ABSTRACT
The highly expressed mouse histone H2a-614 gene is located 800 nt 5
'
of the histone H3-614 gene. There is a 140 nt sequence located 500 nt from the end of the H2-614 mRNA which has been defined as a transcription termination site
for RNA polymerase II. We established an
in vitro
transcription system in which both 3
'
end processing and transcription termination occur. A template containing the
adenovirus major late promoter, a portion of the histone H2a-614 coding region, its 3
'
processing signal, followed by the transcription termination site was
transcribed in a nuclear extract prepared from mouse myeloma cells. Some of the
transcripts synthesized in the extract were cleaved at the histone processing
site in a reaction which was dependent both on the hairpin binding factor and
the U7 snRNP. The efficiency of histone 3
'
end formation was similar both on synthetic transcripts and transcripts
synthesized by RNA polymerase II. Defined transcripts, which were not processed
and which mapped to the transcription termination site, were released from the
template, suggesting that they were formed by transcription termination. Termination
in vitro
was dependent on a functional histone processing signal.
The final step in transcription of RNA from the DNA template is termination of
transcription and release of both the nascent RNA product and RNA polymerase
from the template. Efficient transcription termination is important for
recycling RNA polymerase molecules and preventing transcription interference
from the upstream run-on transcription. Recent studies have shown that transcription termination
is tightly coupled to 3' end processing (reviewed in ref.
1
). To understand the molecular basis of transcription termination, it is
necessary to have a system in which the various events of RNA metabolism all
occur, starting with transcription from a DNA template. As a start in this
direction, we have established an
in vitro
system capable of both 3' end formation and transcription termination from the mouse histone H2a-614 gene.
The mechanism of termination by RNA polymerase II (pol II) is not well
understood. RNA polymerase II transcribes three classes of genes: (i) genes
encoding the polyadenylated mRNAs, (ii) the replication-dependent histone genes and (iii) the capped small nuclear RNA genes. The
3' end of polyadenylated RNAs is formed by cleavage of the nascent
transcript and transcription continues past the cleavage site. Transcription
termination on genes encoding polyadenylated mRNAs is dependent on the presence
of a functional polyadenylation site (
2
-
6
). In cases where there are two genes which are close together, transcription
must terminate between the two genes to prevent disruption of the transcription
complex on the downstream promoter by a polymerase transcribing the upstream
gene. In several cases where there are two closely positioned genes,
transcription termination sites have been identified (
7
,
8
). For some genes protein factors which bind the sequence required for
transcription termination have been identified (
9
-
12
). In cases where genes are relatively far apart, transcription does not
terminate precisely but rather ends in a broad region 3' of the polyadenylation site (
13
,
14
). The prevailing model is that a polyadenylation site and a transcription pause
site(s) combine to form a complete termination site (
15
,
16
). Thus cleavage at the polyadenylation site, followed by degradation of the free 5' end of the nascent RNA by a 5' -> 3' nuclease, is a critical step in transcription
termination (
1
).
Histone mRNAs are the only mRNAs which do not end in a polyA tail (
17
). The 3' end processing signal contains a 16 nt stem-loop followed by a purine-rich U7 snRNP binding site (
18
-
21
). The 3' end of histone mRNAs is formed by a cleavage reaction between the stem-loop and the purine-rich sequence (
22
), with transcription continuing for at least a few hundred nt past the 3' end of the mRNA (
23
,
24
). As in polyadenylated mRNAs, termination of transcription requires a
functional histone 3' processing signal (
24
). The 3' end processing reaction requires a U7 snRNP binding to a purine-rich sequence and a stem-loop binding protein (SLBP) which recognizes the stem-loop (
20
,
25
,
26
).
