RNA polymerases IIA and IIO have distinct roles during transcription from the TATA-less murine dihydrofolate reductase promoter.

The largest subunit of RNA polymerase II (RNAP II) contains a remarkable region of tandem heptapeptide repeats of the consensus sequence Tyr-Ser-Pro-Thr-Ser-Pro-Ser at its carboxyl terminus. This COOH-terminal domain (CTD) is unphosphorylated in RNAP IIA, extensively phosphorylated in RNAP IIO, and absent in RNAP IIB. The reversible phosphorylation of the CTD has been proposed to be integral to each cycle of transcription from the adenovirus-2 major late promoter. The adenovirus-2 major late promoter, however, may not be a good paradigm for the study of CTD function because in vitro transcription from this promoter is not dependent on the CTD. Previous studies suggest that transcription from the murine dihydrofolate reductase (DHFR) promoter requires the CTD. In an effort to investigate the role of the CTD and its phosphorylation, a RNAP II-dependent reconstituted transcription system specific for the DHFR promoter was established. In this reconstituted system, RNAP IIA, but not RNAP IIB, can transcribe from the DHFR promoter. Furthermore, RNAP IIB does not compete with RNAP IIA for preinitiation complex assembly. These results suggest that the CTD plays a critical role in the recruitment of RNAP II to the DHFR promoter. The analysis of preinitiation complexes assembled on the DHFR promoter indicates that RNAP IIA readily assembles into functional preinitiation complexes in contrast to the inefficient assembly of RNAP IIO. However, transcript elongation is catalyzed by RNAP IIO as demonstrated by the photoactivated cross-linking of nascent DHFR transcripts to subunit IIo. These results indicate that transcription from the DHFR promoter involves the reversible phosphorylation of the CTD and support the idea that RNAPs IIA and IIO have essential but distinct functions.

Pro-Ser, at its COOH terminus which is tandemly repeated 26, 44, and 52 times in yeast, Drosophila, and mammalian cells, respectively (Allison et al., 1985;Corden et al., 1985). This unusual domain is commonly referred to as the CTD for "COOH-terminal domain" and is found in a heavily phosphorylated or an unphosphorylated form in vivo (Kim and Dahmus, 1986). The form of the enzyme containing the phosphorylated CTD is referred to as RNAP I10 with its largest subunit (110) having an apparent molecular weight of 240,000. The form of the enzyme containing the unphosphorylated CTD is referred to as RNAP IIA with its largest subunit (IIa) having an apparent molecular weight of 214,000. A third form of the enzyme, designated RNAP IIB, lacks the CTD and is produced by limited proteolysis during the purification of RNAP 11.
Genetic studies in yeast, Drosophila, and mouse have shown that the CTD is essential in vivo (Nonet et al., 1987;Allison et al., 1988;Bartolomei et al., 1988;Zehring et al., 1988). Avariety of studies have been carried out in a n effort to define the involvement of the CTD in transcription. In yeast, the CTD has been implicated to play a role in transcriptional activation. Truncations of the CTD seriously affect the ability of some upstream activating sequences to mediate the induction of transcription both in vivo and in vitro (Scafe et al., 1990;Liao et al., 1991). Conversely, GAL4 deletions are partially suppressed by the presence of additional repeats within the CTD (Allison and Ingles, 1989). In higher eukaryotes, however, the major late transcription factor and Spl are capable of activating transcription from the viral adenovirus-2 major late promoter (Ad-2 MLP) and a chimeric GC box/HSP 70 Drosophila promoter, respectively, in the absence of the CTD (Zehring and Greenleaf, 1990;Buratowski and Sharp, 1990). In addition, the CTD is not required during in vitro transcription from some promoters such as the Ad-2 MLP and Drosophila actin 5C promoter (Kim and Dahmus, 1989;Zehring et al., 1988). Nevertheless, functional interactions, both direct and indirect, have been demonstrated between the CTD and the TATA-binding protein (Conaway et al., 1992;Koleske et aZ., 1992;Usheva et al., 19921, a negative regulator of transcription encoded by SIN I , and the initiator protein HIP1 (Buermeyer et al., 1992). Although the essential nature of the CTD in vivo does not apply to all promoters within a cell, this does not discount the role of the CTD in transcription from these promoters.
