Vol. 12, No. 5
MOLECULAR AND CELLULAR BIOLOGY, May 1992, p. 2078-2090
0270-7306/92/052078-13$02.00/0
Copyright © 1992, American Society for Microbiology
Control of Formation of Two Distinct Classes of RNA Polymerase II Elongation Complexes NICHOLAS F. MARSHALL
AND
DAVID H. PRICE*
Department of Biochemistry, University of Iowa, Iowa City, Iowa 52242 Received 7 November 1991/Accepted 13 February 1992
We have examined elongation by RNA polymerase II initiated at a promoter and have identified two classes of elongation complexes. Following initiation at a promoter, all polymerase molecules enter an abortive mode of elongation. Abortive elongation is characterized by the rapid generation of short transcripts due to pausing of the polymerase followed by termination of transcription. Termination of the early elongation complexes can be suppressed by the addition of 250 mM KCI or 1 mg of heparin per ml soon after initiation. Elongation complexes of the second class carry out productive elongation in which long transcripts can be synthesized. Productive elongation complexes are derived from early paused elongation complexes by the action of a factor which we call P-TEF (positive transcription elongation factor). P-TEF is inhibited by 5,6-dichloro-1-B-Dribofuranosylbenzimidazole at concentrations which have no effect on the initiation of transcription. By using templates immobilized on paramagnetic particles, we show that isolated preinitiation complexes lack P-TEF and give rise to transcription complexes which can carry out only abortive elongation. The ability to carry out productive elongation can be restored to isolated transcription complexes by the addition of P-TEF after initiation. A model is presented which describes the role of elongation factors in the formation and maintenance of elongation complexes. The model is consistent with the available in vivo data concerning control of elongation and is used to predict the outcome of other potential in vitro and in vivo experiments.
It is now clear that the transcription of eucaryotic genes is controlled during the elongation phase as well as at initiation. The number of genes for which elongational control has been implicated is growing (80). The three proto-oncogenes c-myc (52, 61, 65, 79, 85), c-myb (4, 62), and c-fos (16, 70) have been shown to be controlled at elongation. Adenovirus (37, 66, 73), simian virus 40 (36, 67), minute virus of mice (39), and human immunodeficiency virus (HIV) (40, 41, 75, 83) have been demonstrated to have specific blocks to transcription elongation. The mRNA levels for the adenosine deaminase genes of humans and mice are at least partly controlled by a regulated block to elongation (13, 14, 42, 48, 58). Many genes in Drosophila melanogaster have RNA polymerase II molecules arrested early after initiation (68, 69). While control of elongation has been implicated in these examples, very little is known about the molecular mechanisms involved. A number of factors that influence elongation and termination by procaryotic RNA polymerase have been defined (86). In particular, the N and Q protein-mediated antitermination systems of lambda and similar bacteriophages provide a model for how specific gene expression can be controlled by modifying RNA polymerase elongation. The mechanism by which lambda Q protein functions has been partially elucidated (87). A key feature in this process is the pausing of the RNA polymerase downstream of the initiation site where Q protein is added to the elongation complex with the aid of another elongation factor, NusA (24, 87). This Q-modified RNA polymerase is then able to read through downstream pause sites and termination sites of both the rhodependent and rho-independent varieties. The lambda N protein antitermination system involves at least six proteins and may proceed by a mechanism similar to but more complex than lambda Q antitermination (51, 86). Unraveling *
this complicated system has involved purifying the individual components, examining the magnitude of their interactiofls, and resolving the multiple steps kinetically (51, 86). The mechanisms cqptrolling the elongation phase of transcription by RNA polymerase II are not understood, but four protein factors that affect the elongation characteristics of RNA: polymerase II haye been identified. The first factor identified, S-II, was orjginally discovered in 'the mouse and has been found to suppress pausing by RNA polymerase II at specific sites (29, 59, 63, 64, 74, 77, 78). The second factor, factor 5 from D. melanogaster (or mammalian TFIFF or RAP 30/74), is required for initiation and also stimulates the elongation rate of RNA polymerase II (5, 9, 19, 56, 57). The third factor, TFIIX, was identified in HeLa cell extract and stimulates elongation by RNA polymerase 11 (5, 38). Finally, a recently identified yeast protein, YES, stimulates the elongation rate of RNA polymerase II (12). Although the biochemical studies listed above suggest a role for the factors during elongation, the details of their involvement are not known. There are a number of cases in which transcription elongation and termination are affected by events that take place before or shortly after initiation from a specific promoter. The Ul and U2 small nuclear RNA (snRNA) genes are good examples of this, as the promoters of these genes generate transcription complexes capable of specific recognition of termination signals (27, 28, 53). When a different promoter that normally generates poly(A)+ RNA is ligated to an snRNA sequence, the termination signal is not recognized (27, 28, 53). In the c-myc gene, the appearance of a block to elongation is dependent on which of the two promoters is used (52, 81). When transcription is initiated at a different promoter, the specific block to elongation is not functional (7). It has been shown that transcription complexes that initiate from several Drosophila promoters are arrested in the synthesis of the first 50 nucleotides (nt) (68, 69). The biochemical details of the promoter proximal effects listed
Corresponding author. 2078
VOL. 12, 1992
CONTROL OF FORMATION OF ELONGATION COMPLEXES