In one histone gene, the mouse histone H2a-614 gene which is only 800 nt upstream of the H3-614 gene, a transcription termination site has been identified (
24
). This termination region is 140 nt long, GC-rich (75%), and located 550 nt downstream of the processing site of the H2a-614 gene. For this gene `terminated' H2a-614 pre-mRNAs were detected in cells suggesting that cleavage of the nascent transcript is not a prerequisite for
transcription termination (
24
). Here, we report conditions that allow both transcription and processing of a
transcript from a template that contains both a mouse histone H2a-614 3' processing signal and transcription termination site. In addition
to the processed transcripts, we also detected and mapped full-length transcripts which terminated at the termination site as judged by
their release from the supercoiled template.
Mouse myeloma cells were grown in suspension culture in Dulbecco's Modified
Eagle's Medium plus 10% horse serum and harvested at a concentration of 4-6 * 10
5
cells/ml. Nuclei were prepared essentially by the method of Shapiro
et al
. (
27
), as previously described (
28
). The nuclei were extracted with varying salt concentrations ranging from 0.22
M KCl, optimal for histone 3' processing, to 0.35-0.6 M KCl, optimal for
in vitro
transcription. For low salt extracts, where the nuclei did not break and the
chromatin did not swell, the nuclei were removed by centrifugation at 20 000
g
for 30 min. For high salt extracts, the chromatin was removed by centrifugation
at 100 000
g
for 1 h. The resulting supernatant was dialyzed against 20 mM HEPES, pH 7.9,
20% glycerol, 100 mM KCl, 0.2 mM EDTA and 0.5 mM DTT. Precipitated material was
removed by centrifugation, and the supernatant was stored at -80oC in small aliquots. Typical protein concentrations were 4-6 mg/ml.
The HLST gene contains the adenovirus major late promoter (MLP) fused to a
portion of the mouse histone H2a-614 gene containing the 3' processing signal, followed by the transcription termination
region (Fig.
2
A). The genes HLT, HLST
M1
and HLST
M2
were constructed from the HLST gene by substituting the appropriate terminator
mutations (HLST
M1
and HLST
M2
genes) or by deleting the 92 nt starting at the
Sac
I site in the 3' UTR and extending site past the U7 snRNP binding site (HLT gene; Fig.
2
A). Either supercoiled or linear templates cleaved 382 nt from the MLP start
site were used. Transcription was carried out in a 20 [mu]l reaction containing 1 [mu]g template DNA, 15 [mu]l (75 [mu]g protein) nuclear extract, 7.5 mM MgCl
2
,
1 mM ATP, CTP, GTP, 10 [mu]M UTP and 10 [mu]Ci [[alpha]-
32
P]UTP (3000 Ci/mmol) and 30 [mu]g yeast tRNA at 25oC. After addition of 1% SDS and 10 mM EDTA to stop the reaction, RNA
was prepared by extraction with phenol and recovered by precipitation with
ethanol. The RNA products were resolved by electrophoresis on 6% polyacrylamide-7 M urea gels and detected by autoradiography and quantified on a
PhosphorImager (Molecular Dynamics).
To map the transcripts antisense oligonucleotides (1 [mu]g) were included in the reaction. These oligonucleotides result in cleavage
of the transcripts at specific sites by the endogenous RNase H in the extract (
29
).
A 322 nt substrate containing a portion of the H2a-614 coding region, the 3' end and processing signal and 60 nt of flanking sequence was
synthesized using T7 RNA polymerase as previously described (
28
). Reactions were carried out in a final volume of 20 [mu]l containing 17 [mu]l nuclear extract (85 [mu]g protein), 10 000 c.p.m. pre-mRNA transcript (3 * 10
4
c.p.m./[mu]g; 0.33 pmol) and 20 mM EDTA. The reactions were incubated at 32oC for 30 min and were stopped by the addition of NaOAc and extracted with phenol.
The RNA was recovered by ethanol precipitation and analyzed on a 6%
polyacrylamide-7 M urea gel. The RNA was detected by autoradiography and quantified using a PhosphorImager (Molecular Dynamics).