Recent in vitro studies to determine the role of the CTD of RNAP I1 in transcription have led to a model which involves the reversible phosphorylation of the CTD during each transcription cycle. The model proposes that RNAP IIA initially associates with the general transcription factors and the promoter to form a stable preinitiation complex Payne et al., 1989;Laybourn and Dahmus, 1990;Lu et al., 1991;Chesnut et al., 1992). Prior to synthesis of the first phosphodiester bond, RNAP IIA is phosphorylated to RNAP 25033 Role of RNA Polymerases IIA and IIO in DHFR 7Fanscription I10 by a CTD kinase($ which is itself stably associated with the preinitiation complex (Payne et al., 1989;Laybourn and Dahmus, 1990;Arias et al., 1991). RNAP I10 then catalyzes elongation of the transcript (Bartholomew et al., 1986;Cadena and Dahmus, 1987;Payne et al., 1989). A CTD phosphatase, whose activity has recently been identified in HeLa extracts, presumably dephosphorylates RNAP I10 after its release from the template to regenerate RNAP IIA (Chesnut et al., 1992). RNAP IIA can then begin another round of transcription. The observation that CTD phosphorylation occurs at some time during the transition from initiation to elongation suggests that this phosphorylation event could be of regulatory significance.
The RNAP I M I O transcription model discussed above is based primarily on studies utilizing the Ad-2 MLP which does not require the CTD for in vitro transcription (Kim and Dahmus, 1989). Consequently, this promoter is not ideally suited for the analysis of CTD function. Previous studies by Thompson et al. (1989) using HeLa nuclear extracts in which the endogenous RNAP I1 was inactivated by anti-CTD monoclonal antibodies suggest that the CTD is required for transcription from the murine dihydrofolate reductase (DHFR) promoter. The structure of the DHFR promoter is different from that of the Ad-2 MLP in several important aspects. The DHFR promoter does not contain a TATAbox in contrast to the strong consensus TATA sequence present in the Ad-2 MLP. Furthermore, the DHFR promoter comprises four GC boxes, an initiator element containing two overlapping E2F binding sites, and two downstream elements important for regulating and modulating the efficiency of transcription (Dynan et al., 1986;Blake and Azizkhan, 1989;Fanham and Means, 1990;Schmidt et al., 1990). This paper addresses the involvement of the CTD in transcription from the murine DHFR promoter using transcription reactions dependent on the addition of exogenous RNAP 11. The reconstituted transcription system allows for the direct assessment of the role of the CTD in DHFR transcription since the absence/presence and state of phosphorylation of the CTD can be controlled. The studies presented here indicate that the CTD plays a direct role in the recruitment of RNAP I1 to the DHFR promoter. The unphosphorylated CTD is required for preinitiation complex assembly, whereas transcript elongation is catalyzed by the phosphorylated form of RNAP 11. Since the RNAP IIAIIIO phosphorylation cycle appears to be the same in both a TATA-containing and TATA-less promoter, this may be a general mechanism for the transcription of class I1 promoters.

EXPERIMENTAL PROCEDURES
Materials-Ultrapure nucleotides were purchased from Pharmacia LKB Biotechnology Inc. Radiolabeled ribonucleotides [CY-~~PICTP (3000 Ci/mmol) and [y-32PlATP (3000 Ci/mmol) were purchased from Amersham Corp. Wheat germ agglutinin was purchased from Vector Laboratories, Inc. and Sepharose CL-4B was purchased from Sigma. Synthetic oligodeoxyribonucleotides were made at the University of California Davis Protein Structure Laboratory. The Gen-Pak FAX (0.46 x 10 cm) and Protein-Pak Glass DEAE-5PW (10 pm, 0.8 x 7.5 cm) columns were purchased from Waters Chromatography. The alkyl-Superose HR 5/5 and Mono Q HR 5/5 columns were purchased from Pharmacia. The plasmid pSS625 was kindly provided by Robert Schimke (Stanford University).