above are not understood, but in the case of the HIV long terminal repeat promoter there is a specific sequence element, which includes sequence from just upstream of the initiation site to approximately +82 downstream, that seems to control the elongation characteristics of polymerases which initiated at the long terminal repeat (60). This sequence includes the TAR element, which in its RNA form confers responsiveness to the transactivator Tat (32, 40, 75, 76). The viral Tat protein, a cellular protein called p68, and other cellular factors are involved in suppressing premature termination after initiation from the HIV promoter (17, 25, 40, 49, 75, 76). A unifying theory explaining all of the promoter proximal effects has not been proposed. The purine nucleotide analog 5,6-dichloro-1-0-D-ribofuranosylbenzimidazole (DRB) has been shown to affect transcription both in vivo (20, 72) and in vitro (15, 89). Originally, DRB was shown to reduce the ratio of long capped transcripts to short ones when added to cells in culture (20, 21, 71, 82). In these studies, it was noted that DRB inhibited heterogeneous nucle'i RNA synthesis by 60 to 75% while at the same time inhibiting mRNA production by >95%. This finding was variously interpreted to mean that DRB inhibits initiation or causes kremature termination (20, 82). DRB has been used to inhibit transcription in vitro in HeLa extract systems, and again, the conclusion that DRB inhibits initiation was reached (88, 89). Most recently, the hypothesis that DRB inhibits the elongation step of transcription through an elongation factor has been suggested (15). One group has described inhibition of casein kinase II by concentrations of DRB identical to the ones used to inhibit transcription in vivo and in vitro (90). It has been proposed that a DRBsensitive kinase phosphorylates either RNA polymerase II or a transcription factor (90). In our effort td understand the mechanism of elongation control, we have begun to study the formation of elongation complexes in vitro, using Drosophila Kc cell nuclear extract. We observed that the majority of RNA polymerase II molecules that initiated did not normally progress to the end of a 460-nt-long template (35a) but rather paused and then terminated close to the promoter. We also found that DRB inhibited the formation of long transcripts. In trying to understand the molecular basis of these observations, we first developed a system using templates immobilized to paramagnetic beads and then used the system to analyze the properties of elongation complexes formed from isolated preinitiation complexes. We were surprised to find that these complexes were defertive in elongation, synthesizing only short transcripts. The ability to synthesize long transcripts was dependent upon the addition of a DRB-sensitive factor, P-TEF (positive transcription elongation factor), which acts after initiation. MATERIALS AND METHODS Materials. [a-32P]CTP and [a-32P]UTP were from ICN. Streptavidin-coated paramagnetic beads (Dynabeads M280) were obtained from Dynal (Great Neck, N.Y.). Ribonucleoside triphosphates were from Pharmacia-LKB Biotechnology. Nucleoside triphosphates (NTPs), Taq polymerase, and other polymerase chain reaction (PCR) supplies were from Perkin-Elmer/Cetus. DNA oligonucleotides were synthesized at the DNA core facility at the University of Iowa. DRB (Sigma) was dissolved in ethanol to 20 mM and stored at 4°C. At each use, the DRB was warmed to room temperature to dissolve any precipitate and then diluted into 3% ethanol and 20 mM N-2-hydroxyethylpiperazine-N'-2-
2079
ethanesulfonic acid (HEPES; pH 7.6) to make a 240 pzM solution. All other chemicals were reagent grade. Cells and cell extracts. Growth of KC cells and preparation of nuclear extracts were basically as described by Price et al. (56). For use in pulse-chase reactions, Kc cell nuclear extract was passed over a fast protein liquid chromatography fastdesalting column (Sephadex G-25). Approximately 700 Ri of nuclear extract was loaded onto an HR 10/10 column (Pharmacia) that had been equilibrated with 120 mM HGKEDP (25 mM HEPES buffer [pH 7.6], 15% glycerol, 120 mM KCI, 0.1 mM EDTA, 1 mM dithiothreitol, 0.1% of a saturated solution of phenylmethylsulfonyl fluoride in isopropanol). One peak fraction which contained about 90% of the extract protein was collected. The protein concentration of the resulting desalted extract is 15.5 mg/ml, compared with 36.0 mg/ml for undesalted extract. DNA templates. The actin 5C (A2) template was prepared as described previously (56). The Yellow template was prepared by digesting plasmid D-2873 (23), which contains the entire Yellow coding sequence plus approximately 3 kb of flanking sequence, with SpeI. Transcription of this DNA gives a runoff RNA of approximately 660 nt, none of which is bacterial sequence. The Kruppel template was prepared by digesting plasmid pKr (33) with EcoRI. Transcription of the linearized plasmid yields a runoff of about 480 nt, of which only the last 25 nt are from bacterial sequence. The H3 template was constructed by cloning the 450-bp TaqI fragment of the aDm 3000 clone (54) into the ClaI restriction site of pBR322. This plasmid was then digested with both AvaI and PstI to yield a fragment of approximately 1 kb which contains the H3 transcription start site and about 135 bp of upstream sequence from between the H3 and H4 promoters. Transcription of this purified fragment yields an 870-nt runoff transcript, most of which is bacterial sequence. The heat shock template is the EcoRI-BglI fragment from plasmid 56H8 (3), which gives a runoff of about 400 nt. PCR template. Template DNA containing the Drosophila actin SC promoter (56) was synthesized by standard PCR techniques (Perkin-Elmer/Cetus). The PCR was performed by using the A2 clone (56) linearized with HindIlI as the template and primers complementary to regions around -200 and +1250, relative the start point of transcription. The sequence of the upstream primer was 5'-CACTCTTl CATGGCGATATAC-3', and the sequence of the downstream primer was 5'-CTGCCTAAAACTCGAATTGG-3'. The upstream primer was modified by coupling a biotin NHS ester (Clontech) to a high-pressure liquid chromatographypurified oligonucleotide containing an amino group on the 5' end. The amino group was derived from the incorporation of 6 -(4-monomethoxytritylamino)hexyl - (2 - cyanoethyl) - (NNdiisopropyl)-phosphoramidite during the original synthesis of the oligonucleotide (Glen Research). Unincorporated biotin reagent was separated from the biotinylated primer by chromatography on Sephadex G-25. Fresh dilutions of PCR primers were made before each reaction. The template when transcribed unmodified yields a runoff of approximately 1,250 nt and when restricted with SalI yields a runoff of approximately 1,000 nt. In vitro transcription. All transcription reactions were performed with the desalted Kc cell nuclear extracts described above and in accordance with a pulse-chase protocol, except where noted. Typical transcription reactions were begun with a 10-min preincubation (8 RI per reaction) of a mixture containing 20 mM HEPES, 5 mM MgCl2, 60 mM KCI, 10 to 40 ,ug of DNA template per ml, and extract. Transcription was initiated by the addition of 2 pI of pulse
2080
MOL. CELL. BIOL.
