comparing the ars
-
mutant versus the wild-type H4ARS. In addition, the orientation of the ars
-
derivative has no effect on termination function. Thus, an ORC roadblock alone
does not appear to be responsible for transcription termination.
The roadblock mechanism might be mediated by DNA-protein interactions other than ORC. With this in mind, we further
investigated the mechanism of termination function by focusing on ARS1 because
the ARS1 sequence is one of the ARSs that has been extensively analyzed by
linker scanning analysis (
24
). In addition to the 11 bp ACS, there is a region of DNA 3' to the ACS, called the B region, which has been further defined as B1,
B2 and B3 elements. The transcription factor Abf1p binds at the B3 region.
Tanaka and co-workers showed that when ARS1 is placed downstream of a promoter,
transcripts terminated in the B3 region (
3
). This is consistent with a model that suggests a role for Abf1p in the
termination mechanism. Although not all ARSs contain Abf1p binding sites, other
unidentified transcription factor binding sites may be present as a general
feature of ARS composition. Interestingly, the transcription factor, Reb1p, has
been shown to play a role in transcription termination of RNA Pol I (
47
). Therefore the presence of the Abf1p binding site in the ARS sequences led us
to examine the role of Abf1p in transcription termination of RNA Pol II. A 32
bp double-stranded oligomer containing a high affinity Abf1p binding site (see
Methods) was inserted into the intron of pHZ18[Delta]2, to create plasmids pL703 (+) and pL704 (-). After introduction into yeast cells, we determined that the
Abf1p site caused a small reduction in [beta]-galactosidase activity (Table
3
). The slight increase in termination caused by the Abf1 site is probably not
due to binding, as an inactivated version of this construct (pL705 and pL706),
containing a double point mutation in the Abf1p binding site which does not
bind Abf1p (
33
), also showed similar levels of activity (Table
3
). Therefore, although a previous study suggested that a protein-DNA mediated by Abf1p binding may be important for terminating
transcription, we show here that the Abf1 site acts weakly as a terminator in a
manner that is independent of Abf1p binding. From this we conclude that the
Abf1p binding to DNA is insufficient by itself to create a roadblock to
transcription.
The Northern blot data shown in Figure
3
allows us to conclude that termination occurs within the ARSs tested. However,
a more detailed analysis of the polyadenylation site(s) might help elucidate
the relationship between the sequences important for ARS function and the
sequences specifying 3' end formation. To address this issue we examined the polyadenylation
sites of all four ARS sequences, in both the (+) and the (-) orientations. This was accomplished by sequencing cDNA clones derived
from reverse transcription and PCR amplification of RNAs derived from each of
the test strains. The results are depicted in Figure
5
. The polyadenylation sites are heterogeneous. In general, termination occurs 3' to the ACS when the inserts are present in the (-) orientation, but appear to be less specific in the constructs in
which the ARS is in the (+) orientation. However, in all cases poly(A) sites
never map within the ACS itself. In the case of the TRP1ARS1 terminator (pL702)
there is a notable preference for termination within the B3 sequence. This
result confirms the observation of Tanaka
et al.
, and thus validates this method for mapping the 3' ends of mRNAs. No obvious correlation between the sequence elements
defined for ARS305 function (
20
) and poly(A) sites in the pL801 and pL802 constructs is apparent. An additional
aspect of this study points to the heterogeneity of poly(A) site selection,
which has also been observed for other yeast genes (
9
,
42
).
In higher eukaryotes the signal(s) that govern the positioning of the 3' end of an mRNA molecule are simple and well defined. They consist of the
highly conserved hexamer AAUAAA positioned between 10 and 30 nt upstream of the
polyadenylation site, and a less well defined downstream region that is usually
rich in G and/or U residues (for review see
6
,
8
). The signals for the same process in
S.cerevisiae
are less well defined due to the apparent complexity or redundancy of the
sequences involved, although the mechanism of 3' end formation is conserved. In an attempt to define the sequence
characteristics that are important for transcription termination and
polyadenylation, we designed an assay to test the ability of specific sequences
to direct the termination/polyadenylation reaction (
30
). In this study we present evidence that within or near the same sequences that
direct DNA replication, there is also a separable
cis
-acting element (or elements) that directs transcription termination.
The proximity of the sequences that direct DNA replication and transcription
termination is intriguing. At least three explanations for this relationship
are possible. The first is that the placement of the signals found within these
relatively short regions of the DNA is coincidental and the relationship is
insignificant. We consider this unlikely as both sequences can be accomodated
within the space provided in a typical intergeneic region. Although formally we
have not ruled out this explanation,
it is the least interesting of the possibilities, as it lends no insights as to
the nature of either a termination or replication signal. The fact that all
four of the ARSs we tested contain termination activity argues against a chance
co-localization of ARSs and terminators.