Preparation of DNA Templates-The plasmid pSS625 containing the murine dihydrofolate reductase gene from positions -356 to +275 ) was digested sequentially with SmaI and Hin-dIII to produce a 651-base pair fragment. This fragment was purified from plasmid DNA on a Gen-Pak FAX column using a 15-ml linear gradient of 0.60-0.75 M NaCl in 25 rn Tris-HC1, pH 7.5, at a flow rate of 0.5 mVmin. The purified DHFR template gives a run-off transcript of 295 nucleotides. The Ad-2 MLP template (positions -260 to +560), prepared as described in Dahmus and Kedinger (19831, was also purified on a Gen-Pak FAX column using the same linear gradient as described above. The Ad-2 MLP template gives a 560-nucleotide run-off tran-Preparation of Banscription Extract-The S-100 transcription extract was prepared from 2 x 1O'O HeLa cells by the method of Weil et al. (1979) as modified by Dahmus and Kedinger (1983). The 5-100 extract was chromatographed on heparin-Sepharose CL-4B as described by Laybourn and Dahmus (1990), and the transcriptional activity was assayed in the 0.6 M KC1 eluted peak (designated HS0.6 extract). The HS0.6 peak fractions were pooled and chromatographed on a DEW-5PW column as described by Laybourn and Dahmus (1990) with the exception that all of the general transcription factors were step-eluted at 0.25 M KC1 (designated DE0.25 extract as in nomenclature described in Chesnut et al. (1992)). RNAP I1 remained bound to the column. The HS0.6 and DE0.25 extracts were dialyzed against Buffer A.
The transcription factor Spl was purified by the method of Jackson and Tjian (1989). Fractions from the DNA affinity column were analyzed by electrophoresis on a 7.5% polyacrylamide-SDS gel. The peak Spl fractions were dialyzed into Buffer B containing 0.1 M KC1 and assayed for transcriptional activity to determine the optimum amount to use in a standard transcription reaction.
Purification of RNAP II-RNAP IIA and RNAP IIB were purified from calf thymus by the method of Hod0 and Blatti (1977) with the modifications described in Laybourn and Dahmus (1990). RNAP IIA was further purified by chromatography on alkyl-Superose using a 15-ml linear gradient of 1.54 M (NH,),SO, in Buffer C at a flow rate of 0.2 mVmin. RNAP IIB was purified further on Mono Q using a 15-ml linear gradient of 0.24-0.39 M (NH,),SO, in Buffer C at a flow rate of 0.4 ml/min. The RNAP IIB fraction used in Fig. 1 was purified on alkyl-Superose as described above prior to purification on Mono &. Each preparation of RNAP I1 was dialyzed against Buffer D and assayed for promoter-independent transcriptional activity (Kim and Dahmus, 1988). The saturating amount of RNAP IIA varied between different DE0.25 extract preparations. Preinitiation complexes were assembled in the absence of ribonucleotides in a volume of 20 pl for 45 min at 24 "C. Following the addition of ribonucleotides (5 pl), reactions were incubated for 15 min at 24 "C. In transcription reactions involving gel filtration-purified preinitiation complexes, ribonucleotides were added and the reactions incubated for 5 min at 24 "C. Sarkosyl was then added to a final concentration of 0.08% and the reaction incubated for an additional 15 min at 24 "C. Reactions were stopped with 125 pl of stop solution (100 rn NaOAc, pH 5.2,0.4% SDS, 0.2 mg/ml Escherichia coli tRNA) and extracted with 150 pl of phenol followed by 150 pl of chlorofonrdisoarnyl alcohol (241). In the transcription experiment shown in Fig. 6, a 147-base pair 32P-labeled DNA fragment was added to the stop solution to serve as an internal control for recovery of 32P-labeled RNA during the extraction steps. The nucleic acids were precipitated from the aqueous phase by the addition of 10 p1 of 5 M NH,OAc and 600 pl of ethanol and analyzed by electrophoresis on a 5% polyacrylamide, 8 M urea gel which was dried and exposed to x-ray film. The transcript bands were quantitated on a Betascope 603 Blot Analyzer (Betagen) or a BAS1000 phosphorimager (Fuji) as indicated.

In
Formation and Fractionation of Preinitintion Complexes Containing 32P-Labeled RNAps IIA and IIO-RNAP IIA was labeled with casein kinase I1 and [y-32PlATP as previously described (Chesnut et al., 1992).