To determine whether the transcript was released from the template, the reaction
was adjusted to 0.5 M NaCl and applied to a 1.5 ml (0.6 * 30 cm) Agarose A15M (BioRad) column equilibrated with 0.5 M NaCl, 10 mM
HEPES, pH 7.2, 0.1 mM EDTA in a 2 ml disposable pipette. Five drop fractions
were collected. RNA was prepared from each fraction by extraction with phenol-chloroform, recovered by ethanol precipitation, and resolved by
electrophoresis on a 6% polyacrylamide-7 M urea gel.
The synthesis of histone mRNA is a much simpler process than the synthesis of
most mRNAs. Once the transcript has been initiated, there are only two
additional steps: cleavage of the nascent transcript to produce the mature mRNA
and termination of the RNA transcript releasing the RNA polymerase. Conditions
for efficient processing of synthetic histone pre-mRNAs have been developed (
21
,
30
). However, the reaction proceeds
in vitro
optimally in the absence of Mg
2+
, conditions incompatible with transcription. We sought to develop conditions
which were compatible with both transcription by RNA polymerase II and histone
3' end formation.
Previously we developed an efficient
in vitro
processing system from mouse myeloma cell nuclei by extracting the nuclei with
0.22 M KCl. Higher concentrations of KCl resulted in extracts with diminished
processing activity (
28
). The extracts prepared at 0.22 M KCl were not active in transcription (data
not shown). However, extraction of nuclei at a higher salt concentration, 0.35-0.6 M KCl, resulted in extracts that were active in transcription by RNA
polymerase II, and were less active in histone pre-mRNA processing.
We tested the various extracts for the ability to cleave histone pre-mRNAs, and also tested to see whether conditions compatible with transcription and the cleavage reaction could be obtained. The
histone processing reaction occurred efficiently in the low salt extract in the
absence of Mg
2+
(Fig.
1
A, lane 2), but there was only a small amount of processing in a high salt
extract (Fig.
1
A, lane 1) using the same conditions. To improve the processing efficiency
without reducing the transcription activity, we mixed the processing extract
and the transcription extract. The optimal ratio which allowed both processing and transcription at reasonable rates was a 1:3 ratio of processing extract to transcription extract. When the
two extracts were mixed, there was an intermediate processing activity in the
absence of divalent cations (Fig.
1
A, lane 3). However, when the reaction was carried out in the presence of Mg
2+
(Fig.
1
A, lane 4) or in the presence of Mg
2+
and ribonucleotide triphosphates (Fig.
1
A, lane 5), there was very little accumulation of the processed mRNA. Chelation
of the Mg
2+
by adding EDTA restored the processing activity (Fig.
1
A, lane 6). Thus, under normal conditions for
in vitro
transcription the histone pre-mRNA cleavage reaction was very inefficient.
To determine whether we could transcribe and process a histone pre-mRNA
in vitro
, we constructed the HLST gene, which contains the adenovirus major late
promoter, a portion of the histone H2a-614 coding region, 3' end and a 140 nt region necessary for transcription termination
(Fig.
2
A). A processed histone mRNA produced from this gene would be 155 nt long, and
an RNA terminated at the site where transcription termination was mapped
in vivo
would be ~320 nt long. The sequence of the terminator region and the antisense
oligonucleotides used to map the transcripts are also shown in Figure
2
A. The HLST gene was used as a template in the
in vitro
reactions either as a linear template ending 382 nt (Fig.
2
B) from the transcription start site, or as a supercoiled template (Figs
3
-
4
).
Figure
Figure
When the gene was linearized at a site 382 nt from the transcription start site
and incubated in either the transcription extract (Fig.
2
B, lane 1) or in the mixed transcription and processing extract (Fig.
2
B, lane 2), several products accumulated. There was a major run-off product (labeled RO), and several distinct smaller products formed,
including two (labeled T) which migrated in the region expected for termination
products. In addition, there was a clear band at about the size expected for
RNAs processed at the histone 3' end (SL
-5
) in the presence of Mg
2+
(Fig.
2
B). Synthesis of all of the RNAs was inhibited by 2 [mu]g/ml [alpha]-amanitin demonstrating that they were synthesized by RNA
polymerase II (data not shown). We precisely mapped the SL
-5
transcript using antisense oligonucleotides (Fig.
