The transcriptional elongation inhibitor 5,6-dichloro-1-beta-D-ribofuranosylbenzimidazole inhibits transcription factor IIH-associated protein kinase.

Regulation of chain elongation by RNA polymerase II can have an important effect on gene expression (Bentley, D.(1995) Curr. Opin. Genet. Dev. 5, 210-216; Yankulov, K., Blau, J., Purton, T., Roberts, S., and Bentley, D.(1994) Cell 77, 749-759); however the mechanisms that control this step in transcription are not well understood. The adenosine analogue 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB) has long been used as an inhibitor of RNA polymerase II elongation, but its target is not known. We show that DRB is a potent inhibitor of Cdk-activating kinase, associated with the general transcription factor TFIIH. Two other inhibitors of this kinase, H-7 and H-8, also inhibited transcriptional elongation. Furthermore, TFIIH kinase bound specifically to the herpes simplex virus VP16 activation domain which stimulates polymerase II elongation in addition to initiation (Yankulov, K., Blau, J., Purton, T., Roberts, S., and Bentley, D.(1994) Cell 77, 749-759). Our results suggest that DRB affects transcription by inhibiting the TFIIH-associated kinase and that this kinase functions in the control of elongation by RNA polymerase II.

The adenosine analogue 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole (DRB) 1 inhibits RNA polymerase II elongation; however, the molecular basis of its effect is poorly understood. DRB inhibits the production of long, but not short, RNAs (Chodosh et al., 1989;Fraser et al., 1978;Tamm, 1977). It does not directly interfere with polymerization, however, since it does not prevent elongation in nuclear run-on reactions (Roberts and Bentley, 1992;Tamm, 1977). DRB only inhibits transcription if it is present during or immediately following initiation (Chodosh et al., 1989;Kephart et al., 1992). Poor elongation appears to be an intrinsic property of a distinct class of pol II transcription complexes, which are resistant to DRB. We and others Roberts and Bentley, 1992;Bentley, 1995) have suggested that a modification of poorly processive complexes is required to achieve efficient elongation. DRB was postulated to antagonize a positive transcriptional elongation factor (P-TEF) responsible for this modification . Recently, a factor P-TEFb required for production of long transcripts and for reconstitution of DRB sensitivity was purified from Drosophila nuclear extract (Marshall and Price, 1995). It is not clear, though, whether P-TEFb is a direct target for DRB.
DRB inhibits casein kinase II (Zandomeni et al., 1986), but not protein kinase A, protein kinase C, tyrosine kinases (Meggio et al., 1990), or mitogen-activated protein kinase (Dubois et al., 1994b). Interestingly, suppression of pol II transcription by DRB in vivo coincides with reduced phosphorylation of the pol II large subunit C-terminal domain (Dubois et al., 1994a(Dubois et al., , 1994b. CTD hyperphosphorylation normally accompanies the transition from initiation to elongation (Payne et al., 1989). Therefore, reduced CTD phosphorylation in response to DRB could be an indirect effect of inhibiting transcription. Alternatively DRB could directly inhibit a CTD kinase.
A pol II CTD kinase, identical with the Cdk-activating kinase, CAK, is closely associated with the core subunits of the general transcription factor TFIIH (Feaver et al., 1994;Roy et al., 1994;Serizawa et al., 1995;Shiekhattar et al., 1995). Two distinct forms of TFIIH function in DNA repair and pol II transcription in yeast. Only the latter form is complexed with CAK (Feaver et al., 1994;Svejstrup et al., 1995). Core TFIIH subunits bind to acidic activation domains (Xiao et al., 1994), but it is not known whether CAK also associates with activators. In this paper we show that TFIIH kinase associates with an acidic activation domain and that DRB is a potent inhibitor of this kinase. We suggest a role of TFIIH-associated CAK in regulation of pol II transcriptional elongation.