To make 32P-labeled RNAP 110, 3ZP-labeled RNAP IIA was incubated with excess cold ATP and a partially purified fraction of CTD kinase (Chesnut et al., 1992). Both 32P-labeled RNAPs IIA and I10 were purified as previously described (Chestnut et al., 1992), dialyzed into Buffer D and assayed for promoter-independent transcriptional activity (Kim and Dahmus, 1988). script.
[a-32P]CTP (8 Ci/mmol), 600 GTP, 600 p~ UTP, and 200 p~ ATP, 80 Preinitiation complexes were assembled on the DHFR promoter with 32P-labeled RNAP IIA, 110, or a mixture of RNAPs IIA and I10 (milliunits indicated in the figure legends) in a 3 x transcription reaction (total volume was 60 p1) by incubation for 45 min a t 24 "C in the absence of ribonucleotides. The complexes were fractionated from unbound RNAP I1 by gel filtration on a 3.5-ml Sepharose CL4B column equilibrated with Buffer E. Three-drop fractions (150 pl) were collected. One 50-pl aliquot was analyzed directly by electrophoresis on a 5% polyacrylamide-SDS gel which was silver stained (Wray et al., 1981), dried, and exposed to x-ray film. In Fig. 2, a second aliquot was assayed for transcriptional activity as described in the previous section. A third aliquot was assayed for the presence of CTD kinase(s) by incubation with 200 w ATP, dATP, or GTP. The reaction was incubated for 15 min a t 24 "C and analyzed by electrophoresis on a 5% polyacrylamide-SDS gel which was subsequently stained, dried, and exposed to x-ray film.
Photoaffinity Lubeling of RNAP II Subunits with 32P-Labeled RNA " T h e photoaffinity labeling reagent 4-thio-UTP was synthesized as described by Bartholomew et al. (1986) except for the following modification. The reaction mixture (5 ml) was incubated for 60 min a t 37 "C and loaded directly onto a DEAE-5PW column. The 4-thio-u"P was eluted with a 40-ml linear gradient of 0-0.9 M NH40Ac, pH 8.0, a t a flow rate of 0.5 mumin. Care was taken to minimize the exposure of the reagent to light.
The photoaffinity labeling reactions contained the same final buffer conditions as the transcription reactions described above in a total volume of 25 pl. Preincubation reactions contained 12 pl of DE0.25, 30-50 ng of Spl, and 26 milliunits of RNAP IIA or 10 pl of HS0.6 extract. Each reaction containing 80 ng of DHFR template was incubated for 45 min at 24 "C. Ribonucleotides were then added to a final concentration of 5 w [a-32PlCTP (400 Cilmmol), 2.5 p m GTP, 1.0 m 3'-O-methyl-GTP, 200 w I-thio-UTP, and 200 w ATP. The complete transcription reactions were incubated for an additional 15 min at 24 "C. The reaction mixture was irradiated with near ultraviolet light as described by Bartholomew et al. (1986) and loaded onto a 1-ml Sepharose CL4B column equilibrated with Buffer E. Column fractions (50 PI) were treated with DNase I (0.50 unit) for 30 min on ice and analyzed by electrophoresis on a 5% polyacrylamide-SDS gel by silver staining and autoradiography.