MARSHALL AND PRICE
solution which contained 1 to 8 ,uCi [a-32P]CTP and brought the reaction mixture to 600 p.M in GTP, ATP, and UTP. The pulse was continued for 15 s, after which the reaction mixture was brought to 1.2 mM cold CTP by the addition of 2 ,ul of chase solution. During the pulse and chase, the buffer conditions remained constant at 20 mM HEPES-5 mM MgCI2-60 mM KCI unless otherwise indicated. After indicated times, the reactions were stopped by the addition of 200 p.l of a Sarkosyl solution (1% Sarkosyl, 100 mM NaCl, 100 mM Tris [pH 8], 10 mM EDTA, and 100 ,ug of tRNA per ml). Sample workup and analysis of labeled transcripts in 6 and 13% gels were done as described previously (56). Some transcription reactions were performed by a continuous labeling protocol, basically as described previously (56). These reaction mixtures contained 20 mM HEPES, 5 mM MgCl2, 600 p.M each GTP, ATP, and UTP, 30 p.M CTP, 60 mM KCI, 1 to 5 ,uCi of [a-32P]CTP, 0 to 60 ,ug of DNA template per ml, and 4 to 5 p.l of undesalted Kc cell nuclear extract in a total volume of 12.5 p.l. A cocktail containing the buffer, NTPs, and MgCl2 was added last to start the reactions. Reaction mixtures were incubated for 20 min at 25°C, and the reactions were stopped as described above. Immobilized template transcription. To begin, the biotinylated DNA was incubated with streptavidin-conjugated Dynabeads M280 (Dynal) in TE (10 mM Tris [pH 7.6], 0.1 mM EDTA) for 30 to 45 minutes (0.5 pmol of 1.5-kb DNA per mg of beads). Beads with bound DNA were concentrated in a microfuge tube using a common bar magnet and then washed with TE five times by repeated resuspension and concentration steps. DNA-beads were then preincubated with extract according the protocol outlined above. At the end of the preincubation, the beads, with preinitiation complexes, were concentrated magnetically and washed two or three times with 60 mM HGKM (20 mM HEPES [pH 7.6], 15% glycerol, 60 mM KCI, 5 mM MgCl2). The beads were then resuspended in a small volume and aliquoted to individual reaction tubes. Transcriptions using washed preinitiation complexes on immobilized templates were done by a pulse-chase protocol similar to the one described above, with the following exceptions. For most reactions, the aliquoted complexes were resuspended with 60 mM HGKM to 8 ,u1, which is equivalent to the unwashed preincubation volume. These reaction mixtures were then pulsed and chased as outlined above. Because of the properties of the paramagnetic beads, it is possible to perform reactions under more concentrated conditions. When larger amounts of extract were to be added after the pulse, washed preinitiation complexes were resuspended in 4 of 60 mM HGKM per reaction. The reaction mixtures were then pulsed with an additional 2 p,l, which brought each reaction mixture to 1.2 mM in ATP, GTP, and UTP while keeping the concentrations of buffers and salt (60 mM KCI) identical. To begin the chase, 12 of solution was added, which brought the reaction mixture to 600 p.M in ATP, GTP, and UTP as well as 1.2 mM in CTP in a final volume of 18 p.l. This large chase volume allows for the addition of a range of amounts of extract without increasing the final size of the reaction mixture by a significant factor. RESULTS Use of various inhibitory agents suggests that multiple forms of elongation complexes exist. Our earlier results suggested that many RNA polymerase II molecules encounter a block to elongation shortly after initiation in vitro and that 250 mM KC1, 1 mg of heparin per ml, or 0.3% Sarkosyl relieves that
15 sec Pulse ORB Preincubation H 10 miri with DRB
60 mM KCI am
0 11
5:20150!501oO
6 rmin Chaise at 60 or 250 rnM KCI
250 mM KCI 1 5 20 50. M DRB Sw|Sz
~~Nuc.
-460
-123
tRNA
*. ...... 'a
A.w .
:20
FIG. 1. DRB inhibition with or without 250 mM KCI. Reactions were carried out with the standard pulse-chase protocol detailed in Materials and Methods with one exception (see diagram above the gel). DRB was added during the preincubation so that the final concentration would be as indicated. The asterisks indicates that the DRB was added to this sample with the chase solution instead of during the preincubation. Transcripts were analyzed in a 13% acrylamide-TBE(Tris-boric acid-EDTA)-urea gel. The positions of DNA markers are indicated at the right. The runoff transcript from the actin 5C template is 460 nt (Nuc.) long. tRNAs which incorporate labeled CTP at the 3' end as a result of nucleotidyltransferase activity in the extract are indicated. ox-am, ac-amanitin.
block and allow polymerase molecules to reach the end of a long template (35a). This means that only a fraction of the polymerase molecules in untreated extracts can generate long RNAs. Since DRB inhibits the production of long transcripts both in vitro and in vivo (15, 20, 72, 89), we tested the subpopulation of RNA polymerase II elongation complexes that normally can make long runoffs for their sensitivities to DRB. An experiment was performed to titrate the effects of DRB on unstimulated and high-salt-stimulated elongation complexes (Fig. 1). In this experiment, the DRB was added to preincubated complexes before initiation in all lanes except the one marked with the asterisk. In all experiments, DRB did not detectably affect initiation, as indicated by the similarity in amount of total RNA synthesized. From the lanes on the left half of Fig. 1, it can be seen that DRB inhibits the production of runoff RNA. These normally productive elongation complexes are inhibited by concentrations of DRB as low as 1 ,uM, and the effect is maximal with 20 ,uM. On the other hand, the complexes that lead to short (30- to 200-nt) RNAs are completely insensitive to concentrations of DRB as high as 50 p.M. The lanes on the right half of the gel demonstrate that the pattern of short RNA does not appear under high-salt elongation conditions. The addition of 250 mM KCl after initiation allows elongation complexes to synthesize longer RNA than they normally do in untreated extracts. These salt-stimulated complexes are resistant to DRB (right half of Fig. 1). The salt-stimulated complexes have a much slower elongation rate than do
VOL. 12, 1992
CONTROL OF FORMATION OF ELONGATION COMPLEXES Preincubation
15 sec
l0min+/-DRB
1pulse1
A
CTi -:P exes 15seJ ITin n washed cein'eLDaaign 60 mM complex not wash washed KCI '-Eh chase
B
O-l0minchase
no addition 20 eM DRB P 0.5 1 3 5 10 0.5 1 3 5 1i0 minutes
Template DNA
soluble immobile 0 30 60 30 60 Pg.ml mM KCI
-
60 250
60 250
2081
20 min
chase
250 mM KCI 60 250
. .. ._ . .. ..-
I41111
tRNA
li W wo-M-6 V'-*
1: :::M
n,
.:,6: ',, .::
.:.