EXPERIMENTAL PROCEDURES
Affinity Chromatography-GST and GST-VP16(410 -490) were expressed using derivatives of the pGEX2T vector (Pharmacia Biotech Inc.). The proteins were immobilized on glutathione-Sepharose (Pharmacia) at 1 mg/ml. 10-mg samples of HeLa nuclear extract in Buffer D (20 mM KHepes, pH 7.9, 50 mM KCl, 0.5 mM EDTA, 0.5 mM EGTA, 2 mM DTT, 20% (v/v) glycerol) were loaded onto 0.65-ml columns. The columns were washed with 2.7 ml of Buffer D and 2.7 ml of 100 mM KCl in Buffer D. Crude TFIIH was eluted stepwise in 200 and 600 mM KCl in Buffer D. The 200 mM KCl fraction, designated VP16-fraction 5, was used as TFIIH kinase activity.
The protein kinase assay of immunoprecipitated CAK from Xenopus eggs was described previously (Poon et al., 1994). Kinase reactions were with 50 g/ml of GST-CTD or GST-Cdk2(K33R), respectively.
Protein Kinase Substrates-Immunoaffinity-purified calf thymus pol II (Thompson et al., 1990) was a gift of J. Greenblatt. GST-CTD containing all 52 heptad repeats of the mouse CTD was produced from pGCTD (Peterson et al., 1992) or from a derivative, pET21a-GCTD. TFIIE p34 and p56 were produced from vectors supplied by Dr. R. Tjian. The p34 expression vector was modified by insertion of an oligonucleotide encoding a His 6 tag into the NdeI site. The p56-His 6 p34 complex was isolated by mixing bacterial cell lysates containing the two subunits for 20 min on ice prior to purification on Ni 2ϩ agarose (Qiagen). TFIIF (rap74-rap30) was expressed from vectors provided by Dr. Z. Burton. The rap74 vector was modified by insertion of a His 6 tag at the NcoI site and the His 6 rap74-rap30 complex isolated as for TFIIE. The kinase-deficient Cdk2 substrate GST-Cdk2(K33R) was described previously (Poon et al., 1994).
ATPase Assay-ATPase assays were performed under protein kinase reaction conditions. 0.2 l of TFIIH HAP fraction, 300 ng of VP16fraction 5, or 5 g of HeLa nuclear extract were incubated for 2 h at 30°C. Under these conditions the ATPase reaction was linear for more than 3 h. Reactions were terminated by adding 0.8 ml of ice-cold 5% activated charcoal (Sigma) in 7 mM H 3 P0 4 . The mixture was centrifuged, and released inorganic phosphate was measured by counting aliquots of the supernatant.
In Vitro Transcription-RNase protection, purification of recombinant GAL4-AH and GAL4-VP16 and the plasmids pSPVA (adenovirus VA1) and pGal 5 -HIV2 CAT have been described (Yankulov et al., 1994). Transcription reactions (20 l) contained 250 ng of supercoiled plasmids pGal 5 -HIV2 CAT and pSPVA, 80 g of HeLa nuclear extract (Dignam et al., 1983), 100 ng of GAL4-AH or GAL4-VP16, 1 mM MgCl 2 , 1 mM spermidine, 4% polyethylene glycol 8000, 0.5 mM NTPs, 50 mM KCl, 12 mM Hepes, pH 7.9, 10 M ZnCl 2 , 2 mM DTT, 20 mM creatine phosphate, 0.01% Nonidet P-40, 15 units of RNAguard (Pharmacia). Reactions were incubated at 30°C for 1 h, stopped by addition of ␣-amanitin (5 g/ml), 1 mM CaCl 2 , and 2 l of RQ DNase I (Promega). After 10 min at 30°C, SDS (0.4%) and proteinase K (400 g/ml) were added and incubated 10 min at 37°C. The reactions were processed for RNase protection using antisense probes for HIV2 and VA (Yankulov et al., 1994). RESULTS We previously observed that an excess of the herpes simplex virus VP16 activation domain in trans ("squelching") (Gill and Ptashne, 1988) specifically inhibited production of long transcripts in Xenopus