RESULTS
The DHFR Promoter Requires the CTD for in Vitro Transcription-In an effort to establish more directly the requirement of the CTD for transcription from the DHFR promoter, a RNAP 11-dependent reconstituted transcription system was developed which was specific for the DHFR promoter. The ability of purified RNAP IIA and RNAP IIB to support transcription from the DHFR promoter was examined. The Ad-2 MLP, which also can be transcribed in this reconstituted system, was included in the same reaction to serve as an internal control. Increasing amounts of either RNAP IIA or RNAP IIB were added to the reconstituted transcription reaction in the presence of the DHFR promoter and the Ad-2 MLP. RNAP IIA was able to transcribe from both promoters (Fig. IA, lanes 3-6). In contrast, RNAP IIB was able to transcribe from the Ad-2 MLP but was unable to transcribe from the DHFR promoter (lanes 7-10). The amount of RNA transcript synthesized by RNAP IIA and RNAP IIB from each of the DNA templates was quantitated and is shown in Fig. lB. This result gives clear and direct evidence that in vitro transcription from the DHFR promoter requires the CTD of RNAP 11. RNAP ZZA Forms a Stable and Functional Preinitiation Complex on the DHFR Promoter-Using the reconstituted transcription reaction, preinitiation complexes were assembled on the DHFR promoter with 32P-labeled RNAP IIA in the absence of ATP and fractionated from free RNAP IIA by gel filtration. The distribution of RNAP IIA in column fractions was determined by SDS-PAGE and autoradiography. Preinitiation complexes containing 32P-labeled RNAP IIA chromatographed in the excluded volume ( Fig. 2 4 , -ATPpanel, lanes 2 3 ) while the uncomplexed RNAP IIA was recovered in the included volume ( Fig. 2A, -ATPpanel, lanes 6-9) (Laybourn and Dahmus, 1990;Chesnut et al., 1992). To assay for transcriptional activity, the ribonucleotides ATP, GTP, UTP, and [a-32P]CTP were added to aliquots of the same fractions shown in the -ATPpanel of Fig.  2 A . Only the excluded fractions which contain preinitiation complexes supported transcription from the DHFR promoter ( Fig. 2 B , lanes 2 3 ) . The included fractions which contain the major portion of RNAP IIA were not transcriptionally active (Fig. 2B, lanes 6-8). These results demonstrate that stable and functional preinitiation complexes containing RNAP IIA as-

Autoradiography
FIG. 3. Experimental scheme for photoaffinity labeling of RNA polymerase. Based on the method of photoaffinity labeling described by Bartholomew et al. (1986), elongation of the DHFR transcript was terminated at various positions. The complexes were irradiated with near ultraviolet light, purified by gel filtration, and analyzed by electrophoresis on a 5% polyacrylamide-SDS gel and autoradiography. semble on the DHFR promoter in the absence of ATP.
To determine whether these preinitiation complexes contain a stably associated CTD kinase(s), fractions collected from reactions carried out in the absence of ATP were incubated with several nucleotides. 32P-Labeled RNAP IIA in DHFR preinitiation complexes was converted to RNAP I10 in the presence of ATP, dATP, or GTP ( Fig. 2A, lanes 2-3 in +ATP, + M P , and +GTP panels). The uncomplexed RNAP IIA contained in the included fractions, however, was not converted to RNAP I10 ( Fig. 2A, lanes 6-9 in +ATP, + M P , and +GTPpanels). Therefore, preinitiation complexes assembled on the DHFR promoter contain an associated CTD kinase(s1 which purifies with the complexes through gel filtration. Furthermore, the CTD kinase(s) can utilize ATP, dATP, or GTP as nucleotide substrates. Fig. 2 demonstrate that RNAP IIA and a CTD kinase(s) assemble into stable and functional preinitiation complexes on the DHFR promoter. In order to determine the phosphorylation state of RNAP I1 involved in elongation, nascent transcripts were cross-linked to RNAP I1 by photoaffinity labeling (Bartholomew et al., 1986). Preinitiation complexes were assembled on the DHFR promoter using either a crude transcription extract (HS0.6) which contains endogenous RNAP I1 or the reconstituted transcription extract in which unlabeled RNAP IIA was added. Ribonucleotides [(u-~~PICTP, ATP, GTP, 3'-O-methyl-GTP, and the photoprobe 4-thio-UTP were added to allow elongation. Elongation complexes were irradiated with near ultraviolet light, purified by gel filtration, and the fractions analyzed by SDS-PAGE and autoradiography as schematically diagrammed in Fig. 3. When the crude transcription extract (HS0.6) was used to form photoaffinity labeled elongation complexes, the 32P-labeled DHFR transcript was found cross-linked to endogenous RNAP subunits 110 and IIc (Fig. 4A, lanes 3 4 1. Utilizing the RNAP 11-dependent reconstituted transcription system, purified RNAP IIA was added (shown in Fig. 4B under Stained Gel) and photoaffinity labeled complexes were isolated and analyzed. In these reactions, nascent DHFR transcripts were again cross-linked to subunits 110 and IIc (Fig. 4B, complete panel, lanes 3-4). Thus, RNAP I10 is the elongating enzyme even in reconstituted transcription reactions in which the input enzyme was RNAP IIA. In control experiments, photoaffinity labeled proteins were not detected in reactions lacking the DHFR template or RNAP I1 (Fig. 4B, right two panels).