- u
:::,:,i.,,
:f-.
...
I tRNA
FIG. 2. Time course of transcription with and without DRB. Reactions were carried out with the standard pulse-chase protocol detailed in Materials and Methods and diagrammed above the gel. Preincubations were carried out with or without DRB. Aliquots of 12 ,ul were withdrawn from large reaction mixtures and stopped at the indicated times. Transcripts were analyzed in a 13% acrylamideTBE-urea gel. The arrow indicates the position of the 460-nt runoff transcript. P, RNA at end of pulse.
productive complexes owing to the loss of elongation factor function (35a). It is also interesting to note that inhibition of productive elongation complexes is nearly complete when DRB is added with the chase (lane marked with an asterisk in Fig. 1). This finding indicates that the DRB-sensitive step occurs after initiation. To investigate the kinetics of the formation of the pattern of short RNAs and runoff RNA, a time course of synthesis was performed (Fig. 2). Under normal conditions, the formation of the pattern of short RNAs is very rapid, being essentially complete in 30 s (Fig. 2, left lanes). Since the pattern of short transcripts does not change with increasing chase time, we will call these short RNAs abortive transcripts. The appearance of runoff RNA is much slower, taking approximately 5 min to reach its maximum. In the presence of 20 ,uM DRB, the production of runoff is almost completely inhibited (Fig. 2, right lanes). However, the kinetics of accumulation as well as final pattern of the short transcripts are unchanged. Regardless of the presence or absence of DRB, the transcripts are partially degraded by the action of nucleases in the extract. The maximal elongation rate of complexes synthesizing short RNAs is initially very high but decays rapidly to zero. Productive elongation complexes exhibit a maximum elongation rate of about 1,000 nt/min (35a). In this experiment, there is little 460-nt runoff until at least 3 min of transcription. As suggested in the accompanying report (35a), this finding implies the existence of a slow step in the formation of productive elongation complexes.
FIG. 3. Transcription of the PCR template in soluble and immobilized forms. (A) The biotinylated actin 5C promoter template generated by PCR was transcribed by using the indicated amount of DNA either directly or immobilized to Dynabeads. Reactions were carried out by the continuous-labeling protocol described in Materials and Methods. Transcripts were analyzed in a 6% acrylamideTBE-urea gel. (B) Preincubated immobilized templates were transcribed either with or without washing of the preinitiation complexes. Reactions were carried out with the standard pulsechase protocol detailed in Materials and Methods and diagrammed above the gel. Complexes were washed by removing the extract from the magnetically concentrated beads and performing three 200-,ul washes with the indicated KCI concentration in HGKM. The high-salt-washed complexes were washed again with low-salt buffer to allow initiation. Transcripts were analyzed in a 13% acrylamideTBE-urea gel. The arrows point to the 1,250-nt runoff transcript.
Elongation complexes derived from washed preinitiation complexes generate only short transcripts. The use of DNA templates immobilized on a solid support is a valuable tool for the study of initiation of transcription (1, 2, 22). Previous studies have used biotinylated DNA coupled to streptavidinagarose beads. We have developed an improved method in which the template DNA is coupled to paramagnetic particles. This allows the rapid purification of RNA polymerase II transcription complexes. The coupling of the biotinylated DNA to the beads did not interfere with transcription of the DNA by Drosophila Kc cell nuclear extract (Fig. 3A). Preinitiation complexes formed by incubation of DNA-beads with extract were able to initiate with equal efficiency before and after washing (compare the no-chase lanes for the not washed and 60 mM KCI-washed complexes). Note that the highly labeled tRNAs present in the unwashed-complex lanes are completely absent in the lanes showing transcription of washed complexes. Elongation complexes formed from washed preinitiation complexes were not able to proceed to the end of the template when chased at 60 mM KCI. The transcripts visible on the autoradiograph after 20 min of chase do not exceed 400 to 600 nt in length. These transcripts are somewhat longer on average than abortive transcripts detected when
2082
:~ ~ .s}E*f;S.wJ" '
MARSHALL AND PRICE
10 min Preincubation I
Isolation of
complexes
250 mM
no
0 - 10 min Chase Lwith or without additions
|: _r..XS,:., '.w:_. 15 sec
B
15sec
A
Isolation ofPutse 0 - 10 min Chase with or without additions complexes
10 min
Preincubation I
250 mM KCI
no
1 mg/ml
KCI addition heparin P 0.5 1 13 1010.5! 113 10 i.511 31 101minutes;
addition
1 mg/ml eI parin
P'0.511310iO.5 1_30t.5113'tOLniinuies
Nuc.
-492
-
Sh. [:
-246
. .!. .
|wk F .
-123
~ ~ ~ ~ ~ ~ ~ ~ ~ . . . ...:
-738 -615
-492 -369
.;
_
... .... _ m
_
Nuc. -1107
.g x .8.
....
:.: . ...... : * .
MOL. CELL. BIOL.