2
C). When the transcription and processing extracts were mixed, there was a
reduction in the transcription activity but an increase in the amount of the
processed RNA formed (Fig.
2
B, compare lane 2 with lane 1), consistent with the better processing efficiency
of these extracts. It is difficult to estimate the processing efficiency in
these reactions since many of the products are heterogeneous, but the
efficiency was certainly not higher on the transcripts formed
in vitro
, even taking into account that the uniformly labeled run-off transcript contains about three times more radioactivity per mole than
the 155 nt processed RNA. Thus there is no indication that transcripts
synthesized in the extract by RNA polymerase II were processed any better than
synthetic RNAs which were added to the extract.
Several experiments demonstrated that the SL
-5
product ended at the end of the stem-loop and was formed by the histone mRNA processing reaction. Short
transcripts can also result from the RNA polymerase stalling during
transcription, often at sites of potential secondary structure in the template
or RNA product. To demonstrate that some of the SL
-5
product was formed by histone 3' processing, we used inhibitors of the two
trans
-acting factors involved in histone pre-mRNA processing, the hairpin binding factor (HBF, which binds to the
stem-loop) and U7 snRNP (which base-pairs with the downstream cleavage site). Formation of the SL
-5
product was partially inhibited by addition of a 30 nt RNA containing the wild-type stem-loop sequence (Fig.
2
B, lane 6). This RNA binds to the hairpin binding factor and reduces histone mRNA
processing by ~90% in these extracts (
28
). The formation of the other transcripts was not affected by this competitor
RNA, consistent with it having a specific effect of RNA processing. Formation
of the SL
-5
product was not blocked by a 30 nt RNA with the stem sequence reversed (Fig.
2
, lane 7), and this RNA does not bind the hairpin factor or block processing (
28
).
The histone 3' processing reaction is absolutely dependent on U7 snRNP and antisense
oligonucleotides to the 5' end of U7 snRNP completely abolish processing (Fig.
1
C). Formation of the SL
-5
transcript was partially inhibited by an antisense oligonucleotide to the 5' end of U7 snRNA (Fig.
2
C, compare lanes 1 and 4), while the overall level of transcription was not
affected. Thus some of the SL
-5
transcripts were formed by processing (50-80% in different experiments) and some were presumably formed by
exonuclease activity, particularly in cases where the transcripts were first
cleaved by RNase H, as in Figure
2
C. It is also possible that some of the SL
-5
product was a result of RNA polymerase II stalling at the stem-loop. However, the production of >50% of the SL
-5
product was dependent on both the hairpin binding factor and the U7 snRNP, and
hence was formed by the histone mRNA processing reaction.
To map the 3' end of the SL
-5
transcript, we used antisense oligonucleotides. Previously we have shown that
inclusion of antisense nucleotides in the transcription reaction results in
cleavage of the
in vitro
transcripts by RNase H and accumulation of a single product from all
transcripts which extended beyond the position where the antisense
oligonucleotide hybridized to the RNA transcript (
29
). Thus inclusion of oligonucleotide A (shown in Fig.
2
A) in the reaction resulted in conversion of all transcripts longer than 255 nt
to a single 255 nt RNA (Fig.
2
C, lane 1). In addition the SL
-5
product accumulated in the reaction. To map these transcripts, we used two
antisense oligonucleotides. Oligonucleotide B extended to the histone pre-mRNA cleavage site 5 nt from the end of the stem-loop and oligonucleotide C extended to the base of the stem (Fig.
2
C). Inclusion of oligonucleotide B in the reaction resulted in accumulation of
two major transcripts, one ending at the histone processing site and one ending
at the base of the stem (Fig.
2
C, lane 2). The latter transcript was probably formed by the same activity in
the extract which trims the processed histone mRNA. Inclusion of
oligonucleotide C resulted in formation of a major product which has the same
mobility as the processed product formed in the extract (Fig.
2
C, lane 3), demonstrating that the SL
-5
product is the product which accumulates in the extract.