The second possibility is that transcription termination and initiation of
replication are functionally interrelated. For example, it has been noted that
ARSs are located only in nontranscribed regions of the chromosome (
12
), suggesting this may be a prerequisite for ARS function. Thus, one way to
ensure that the ARS region remains transcriptionally inactive is to build
termination signals in or around ARSs. It follows that mutations that disturb
termination function should also interfere with ARS function. We have isolated
termination mutants with the ARS305 sequence but do not yet know if these
mutants will have an effect on replication (unpublished results). As
termination signals are generally very ill defined it is difficult to assess
whether ARS associated termination signals will be different from non-ARS terminators. For example, it remains to be established if there are
multiple termination signals or a single signal associated with the ARSs we
have examined [with the exception of the H4ARS, which we know has multiple
termination functions, i.e., those adjacent to the fragment we examined that
are associated with histone H4 transcripts (
38
)]. If terminators are reiterated to create a transcript-free chromatin organization at replication origins then combined
mutagenesis of the redundant terminators to give full transcription through the
ARS may be required to visualize replication defects. Interestingly, a similar
model of multiple termination signals surrounding yeast centromeric sequences
has been proposed (
48
), and a transcriptional terminator has been mapped to the promoter region of
the
URA3
gene (
49
). Thus, transcriptional terminators may generally be associated with several
regulatory sequences in yeast.
It is much less likely that ARS function is required to terminate transcription,
as there are far more terminators in the genome than there are ARSs. However,
it is possible that different mechanisms may be responsible for termination at
different sites, and that ARS-associated termination could represent a particular class of terminator.
It is noteworthy that each of the ARSs we have examined terminate transcription
in an orientation-independent fashion. It has been suggested that terminators that operate
in either orientation constitute a class of terminators that is distinct from
those that work in an orientation-dependent manner (
42
,
43
). Thus the idea that the differences between these classes could be due to the
presence or absence of an ARS is intriguing. Our data eliminate the possibility
that ARS function is required for termination, as severe mutations in the ACS
have no apparent effect on termination. In addition, the possibility that a
simple roadblock mechanism is responsible for the termination activity by ORC
or by Abf1p is unlikely. We have not excluded the possibility that Abf1p
binding is necessary, but not sufficient, to cause transcription termination.
This possibility should be considered in light of what is known about Pol I
termination where the transcriptional activator Reb1p plays an important role
in stopping transcription, but Reb1p binding alone does not account for the
termination effect (
47
). In fact, replacement of Reb1p with an unrelated DNA binding protein, the lac
repressor, also contributed to termination of a yeast Pol I transcript
in vitro
(
50
). The authors propose that Pol I termination occurs as a result of two stepwise
events, requiring the polymerase to pause in the proximity of a release
element. Reb1p binding serves as a mediator of the pausing, but the lac
repressor can functionally replace the endogenous termination factor.
The third explanation for the proximity of the ARS/terminator signals is that
the same features required for initiation of replication may also be important
for termination of transcription. For example, the ease of DNA unwinding is an
important feature of an ARS (
22
). It is possible that this same characteristic is important in terminators, and
therefore both types of signals have evolved from the same region of the
chromosome. Why or how unwinding might facilitate termination is unclear,
however several possibilities exist. As the ease of unwinding reflects the
strength of DNA-DNA strand interactions, it may also affect the strength of a potential
RNA-DNA hybrid and/or intramolecular base pairing and RNA secondary structure
formation. The sequence composition, reflected in the overall A-T-richness of the DNA, is not sufficient to direct termination (
9
). Another characteristic of some ARSs is the presence of a sequence which
causes DNA to bend (
51
). In ARS1, a bent DNA sequence is found in the B3 region, which is also the
region that contains the ABF1 binding site (
3
). However, the bending function in ARS1 can be eliminated without affecting
replication efficiency (
24
). As DNA bending has also been implicated in intrinsic termination (
52
), it is possible that this feature is critical in directing termination, but
unimportant in replication. Thus, it is not yet fully clear which
characteristics of the DNA account for triggering either the initiation of
replication or transcription termination in the yeast genome. Further
experiments may reveal which, if any, of the components within these short ARS
sequences we examined are actually utilized to specify a termination signal.
We would like to thank Claire Moore, Arthur Lustig, Dean Dawson and William H.
Baricos for helpful comments on the manuscript. This work was supported by
grants from the NSF (MCB9316701) and the American Cancer Society (JFRA-500) to L.E.H. and to the Tulane Cancer Center for support of S.C.
Plasmid
Terminator
[beta]-gal units
Plate assay
pHZ18[Delta]2
vector
n/a
blue
pL101
ADH 2 3' end
4 +- 1.8 (259 +- 133)
white
pL603
H4ARS mut (+)
0.4 +- 0.3 (249 +-6.3)
white
pL604
H4ARS mut (-)
0.4 +- 0.3 (249 +- 6.3)
white
pL703
ABF1 (+)
186 +- 7.5 (249 +- 6.8)
blue
pL704
ABF1 (-)
176 +- 6.1 (249 +- 6.8)
blue
pL705
ABF1 mut (+)
183 +- 0.6 (249 +-6.8)
blue
pL706
ABF1 mut (-)
190 +- 18 (249 +- 6.8)
blue
REFERENCES
Return