-246
....... _.
e
MOLECULAR AND CELLULAR BIOLOGY, May 1992, p. 2078-2090
0270-7306/92/052078-13$02.00/0
Copyright © 1992, American Society for Microbiology
Control of Formation of Two Distinct Classes of RNA Polymerase II Elongation Complexes NICHOLAS F. MARSHALL
AND
DAVID H. PRICE*
Department of Biochemistry, University of Iowa, Iowa City, Iowa 52242 Received 7 November 1991/Accepted 13 February 1992
We have examined elongation by RNA polymerase II initiated at a promoter and have identified two classes of elongation complexes. Following initiation at a promoter, all polymerase molecules enter an abortive mode of elongation. Abortive elongation is characterized by the rapid generation of short transcripts due to pausing of the polymerase followed by termination of transcription. Termination of the early elongation complexes can be suppressed by the addition of 250 mM KCI or 1 mg of heparin per ml soon after initiation. Elongation complexes of the second class carry out productive elongation in which long transcripts can be synthesized. Productive elongation complexes are derived from early paused elongation complexes by the action of a factor which we call P-TEF (positive transcription elongation factor). P-TEF is inhibited by 5,6-dichloro-1-B-Dribofuranosylbenzimidazole at concentrations which have no effect on the initiation of transcription. By using templates immobilized on paramagnetic particles, we show that isolated preinitiation complexes lack P-TEF and give rise to transcription complexes which can carry out only abortive elongation. The ability to carry out productive elongation can be restored to isolated transcription complexes by the addition of P-TEF after initiation. A model is presented which describes the role of elongation factors in the formation and maintenance of elongation complexes. The model is consistent with the available in vivo data concerning control of elongation and is used to predict the outcome of other potential in vitro and in vivo experiments.
It is now clear that the transcription of eucaryotic genes is controlled during the elongation phase as well as at initiation. The number of genes for which elongational control has been implicated is growing (80). The three proto-oncogenes c-myc (52, 61, 65, 79, 85), c-myb (4, 62), and c-fos (16, 70) have been shown to be controlled at elongation. Adenovirus (37, 66, 73), simian virus 40 (36, 67), minute virus of mice (39), and human immunodeficiency virus (HIV) (40, 41, 75, 83) have been demonstrated to have specific blocks to transcription elongation. The mRNA levels for the adenosine deaminase genes of humans and mice are at least partly controlled by a regulated block to elongation (13, 14, 42, 48, 58). Many genes in Drosophila melanogaster have RNA polymerase II molecules arrested early after initiation (68, 69). While control of elongation has been implicated in these examples, very little is known about the molecular mechanisms involved. A number of factors that influence elongation and termination by procaryotic RNA polymerase have been defined (86). In particular, the N and Q protein-mediated antitermination systems of lambda and similar bacteriophages provide a model for how specific gene expression can be controlled by modifying RNA polymerase elongation. The mechanism by which lambda Q protein functions has been partially elucidated (87). A key feature in this process is the pausing of the RNA polymerase downstream of the initiation site where Q protein is added to the elongation complex with the aid of another elongation factor, NusA (24, 87). This Q-modified RNA polymerase is then able to read through downstream pause sites and termination sites of both the rhodependent and rho-independent varieties. The lambda N protein antitermination system involves at least six proteins and may proceed by a mechanism similar to but more complex than lambda Q antitermination (51, 86). Unraveling *
this complicated system has involved purifying the individual components, examining the magnitude of their interactiofls, and resolving the multiple steps kinetically (51, 86). The mechanisms cqptrolling the elongation phase of transcription by RNA polymerase II are not understood, but four protein factors that affect the elongation characteristics of RNA: polymerase II haye been identified. The first factor identified, S-II, was orjginally discovered in 'the mouse and has been found to suppress pausing by RNA polymerase II at specific sites (29, 59, 63, 64, 74, 77, 78). The second factor, factor 5 from D. melanogaster (or mammalian TFIFF or RAP 30/74), is required for initiation and also stimulates the elongation rate of RNA polymerase II (5, 9, 19, 56, 57). The third factor, TFIIX, was identified in HeLa cell extract and stimulates elongation by RNA polymerase 11 (5, 38). Finally, a recently identified yeast protein, YES, stimulates the elongation rate of RNA polymerase II (12). Although the biochemical studies listed above suggest a role for the factors during elongation, the details of their involvement are not known. There are a number of cases in which transcription elongation and termination are affected by events that take place before or shortly after initiation from a specific promoter. The Ul and U2 small nuclear RNA (snRNA) genes are good examples of this, as the promoters of these genes generate transcription complexes capable of specific recognition of termination signals (27, 28, 53). When a different promoter that normally generates poly(A)+ RNA is ligated to an snRNA sequence, the termination signal is not recognized (27, 28, 53). In the c-myc gene, the appearance of a block to elongation is dependent on which of the two promoters is used (52, 81). When transcription is initiated at a different promoter, the specific block to elongation is not functional (7). It has been shown that transcription complexes that initiate from several Drosophila promoters are arrested in the synthesis of the first 50 nucleotides (nt) (68, 69). The biochemical details of the promoter proximal effects listed
Corresponding author. 2078
VOL. 12, 1992
CONTROL OF FORMATION OF ELONGATION COMPLEXES