RNAP 110 Catalyzes Elongation from the DHFR Promoter "Results presented in
Similarly, no photoaffinity labeling was observed in reactions which contained a-amanitin or which were not irradiated (data not shown). These results clearly indicate that RNAP I10 catalyzes elongation of the DHFR transcript. Similar to the results presented in Fig. 2, RNAP IIA efficiently assembled into preinitiation complexes on the DHFR promoter as shown by the presence of 32P-labeled RNAP IIA in excluded fractions (Fig. 5, ZZA panel, lanes 5-6, and graph below). In the reconstituted reaction containing 32P-labeled RNAP 110, only a minor fraction of RNAP I10 assembled into preinitiation com-plexes as shown by the small amount of 32P-labeled RNA€' I10 found in excluded fractions (Fig. 5, ZZO panel, lane 5). 32P-Labeled RNAP IIA, generated from labeled RNAP I10 by a CTD phosphatase present in the DE0.25 extract, was also detectable in excluded and included fractions (Fig. 5, ZZO panel,   lanes 1,2,5, and 11-13, see graph below). The large difference in the efficiency with which RNAP IIA and RNAP I10 assemble into preinitiation complexes is evident in the reconstituted reaction containing equimolar amounts of 32P-labeled R N A P s IIA and 110. A significantly larger fraction of 32P-labeled RNAP IIA assembled into preinitiation complexes as compared to 32Plabeled RNAP I10 (Fig. 5, ZLAlZZO panel, lanes 5-6, see graph   below). These results indicate that phosphorylation of the CTD hinders the assembly of RNAP I1 into preinitiation complexes on the DHFR promoter. Fig. 1, RNAP IIB is unable to transcribe from the DHFR promoter. To determine whether the absence of the CTD prevents the assembly of RNAP IIB into preinitiation complexes, transcription was camed out in the presence of increasing amounts of RNAP IIB in reactions which contained a limiting amount of RNAP IIA. If RNAP IIB assembled into nonfunctional preinitiation complexes on the DHFR promoter, then transcription by RNAP IIA should diminish in the presence of increasing amounts of RNAP IIB. Conversely, transcription from the Ad-2 MLP catalyzed by a limiting amount of RNAP ILA should increase in the presence of increasing amounts of RNAP IIB. In the reconstituted reaction used for these experiments, the addition of 10 milliunits of RNAP IIA resulted in the maximal amount of transcription from both the DHFR promoter and the Ad-2 MLP (Fig. 6, A and  B, lane 3). The addition of 2.5 milliunits of RNAP IIA gave a level of transcription which was approximately 2-and 4-fold less than the maximal level of transcription obtained from the DHFR promoter and the Ad-2 MLP, respectively (Fig. 6, A and  B , lanes 2 and 6). In agreement with the results presented in Fig. 1, the addition of 2.5 and 25 milliunits of RNAP IIB to reactions containing the DHFR promoter did not result in appreciable transcription from the DHFR promoter ( Fig. 6 A ,   lanes 4 4 ) . The addition of 2.5 and 25 milliunits of RNAP IIB to reactions containing the Ad-2 MLP resulted in a level of transcription which was comparable with that obtained from RNAP IIA (Fig. 6B, compare lanes 2-3 with lanes 4 4 ) . Specific transcription from the DHFR promoter catalyzed by RNAP IIA was not diminished in the presence of increasing amounts of RNAP IIB, even up to a 10-fold excess (Fig. 6 A , lanes 6-10, and 6 0 . Additionally, increasing amounts of RNAP IIB did not compete with 32P-labeled RNAP IIA in preinitiation complex formation as analyzed by gel filtration, SDS-PAGE, and autoradiography (data not shown). These results indicate that RNAP IIB is incapable of assembling into preinitiation. complexes on the DHFR promoter and suggest that the CTD plays a direct role in the recruitment of RNAP I1 to this promoter. In contrast, specific transcription from the Ad-2 MLP catalyzed by RNAP IIA was augmented by the presence of increasing amounts of RNAP IIB (Fig. 6B, lanes 6-10, and 6 0 .