Further indication that histone pre-mRNA processing occurs in the extract was provided by examining the
products formed after various times of incubation. In this experiment a
template linearized 682 nt from the transcription start site was used. After
incubation for 10 min, there were run-off transcripts formed and a transcript mapping to the histone pre-mRNA processing site, 5 nt past the end of the stem-loop (Fig.
2
D, lane 2, band SL). After incubation for 30 or 60 min there was an increase in
the overall amount of transcription and in the amount of processed product.
However, after incubation for 30 or 60 min all the `processed' product was the
SL
-5
product (Fig.
2
D, lanes 3 and 4). These results are consistent with accurate processing of the
histone pre-mRNA transcripts synthesized
in vitro
, followed by trimming by a Mg
2+
dependent nuclease in the extract to give the SL
-5
product.
We consistently noted the appearance of transcripts which migrated in the
position expected for transcripts which might have been formed by transcription
termination (Fig.
2
B, bands labeled T). Alternatively these and any other shorter transcripts could
result from the RNA polymerase stalling at defined sites on the template. One
requirement for a transcript formed by termination of transcription is that the
transcript be released from the template. To determine whether any of the
discrete products observed might be a result of transcription termination, we
used supercoiled templates to facilitate the analysis, since the template can
be readily separated from the RNA products.
When a supercoiled template containing the HLST gene was incubated with the
transcription extract, there were several discrete products formed that
migrated between 350 and 450 nt (based on DNA markers) as well as large
heterogeneous transcripts resulting from transcription for various distances
along the supercoiled template. These products were found in different extracts
tested (Fig.
3
A, lanes 1 and 2) and were also present in transcripts from the linear templates
(Fig.
2
B). To map the position of these transcripts we used the antisense
oligonucleotides A, E and F shown in Figure
2
A. Oligonucleotide A is located 5' to the
in vivo
transcription termination site, oligonucleotide E is located right at the
termination site and oligonucleotide F is located 30 nt 3' of the termination site. In the transcription reaction with the
supercoiled template and no added oligonucleotide, there were transcripts which
were the size expected for transcripts extending to near the termination site
(Fig.
3
B, lane 1). Inclusion of oligonucleotide A resulted in cleavage of all of these
transcripts (and the longer heterogeneous transcripts) producing the expected
255 nt RNA (Fig.
3
B, lane 2). Thus all of these transcripts were initiated from the adenovirus
major late promoter. Inclusion of oligonucleotide E resulted in cleavage of the
transcripts yielding a product which was almost identical in mobility to the
major product observed (labeled T, Fig.
3
B, lane 3). Inclusion of oligonucleotide F, which was located at the 3' end of the terminator sequence, had no effect on the transcripts labeled
T, but cleaved all the larger heterogeneous transcripts resulting in the
expected 329 nt product, which were slightly larger than the products labeled T
(Fig.
3
B, lane 4). From the intensity of the two products after cleavage in the
presence of oligonucleotide F, we estimate that 20-30% of the transcripts end at or near the termination site. Note that
since the transcripts are uniformly labeled, there is much more radioactivity
in the long heterogeneous transcripts (Fig.
3
A) than in the terminated transcripts. However, when these long transcripts are
cleaved the 3' fragment is probably rapidly degraded, since uncapped RNAs are rapidly
degraded by a 5' -> 3' exonuclease which is Mg
2+
and ATP-dependent (
37
). The transcripts resulting from cleavage by RNase H and the terminated
transcripts contain about the same amount of radioactivity per molecule. Thus
RNAs were formed in the reaction which were initiated at the major late
promoter and which ended near the expected termination site.
The presence of defined RNA products from the
in vitro
transcription reaction could result from processing of the transcripts,
termination and release of the transcripts or stalling of the RNA polymerase at
defined sites on the template. RNA polymerase II may stall during transcription
and then resume elongation after action of an elongation factor such as TFIIS (
38
-
43
). TFIIS together with RNA polymerase II removes a few nucleotides to allow RNA
polymerase to restart transcription at the pause site (
44
-
46
). A characteristic of a stalled polymerase is that both the polymerase and the
RNA product remain associated with the DNA template. In contrast, terminated
transcripts will be released from the template. Since transcription termination
occurs at specific sequences in the DNA template, we expect most of the RNA
transcripts product to remain associated with the supercoiled templates unless
there is a defined termination site present in the template.