above are not understood, but in the case of the HIV long terminal repeat promoter there is a specific sequence element, which includes sequence from just upstream of the initiation site to approximately +82 downstream, that seems to control the elongation characteristics of polymerases which initiated at the long terminal repeat (60). This sequence includes the TAR element, which in its RNA form confers responsiveness to the transactivator Tat (32, 40, 75, 76). The viral Tat protein, a cellular protein called p68, and other cellular factors are involved in suppressing premature termination after initiation from the HIV promoter (17, 25, 40, 49, 75, 76). A unifying theory explaining all of the promoter proximal effects has not been proposed. The purine nucleotide analog 5,6-dichloro-1-0-D-ribofuranosylbenzimidazole (DRB) has been shown to affect transcription both in vivo (20, 72) and in vitro (15, 89). Originally, DRB was shown to reduce the ratio of long capped transcripts to short ones when added to cells in culture (20, 21, 71, 82). In these studies, it was noted that DRB inhibited heterogeneous nucle'i RNA synthesis by 60 to 75% while at the same time inhibiting mRNA production by >95%. This finding was variously interpreted to mean that DRB inhibits initiation or causes kremature termination (20, 82). DRB has been used to inhibit transcription in vitro in HeLa extract systems, and again, the conclusion that DRB inhibits initiation was reached (88, 89). Most recently, the hypothesis that DRB inhibits the elongation step of transcription through an elongation factor has been suggested (15). One group has described inhibition of casein kinase II by concentrations of DRB identical to the ones used to inhibit transcription in vivo and in vitro (90). It has been proposed that a DRBsensitive kinase phosphorylates either RNA polymerase II or a transcription factor (90). In our effort td understand the mechanism of elongation control, we have begun to study the formation of elongation complexes in vitro, using Drosophila Kc cell nuclear extract. We observed that the majority of RNA polymerase II molecules that initiated did not normally progress to the end of a 460-nt-long template (35a) but rather paused and then terminated close to the promoter. We also found that DRB inhibited the formation of long transcripts. In trying to understand the molecular basis of these observations, we first developed a system using templates immobilized to paramagnetic beads and then used the system to analyze the properties of elongation complexes formed from isolated preinitiation complexes. We were surprised to find that these complexes were defertive in elongation, synthesizing only short transcripts. The ability to synthesize long transcripts was dependent upon the addition of a DRB-sensitive factor, P-TEF (positive transcription elongation factor), which acts after initiation. MATERIALS AND METHODS Materials. [a-32P]CTP and [a-32P]UTP were from ICN. Streptavidin-coated paramagnetic beads (Dynabeads M280) were obtained from Dynal (Great Neck, N.Y.). Ribonucleoside triphosphates were from Pharmacia-LKB Biotechnology. Nucleoside triphosphates (NTPs), Taq polymerase, and other polymerase chain reaction (PCR) supplies were from Perkin-Elmer/Cetus. DNA oligonucleotides were synthesized at the DNA core facility at the University of Iowa. DRB (Sigma) was dissolved in ethanol to 20 mM and stored at 4°C. At each use, the DRB was warmed to room temperature to dissolve any precipitate and then diluted into 3% ethanol and 20 mM N-2-hydroxyethylpiperazine-N'-2-
2079
ethanesulfonic acid (HEPES; pH 7.6) to make a 240 pzM solution. All other chemicals were reagent grade. Cells and cell extracts. Growth of KC cells and preparation of nuclear extracts were basically as described by Price et al. (56). For use in pulse-chase reactions, Kc cell nuclear extract was passed over a fast protein liquid chromatography fastdesalting column (Sephadex G-25). Approximately 700 Ri of nuclear extract was loaded onto an HR 10/10 column (Pharmacia) that had been equilibrated with 120 mM HGKEDP (25 mM HEPES buffer [pH 7.6], 15% glycerol, 120 mM KCI, 0.1 mM EDTA, 1 mM dithiothreitol, 0.1% of a saturated solution of phenylmethylsulfonyl fluoride in isopropanol). One peak fraction which contained about 90% of the extract protein was collected. The protein concentration of the resulting desalted extract is 15.5 mg/ml, compared with 36.0 mg/ml for undesalted extract. DNA templates. The actin 5C (A2) template was prepared as described previously (56). The Yellow template was prepared by digesting plasmid D-2873 (23), which contains the entire Yellow coding sequence plus approximately 3 kb of flanking sequence, with SpeI. Transcription of this DNA gives a runoff RNA of approximately 660 nt, none of which is bacterial sequence. The Kruppel template was prepared by digesting plasmid pKr (33) with EcoRI. Transcription of the linearized plasmid yields a runoff of about 480 nt, of which only the last 25 nt are from bacterial sequence. The H3 template was constructed by cloning the 450-bp TaqI fragment of the aDm 3000 clone (54) into the ClaI restriction site of pBR322. This plasmid was then digested with both AvaI and PstI to yield a fragment of approximately 1 kb which contains the H3 transcription start site and about 135 bp of upstream sequence from between the H3 and H4 promoters. Transcription of this purified fragment yields an 870-nt runoff transcript, most of which is bacterial sequence. The heat shock template is the EcoRI-BglI fragment from plasmid 56H8 (3), which gives a runoff of about 400 nt. PCR template. Template DNA containing the Drosophila actin SC promoter (56) was synthesized by standard PCR techniques (Perkin-Elmer/Cetus). The PCR was performed by using the A2 clone (56) linearized with HindIlI as the template and primers complementary to regions around -200 and +1250, relative the start point of transcription. The sequence of the upstream primer was 5'-CACTCTTl CATGGCGATATAC-3', and the sequence of the downstream primer was 5'-CTGCCTAAAACTCGAATTGG-3'. The upstream primer was modified by coupling a biotin NHS ester (Clontech) to a high-pressure liquid chromatographypurified oligonucleotide containing an amino group on the 5' end. The amino group was derived from the incorporation of 6 -(4-monomethoxytritylamino)hexyl - (2 - cyanoethyl) - (NNdiisopropyl)-phosphoramidite during the original synthesis of the oligonucleotide (Glen Research). Unincorporated biotin reagent was separated from the biotinylated primer by chromatography on Sephadex G-25. Fresh dilutions of PCR primers were made before each reaction. The template when transcribed unmodified yields a runoff of approximately 1,250 nt and when restricted with SalI yields a runoff of approximately 1,000 nt. In vitro transcription. All transcription reactions were performed with the desalted Kc cell nuclear extracts described above and in accordance with a pulse-chase protocol, except where noted. Typical transcription reactions were begun with a 10-min preincubation (8 RI per reaction) of a mixture containing 20 mM HEPES, 5 mM MgCl2, 60 mM KCI, 10 to 40 ,ug of DNA template per ml, and extract. Transcription was initiated by the addition of 2 pI of pulse
2080
MOL. CELL. BIOL.