oocytes presumably by titrating a factor(s) required for efficient elongation (Yankulov et al., 1994). The similar effects of squelching and DRB on transcriptional elongation prompted us to look for a DRB target by VP16 affinity chromatography. Because DRB inhibits pol II CTD phosphorylation in vivo (Dubois et al., 1994a), we looked for a CTD kinase activity that bound to the VP16 activation domain and was inhibited by DRB. HeLa nuclear extract was fractionated on GST-VP16 affinity resin, and fractions were assayed for phosphorylation of a GST-CTD substrate. We detected a CTD kinase activity which bound to the GST-VP16 column ( Fig. 1) but not to control GST or mutant GST-VP16 (GST-SW6 (Walker et al., 1993)) columns (data not shown). The CTD kinase co-eluted with two subunits of TFIIH: p62 and p40 MO15 , the catalytic subunit of CAK. (There was less than expected kinase activity in fraction 6, possibly due to an inhibitor or a phosphatase). Two other CTD kinases, p34 cdc2 (Cisek and Corden, 1989) and DNA-dependent protein kinase monitored by detection of the p70 KU subunit (Peterson et al., 1992), were not retained on the VP16 column (Fig. 1). The CTD kinase activity in VP16-fraction 5 was immunodepleted by anti-p62 antibodies 2 indicating that it was associated with TFIIH. Xenopus TFIIH from oocyte extracts was also specifically retained on a GST-VP16 column. 3 The VP16-associated GST-CTD kinase was assayed for activity on a physiological substrate, purified calf thymus pol II. Phosphorylation of pol II large subunit by VP16-fraction 5 was stimulated by DNA and recombinant TFIIE in accordance with the properties of TFIIH (Lu et al., 1992). Importantly, pol II phosphorylation under these conditions was inhibited by DRB with an IC 50 of about 25 M (Fig. 2, A and D).
Highly purified human TFIIH (HAP fraction, Gerard et al., 1991) and the VP16-associated kinase were compared for sensitivity to DRB using three recombinant substrates: GST-CTD, TFIIE, and TFIIF. Both kinases phosphorylated the p56, but not the p34, subunit of TFIIE and the rap74, but not the rap30 subunit of TFIIF, 4 in agreement with the reported substrate specificity of TFIIH (Ohkuma and Roeder, 1994). The two kinase preparations were inhibited by DRB with IC 50 values between 10 and 50 M for all three recombinant substrates (Fig. 2B) in agreement with the results for calf thymus pol II. Furthermore, CAK prepared by immunoprecipitation from Xenopus egg extract with anti-p40 MO15 anti-peptide antibody (Poon et al., 1994) phosphorylated both GST-Cdk2(K33R) and GST-CTD in a DRB-sensitive manner which closely resembled TFIIH (Fig. 2C).
The previously identified inhibitors of TFIIH kinase, H-7 and H-8 (Serizawa et al., 1993b), also inhibited phosphorylation of GST-CTD (not shown) and pol II by the VP16-associated kinase, but at substantially higher concentrations than DRB (Fig. 2D). In contrast, the DNA-dependent ATPase activity of highly purified TFIIH was not inhibited even at 1 mM DRB ( Fig. 2E and data not shown), nor did DRB inhibit the overall ATPase activity or total autophosphorylation activity of crude HeLa nuclear extract (Fig. 2E). DRB at 1 mM also had no effect 2 K. Yankulov and S. Dhut,unpublished results. 3  on the 5Ј-3Ј and 3Ј-5Ј helicase activities of purified TFIIH. 5 It was previously demonstrated that H-8 had no effect on TFIIH ATPase activity (Serizawa et al., 1993b).