Therefore, to determine whether the RNA products were released from the
template, the reaction was stopped by addition of EDTA, and then immediately
applied to a Biogel A15M column separating the template from the rest of the
reaction. This entire fractionation took less than 10 min. RNA was extracted
from each of the fractions and analyzed by gel electrophoresis. Ten percent of
each fraction was analyzed by agarose gel electrophoresis to locate the DNA
template. RNA was prepared from the remaining 90% of each fraction and analyzed
by electrophoresis. Most of the RNA products eluted from the column with the
DNA template (Fig.
4
, lanes 2 and 3). These included a large number of products smaller than the
expected termination product, indicating that these small transcripts were the
result of stalled polymerases which were still associated with the template
(Fig.
4
, lanes 2 and 3). The major RNAs which were not associated with the template
were the RNAs which ended near the
in vivo
termination site. These were included in the column and eluted in a broad peak
(Fig.
4
, lanes 5-10). We estimate that ~50% of the RNAs ending in the termination region were released from
the template. Since termination may be at least a two-step process, involving first pausing of the RNA polymerase followed by
release of the transcripts and the polymerase, there must be some paused
transcripts which have not been released. The released transcripts are present
presumably as heterogeneous ribonucleoprotein particles accounting for their
broad distribution in the column. There was a small amount of the processed SL
-5
product formed in this reaction and this was also released from the template as
expected (Fig.
4
, lanes 6-9). Thus, in addition to the processed product, an RNA product which
ended at or near the transcription termination site was released from the
template, suggesting that proper transcription termination occurred by some of
the Pol II
in vitro
.
A feature of transcription termination of transcripts from genes encoding mRNAs
is that the termination can only occur after transcription past a functional 3' processing site (
2
,
4
,
6
). This is true for both polyadenylated mRNAs and histone mRNAs (
24
). To determine whether the same requirement exists in this
in vitro
system, we tested the effect of both removal of the 3' processing signal and mutation of the terminator sequence on formation
of the terminated transcripts. Two mutations were made in the terminator
region, one, HLST
M1
, which deletes a portion of the terminator but does not have a significant
effect on termination in cultured cells and the other, HLST
M2
, which alters 9 nt and abolishes termination (our unpublished results). All
four genes were transcribed in the same extract, a combination of the
processing and transcription extracts. A portion of the transcription reaction
containing the HLST gene was treated with the antisense oligonucleotide A which
cleaves the transcripts just prior to the termination site (Fig.
5
, lane 1). There were a series of transcripts in the termination region formed
from the HLST template and from the HLST
M1
template which has a histone processing signal and an active terminator (Fig.
5
, lanes 2 and 4). Only large transcripts were formed from the HLT gene which
lacks the histone processing site (Fig.
5
, lane 3). Since this gene has a deletion, the terminated transcripts would have
been ~200 nt long. A similar result was seen with the HLST
M2
template which lacks a functional terminator (Fig.
5
, lane 5). There were some processed transcripts formed from this template but
there were no transcripts in the region of the terminator; instead all the transcripts extended well past the termination region (Fig.
5
, lane 5). Thus in the
in vitro
system, formation of the transcripts in the termination region requires both a
functional 3' processing signal and an active terminator region.
Figure
Although
in vitro
systems have been developed to allow study of many of the individual reactions
in mRNA biosynthesis, there are few examples of preparations which carry out
multiple reactions in mRNA biosynthesis. The optimal conditions for
in vitro
transcription and splicing are quite different, as are the conditions for
splicing and polyadenylation (
47
). This has made it difficult to develop a single system capable of carrying out
multiple reactions in mRNA biosynthesis. Yet in the cell many of these
reactions are clearly coupled, particularly splicing and polyadenylation (
48
-
50
) and by adjusting conditions appropriately coupling has been demonstrated
in vitro
(
47
,
51
). In the case of histone pre-mRNA only a single cleavage reaction is required to form the histone mRNA
and this cleavage could function to release the nascent transcript from the
template.