DISCUSSION
The hypothesis that reversible phosphorylation of the CTD plays an integral role in the RNAP I1 transcription cycle has been proposed based on studies of a single promoter, the Ad-2 MLP. In this model, RNAP IIAinitially binds to the promoter to form a preinitiation complex and is subsequently phosphorylated to RNAP I10 by a CTD kinase which may be a basal transcription factor (Payne et al., 1989;Laybourn and Dahmus, 1990;Feaver et al., 1991;Chesnut et al., 1992;Lu et al., 1992;Serizawa et al., 1992). RNAP I10 then catalyzes the elongation of the transcript (Bartholomew et al., 1986;Cadena and Dahmus, 1987;Payne et al., 1989).
The DHFR promoter is involved in the regulation of a cellular, housekeeping gene, and its promoter elements differ significantly from that of the viral Ad-2 MLP. A previous study by Thompson et al. (1989) suggests that transcription from the DHFR promoter requires the CTD of RNAP I1 in contrast to the Ad-2 MLP which had previously been shown to be efficiently transcribed by RNAP IIB (Kim and Dahmus, 1989). in the inhibited nuclear extract. The studies presented here utilize a RNAP 11-dependent reconstituted transcription system to directly assess the ability of RNAP IIA and RNAP IIB to transcribe from the DHFR promoter. In agreement with the results of Thompson et al. (1989) these studies demonstrate that RNAP IIA, but not RNAF' IIB, is capable of transcribing from the DHFR promoter in the reconstituted system. This characteristic of transcription dependence on the CTD, which is lacking in the Ad-2 MLP, makes the DHFR promoter an attractive system for studies on the role of the CTD and its reversible phosphorylation during transcription.
These studies establish that RNAP IIA efficiently assembles into functional preinitiation complexes on the DHFR promoter in the absence ofATP. This is supported by the observation that preinitiation complexes containing RNAP IIA can be purified by gel filtration and can synthesize a run-off transcript upon the addition of ribonucleotides. In contrast, RNAP I10 is inefficient in assembling into preinitiation complexes on the DHFR promoter. Therefore, a n unphosphorylated CTD is required for the efficient interaction of RNAP I1 with the preinitiation complex.
The observation that increasing amounts of RNAP IIB do not inhibit transcription from the DHFR promoter in reactions containing a limiting amount of RNAP IIA indicates that RNAP IIB cannot compete with RNAP IIA for assembly into preinitiation complexes. Consequently, the inability of RNAP IIB to transcribe from the DHFR promoter must stem from its failure to assemble into preinitiation complexes. This result rules out the possibility that RNAP IIB is stalled at some point in initiation or early in transcript elongation and supports a direct role for the CTD in the recruitment of RNAP I1 to the DHFR promoter.
The CTD requirement observed in the association of RNAP I1 with DHFR preinitiation complexes is not observed in the case of the Ad-2 MLP. This is supported by the observation that transcription from the Ad-2 MLP in the presence of a limiting amount of RNAP IIA increases with the addition of increasing amounts of RNAP IIB. It is interesting, however, that when the CTD is present in RNAP 11, as it is in vivo, it must be in the unphosphorylated form in order to assemble efficiently into preinitiation complexes on the Ad-2 MLP (Lu et al., 1991;Chesnut et al., 1992). Recent studies demonstrate that not all promoters use the same set of "general" basal transcription factors (Parvin et al., 1992). Thus, it is possible that each promoter assembles its own distinct array of general and specific transcription factors involving different protein-protein and protein-DNA interactions. The preinitiation complex that assembles on the Ad-2 MLP with RNAP IIB may be stable due to strong interactions which can compensate for the lack of the CTD. On the other hand, the stability of the preinitiation complex that assembles on the DHFR promoter depends on the interactionb) of the CTD with components of that complex. Preinitiation complexes assembled on the DHFR promoter also contain a stably associated CTD kinase activity which is able to utilize ATP, dATP, or GTP as nucleotide substrates in the conversion of RNAP IIA to 110. This IIA to I10 conversion is significant in that RNAP I10 subunits are found cross-linked to the nascent DHFR transcript, thus implicating RNAP I10 as the elongating form of the enzyme in DHFR transcription. It is presumed that RNAP I10 dissociates from the template upon termination and is dephosphorylated to RNAP IIA in order to begin another round of transcription. These results establish that phosphorylation of the CTD occurs during transcription from the DHFR promoter and suggest that this may be a common feature in the transcription of class I1 promoters.