MARSHALL AND PRICE
solution which contained 1 to 8 ,uCi [a-32P]CTP and brought the reaction mixture to 600 p.M in GTP, ATP, and UTP. The pulse was continued for 15 s, after which the reaction mixture was brought to 1.2 mM cold CTP by the addition of 2 ,ul of chase solution. During the pulse and chase, the buffer conditions remained constant at 20 mM HEPES-5 mM MgCI2-60 mM KCI unless otherwise indicated. After indicated times, the reactions were stopped by the addition of 200 p.l of a Sarkosyl solution (1% Sarkosyl, 100 mM NaCl, 100 mM Tris [pH 8], 10 mM EDTA, and 100 ,ug of tRNA per ml). Sample workup and analysis of labeled transcripts in 6 and 13% gels were done as described previously (56). Some transcription reactions were performed by a continuous labeling protocol, basically as described previously (56). These reaction mixtures contained 20 mM HEPES, 5 mM MgCl2, 600 p.M each GTP, ATP, and UTP, 30 p.M CTP, 60 mM KCI, 1 to 5 ,uCi of [a-32P]CTP, 0 to 60 ,ug of DNA template per ml, and 4 to 5 p.l of undesalted Kc cell nuclear extract in a total volume of 12.5 p.l. A cocktail containing the buffer, NTPs, and MgCl2 was added last to start the reactions. Reaction mixtures were incubated for 20 min at 25°C, and the reactions were stopped as described above. Immobilized template transcription. To begin, the biotinylated DNA was incubated with streptavidin-conjugated Dynabeads M280 (Dynal) in TE (10 mM Tris [pH 7.6], 0.1 mM EDTA) for 30 to 45 minutes (0.5 pmol of 1.5-kb DNA per mg of beads). Beads with bound DNA were concentrated in a microfuge tube using a common bar magnet and then washed with TE five times by repeated resuspension and concentration steps. DNA-beads were then preincubated with extract according the protocol outlined above. At the end of the preincubation, the beads, with preinitiation complexes, were concentrated magnetically and washed two or three times with 60 mM HGKM (20 mM HEPES [pH 7.6], 15% glycerol, 60 mM KCI, 5 mM MgCl2). The beads were then resuspended in a small volume and aliquoted to individual reaction tubes. Transcriptions using washed preinitiation complexes on immobilized templates were done by a pulse-chase protocol similar to the one described above, with the following exceptions. For most reactions, the aliquoted complexes were resuspended with 60 mM HGKM to 8 ,u1, which is equivalent to the unwashed preincubation volume. These reaction mixtures were then pulsed and chased as outlined above. Because of the properties of the paramagnetic beads, it is possible to perform reactions under more concentrated conditions. When larger amounts of extract were to be added after the pulse, washed preinitiation complexes were resuspended in 4 of 60 mM HGKM per reaction. The reaction mixtures were then pulsed with an additional 2 p,l, which brought each reaction mixture to 1.2 mM in ATP, GTP, and UTP while keeping the concentrations of buffers and salt (60 mM KCI) identical. To begin the chase, 12 of solution was added, which brought the reaction mixture to 600 p.M in ATP, GTP, and UTP as well as 1.2 mM in CTP in a final volume of 18 p.l. This large chase volume allows for the addition of a range of amounts of extract without increasing the final size of the reaction mixture by a significant factor. RESULTS Use of various inhibitory agents suggests that multiple forms of elongation complexes exist. Our earlier results suggested that many RNA polymerase II molecules encounter a block to elongation shortly after initiation in vitro and that 250 mM KC1, 1 mg of heparin per ml, or 0.3% Sarkosyl relieves that
15 sec Pulse ORB Preincubation H 10 miri with DRB
60 mM KCI am
0 11
5:20150!501oO
6 rmin Chaise at 60 or 250 rnM KCI
250 mM KCI 1 5 20 50. M DRB Sw|Sz
~~Nuc.
-460
-123
tRNA
*. ...... 'a
A.w .
:20
FIG. 1. DRB inhibition with or without 250 mM KCI. Reactions were carried out with the standard pulse-chase protocol detailed in Materials and Methods with one exception (see diagram above the gel). DRB was added during the preincubation so that the final concentration would be as indicated. The asterisks indicates that the DRB was added to this sample with the chase solution instead of during the preincubation. Transcripts were analyzed in a 13% acrylamide-TBE(Tris-boric acid-EDTA)-urea gel. The positions of DNA markers are indicated at the right. The runoff transcript from the actin 5C template is 460 nt (Nuc.) long. tRNAs which incorporate labeled CTP at the 3' end as a result of nucleotidyltransferase activity in the extract are indicated. ox-am, ac-amanitin.
block and allow polymerase molecules to reach the end of a long template (35a). This means that only a fraction of the polymerase molecules in untreated extracts can generate long RNAs. Since DRB inhibits the production of long transcripts both in vitro and in vivo (15, 20, 72, 89), we tested the subpopulation of RNA polymerase II elongation complexes that normally can make long runoffs for their sensitivities to DRB. An experiment was performed to titrate the effects of DRB on unstimulated and high-salt-stimulated elongation complexes (Fig. 1). In this experiment, the DRB was added to preincubated complexes before initiation in all lanes except the one marked with the asterisk. In all experiments, DRB did not detectably affect initiation, as indicated by the similarity in amount of total RNA synthesized. From the lanes on the left half of Fig. 1, it can be seen that DRB inhibits the production of runoff RNA. These normally productive elongation complexes are inhibited by concentrations of DRB as low as 1 ,uM, and the effect is maximal with 20 ,uM. On the other hand, the complexes that lead to short (30- to 200-nt) RNAs are completely insensitive to concentrations of DRB as high as 50 p.M. The lanes on the right half of the gel demonstrate that the pattern of short RNA does not appear under high-salt elongation conditions. The addition of 250 mM KCl after initiation allows elongation complexes to synthesize longer RNA than they normally do in untreated extracts. These salt-stimulated complexes are resistant to DRB (right half of Fig. 1). The salt-stimulated complexes have a much slower elongation rate than do
VOL. 12, 1992
CONTROL OF FORMATION OF ELONGATION COMPLEXES Preincubation
15 sec
l0min+/-DRB
1pulse1
A
CTi -:P exes 15seJ ITin n washed cein'eLDaaign 60 mM complex not wash washed KCI '-Eh chase
B
O-l0minchase
no addition 20 eM DRB P 0.5 1 3 5 10 0.5 1 3 5 1i0 minutes
Template DNA
soluble immobile 0 30 60 30 60 Pg.ml mM KCI
-
60 250
60 250
2081
20 min
chase
250 mM KCI 60 250
. .. ._ . .. ..-
I41111
tRNA
li W wo-M-6 V'-*
1: :::M
n,
.:,6: ',, .::
.:.