A second problem is that different experimental protocols are optimal for
preparation of extracts active in processing and extracts active in
transcription. A higher salt concentration is required to extract essential
components for transcription, than to extract the essential processing
components. In addition, there is an inhibitor of histone pre-mRNA processing released from nuclei at higher salt. It is likely that
this is a non-specific nucleic acid binding activity, since histone pre-mRNA processing is stimulated by the addition of yeast tRNA to the
extract.
In vitro
, histone pre-mRNA processing occurs most efficiently in the presence of high
concentrations of EDTA. In reactions containing Mg
2+
it is difficult to demonstrate histone pre-mRNA processing since there are nucleases present which give products very
similar to the mature histone mRNA. An ATP-dependent 3' -> 5' exonuclease has been described in mammalian nuclear
extracts (
34
) and it is possible that this activity is responsible for some of the U7-independent formation of histone mRNA in the presence of Mg
2+
. The secondary structure of the histone mRNA or the protein-RNA complex at the 3' end of histone mRNA may block the nucleases
in vitro
, resulting in accumulation of histone mRNA-sized molecules. However, by determining that the formation of the
processed transcripts is dependent on U7 snRNP, the transcripts formed by
nuclease activity can be distinguished from the processed transcripts.
Even under conditions in which we obtained processing of the histone transcripts
initiated
in vitro
, the efficiency of processing was not any higher on the transcripts initiated
in vitro
than on synthetic transcripts which were added to the reaction. Therefore in this
in vitro
system, we found no evidence for preferential coupling of 3' end formation to transcription in this
in vitro
system. Similarly in two studies where polyadenylation of transcripts initiated
in vitro
was reported there was not preferential processing of these transcripts (
52
,
53
). Thus, although
in vivo
it is very likely that the cleavage event leading to mRNA 3' end formation occurs on transcripts still bound to the DNA template and
serves to release the transcripts from the template, these
in vitro
studies show no evidence of preferential coupling of cleavage of template bound
transcripts compared with synthetic transcripts.
There have been only a few studies of transcription termination
in vitro
by Pol II using natural termination sites (
52
,
54
-
56
). Hyman and Moore have shown that in a yeast cell-free system, Pol II either terminates or pauses at termination sites
mapped
in vivo
(
57
). Sites of RNA polymerase II termination have also been mapped using purified
enzymes initiated on tailed templates (
58
). Whether these sites represent sites at which the enzyme terminates
in vivo
is not clear. Even in these studies, it was not shown that the terminated
transcripts are released from the DNA template so that termination and pausing
could not be distinguished.
Our results also suggest that the mechanism of transcription termination in
histone genes may be distinctly different from termination in a gene encoding
polyadenylated mRNA. In the proposed model for termination on a gene encoding a
polyadenylated RNA, the transcription complex stalls at a site downstream from
the 3' end processing signal, and the cleavage at the polyA site causes the
release of the nascent transcripts from the DNA template. The RNA polymerase is
released from the template after the associated uncapped RNA fragment is
degraded (
1
). Definitive evidence that cleavage must occur prior to termination has not
been obtained, and the results of one study could be interpreted to suggest
that termination does not require cleavage (
3
). After transcription of histone genes, however, some of the intact terminated
transcripts are released from the DNA template both
in vivo
(
24
) and
in vitro
indicating that a more complex event occurs at the termination site besides
simple stalling.
In vitro
there is a dependence on the histone processing signal and the termination
sequence, consistent with the possibility that the results observed
in vitro
are similar to the steps occurring
in vivo
. This system may provide a starting point for characterizing the factors
required for termination.
We thank Zbigniew Dominski and Mike Whitfield for some of the processing
extracts used in this study. This work was supported by grant GM29832 to W.F.M.



REFERENCES
Return