Recent experiments to address the obligatory nature of CTD phosphorylation during the transcription cycle utilized the protein kinase inhibitor N-[2-(methylamino)ethyl]-5-isoquinoline sulfonamide dihydrochloride (H-8) to inhibit CTD kinase activity (Serizawa et al., 1993). In the presence of H-8 concentrations which inhibited CTD phosphorylation by more than 99%, basal transcription from the Ad-2 MLP was virtually unaffected. These results suggest that CTD phosphorylation is not a necessary step during the transition from initiation to elongation. However, the possibility cannot be excluded that phosphorylation of a small percentage of RNAP I1 was not inhibited in these reactions and that this was responsible for the transcription observed both in the absence and presence of H-8. A more definitive experiment would be to photocross-link the nascent transcript to the elongating enzyme in the presence of H-8. Additionally, the interpretation of these H-8 inhibitor experiments is overshadowed by the fact that in vitro transcription from the Ad-2 MLP does not require the CTD. It will be of interest to establish whether or not transcription from the DHFR promoter, which has a dependence on the CTD, is sensitive to H-8.
A CTD kinase(s) is a component of preinitiation complexes assembled on both the DHFR promoter and the Ad-2 MLF' (Payne et al., 1989;Laybourn and Dahmus, 1990;Arias et al., 1991). Two distinct kinase activities, CTDKl and CTDK2, have been partially purified and characterized from HeLa transcription extracts (Payne and Dahmus, 1993). Additionally, other laboratories have identified CTD kinases such as TFIIH (BTF2) (Lu et al., 1992) and the template-associated CTD kinase (Arias et al., 1991) from human, factor S from rat (Serizawa et al., 19921, cdc2 kinase from mouse (Cisek and Corden, 19891, and factor b and CTKl from yeast (Feaver et al., 1991;Lee and Greenleaf, 1991). TFIIH, 6, and factor b have also been identified as basal transcription factors. The multiplicity of CTD kinases identified in in vitro studies suggests that multiple CTD kinases may function in vivo. TATA-containing and TATA-less promoters such as the Ad-2 MLP and DHFR promoter, respectively, could in principle assemble preinitiation complexes which contain distinct CTD kinases. In a broader sense, different promoters may bind different types of CTD kinases. Further studies are necessary to establish whether a single CTD kinase, as opposed to multiple CTD kinases, is involved in the transcription of class I1 promoters.
What specific interactions might serve to direct a promoterspecific CTD kinase to a promoter? A recent study demonstrated that during S phase, the E2F transcription factor is in a complex with cyclin A, Rb-related p107 protein, and p33 cdk2 protein kinase (Devoto et al., 1992). Consequently, these studies propose that the role of E2F may be to localize the cdk2 kinase to the DNA in a sequence-specific manner such that the cdk2 kinase may phosphorylate a n adjacent DNA-bound factor. Devoto et al. (1992) suggest that a good candidate for the target of the cdk2 kinase is the CTD of RNAP 11. Evidence in support of this idea comes from studies which show that the E2F-cyclin A-p107-cdk2 kinase complex can be found on the DHFR promoter at the E2F binding site (Wade et al., 1992).
The observation that phosphorylation of the CTD of RNAP I1 occurs with each round of transcription from both the Ad-2 MLP and the DHFR promoter suggests that this phosphorylation event may play a role in the regulation of transcription.
Once assembled into a preinitiation complex, it is possible that phosphorylation of the CTD is involved in a key regulatory step to permit the transition from initiation to elongation. Alternatively, phosphorylation may be temporally but not causally related to the initiation process. Irrespective of whether CTD phosphorylation plays a role in the initiation, elongation, or termination phase of transcription, the presence of multiple CTD kinases could enhance the regulatory significance of this modification. Clearly, further studies are necessary to establish the functional andor regulatory significance of CTD phosphorylation.