- u
:::,:,i.,,
:f-.
...
I tRNA
FIG. 2. Time course of transcription with and without DRB. Reactions were carried out with the standard pulse-chase protocol detailed in Materials and Methods and diagrammed above the gel. Preincubations were carried out with or without DRB. Aliquots of 12 ,ul were withdrawn from large reaction mixtures and stopped at the indicated times. Transcripts were analyzed in a 13% acrylamideTBE-urea gel. The arrow indicates the position of the 460-nt runoff transcript. P, RNA at end of pulse.
productive complexes owing to the loss of elongation factor function (35a). It is also interesting to note that inhibition of productive elongation complexes is nearly complete when DRB is added with the chase (lane marked with an asterisk in Fig. 1). This finding indicates that the DRB-sensitive step occurs after initiation. To investigate the kinetics of the formation of the pattern of short RNAs and runoff RNA, a time course of synthesis was performed (Fig. 2). Under normal conditions, the formation of the pattern of short RNAs is very rapid, being essentially complete in 30 s (Fig. 2, left lanes). Since the pattern of short transcripts does not change with increasing chase time, we will call these short RNAs abortive transcripts. The appearance of runoff RNA is much slower, taking approximately 5 min to reach its maximum. In the presence of 20 ,uM DRB, the production of runoff is almost completely inhibited (Fig. 2, right lanes). However, the kinetics of accumulation as well as final pattern of the short transcripts are unchanged. Regardless of the presence or absence of DRB, the transcripts are partially degraded by the action of nucleases in the extract. The maximal elongation rate of complexes synthesizing short RNAs is initially very high but decays rapidly to zero. Productive elongation complexes exhibit a maximum elongation rate of about 1,000 nt/min (35a). In this experiment, there is little 460-nt runoff until at least 3 min of transcription. As suggested in the accompanying report (35a), this finding implies the existence of a slow step in the formation of productive elongation complexes.
FIG. 3. Transcription of the PCR template in soluble and immobilized forms. (A) The biotinylated actin 5C promoter template generated by PCR was transcribed by using the indicated amount of DNA either directly or immobilized to Dynabeads. Reactions were carried out by the continuous-labeling protocol described in Materials and Methods. Transcripts were analyzed in a 6% acrylamideTBE-urea gel. (B) Preincubated immobilized templates were transcribed either with or without washing of the preinitiation complexes. Reactions were carried out with the standard pulsechase protocol detailed in Materials and Methods and diagrammed above the gel. Complexes were washed by removing the extract from the magnetically concentrated beads and performing three 200-,ul washes with the indicated KCI concentration in HGKM. The high-salt-washed complexes were washed again with low-salt buffer to allow initiation. Transcripts were analyzed in a 13% acrylamideTBE-urea gel. The arrows point to the 1,250-nt runoff transcript.
Elongation complexes derived from washed preinitiation complexes generate only short transcripts. The use of DNA templates immobilized on a solid support is a valuable tool for the study of initiation of transcription (1, 2, 22). Previous studies have used biotinylated DNA coupled to streptavidinagarose beads. We have developed an improved method in which the template DNA is coupled to paramagnetic particles. This allows the rapid purification of RNA polymerase II transcription complexes. The coupling of the biotinylated DNA to the beads did not interfere with transcription of the DNA by Drosophila Kc cell nuclear extract (Fig. 3A). Preinitiation complexes formed by incubation of DNA-beads with extract were able to initiate with equal efficiency before and after washing (compare the no-chase lanes for the not washed and 60 mM KCI-washed complexes). Note that the highly labeled tRNAs present in the unwashed-complex lanes are completely absent in the lanes showing transcription of washed complexes. Elongation complexes formed from washed preinitiation complexes were not able to proceed to the end of the template when chased at 60 mM KCI. The transcripts visible on the autoradiograph after 20 min of chase do not exceed 400 to 600 nt in length. These transcripts are somewhat longer on average than abortive transcripts detected when
2082
:~ ~ .s}E*f;S.wJ" '
MARSHALL AND PRICE
10 min Preincubation I
Isolation of
complexes
250 mM
no
0 - 10 min Chase Lwith or without additions
|: _r..XS,:., '.w:_. 15 sec
B
15sec
A
Isolation ofPutse 0 - 10 min Chase with or without additions complexes
10 min
Preincubation I
250 mM KCI
no
1 mg/ml
KCI addition heparin P 0.5 1 13 1010.5! 113 10 i.511 31 101minutes;
addition
1 mg/ml eI parin
P'0.511310iO.5 1_30t.5113'tOLniinuies
Nuc.
-492
-
Sh. [:
-246
. .!. .
|wk F .
-123
~ ~ ~ ~ ~ ~ ~ ~ ~ . . . ...:
-738 -615
-492 -369
.;
_
... .... _ m
_
Nuc. -1107
.g x .8.
....
:.: . ...... : * .
MOL. CELL. BIOL.
-246
....... _.
e