To address whether inhibition of elongation is a general property of TFIIH kinase inhibitors, we investigated the effects of H-7 and H-8 on pol II transcription in HeLa nuclear extract (Fig. 3). Transcription reactions were performed with a template containing five GAL4 binding sites fused upstream of the HIV2 TATA box and TAR sequences (pGal 5 -HIV2 CAT). Transcription was activated by recombinant GAL4-AH or GAL4-VP16. Read-through (RT) and prematurely terminated (TM) transcripts were quantified by PhosphorImager analysis of RNase protection assays (Fig. 3A). Addition of DRB, H-7, or H-8 reduced the elongation efficiency (RT/RT ϩ TM) from 70 to 23% in the presence of GAL4-AH (Fig. 3A). Similar results were obtained when transcription was activated by GAL4-VP16 (Fig.  3B). The dose-response curves in Fig. 3B show that whereas DRB significantly inhibited elongation at 10 M, H-7 and H-8 were only effective in excess of 50 M. H-7 and H-8 also inhibited elongation through the mouse c-myc gene in HeLa extract. 6 A similar inhibitory effect (DRB Ͼ H-7, H-8) on elongation of transcripts from the mouse c-myc, HIV2 LTR, and Gal 5 -E1b promoters (Yankulov et al., 1994) was observed in injected Xenopus oocytes. 7

DISCUSSION
In this paper we show that H-7, H-8, and DRB inhibit both TFIIH (CAK) kinase and pol II elongation. The dose-response curves for inhibition of transcription and kinase activity were similar for each compound, and DRB was always more effective 5 R. Roy and J.-M. Egly, unpublished results. 6 D. Bentley, unpublished results. 7 K. Yankulov, unpublished results.

FIG. 2. DRB specifically inhibits TFIIH-associated kinase.
A, phosphorylation of pol II by VP16-associated kinase is stimulated by TFIIE and inhibited by DRB. Kinase reactions contained 5 g/ml calf thymus pol II, 0.8 g/ml recombinant TFIIE, and 1.0 (left panel) or 1.5 g/ml fraction 5 from the VP16 affinity column (right panel). B, kinase reactions with three recombinant substrates using VP16-associated kinase (fraction 5, Fig. 1) or highly purified TFIIH (HAP fraction (Gerard et al., 1991)). Autoradiograms of phosphorylated GST-CTD, TFIIE p56, and TFIIF rap74 are shown. C, CAK, immunoprecipitated from Xenopus egg extract (Poon et al., 1994), is inhibited by DRB. Substrates were GST-CTD (Ⅺ) and GST-Cdk2(K33R), a kinase deficient mutant of Cdk2 (छ). D, dose response curves for DRB (Ⅺ), H-7 (छ), and H-8 (E) inhibition of pol II phosphorylation by VP16-fraction 5 as in A. E, DRB specifically inhibits the kinase, but not the ATPase, activity of TFIIH. GST-CTD phosphorylation by VP16-fraction 5 (Ⅺ) or pure TFIIH (छ) data are from B. ATPase activity in the presence of DNA was determined for pure TFIIH (E), VP16-fraction 5 (Ç) and HeLa nuclear extract (NE) (µ). Total autophosphorylation activity of nuclear extract (NE KINASE, ࡗ) was determined using 1.5 g of HeLa nuclear extract under standard kinase assay conditions without added substrate. than H-7 and H-8. It is particularly significant that TFIIH kinase is highly sensitive to DRB, which has long been known to interfere with transcript elongation (Fraser et al., 1978;Tamm, 1977), although its mechanism of action has remained obscure. Casein kinase II is quite specifically inhibited by DRB (Zandomeni et al., 1986), but it is unlikely to be involved in the effect on elongation, because it is unaffected by H-7 and H-8 at the concentrations we used (Hidaka and Kobayashi, 1992). Conversely protein kinase A and protein kinase C, which are inhibited at low concentrations of H-7 and H-8, are unaffected by DRB (Meggio et al., 1990). We suggest that the relevant target for these inhibitors of pol II elongation is the kinase associated with the general transcription factor TFIIH.
H-8 had no effect on basal transcription in a highly purified reconstituted system (Serizawa et al., 1993a), but inhibited transcription in crude yeast cell extract (Li and Kornberg, 1994). We show that in crude HeLa nuclear extract H-7, H-8, and DRB inhibit pol II elongation (Fig. 3). In Drosophila extract P-ETFb is required for production of long transcripts and for sensitivity to DRB (Marshall and Price, 1995). Sensitivity of pol II transcription to inhibitors of TFIIH kinase may therefore require additional elongation factors such as P-ETFb, which are not present in highly purified systems. It is not clear whether these factors are substrates for TFIIH kinase. Our results suggest a role for TFIIH kinase in stimulation of pol II elongation, but they do not exclude additional functions in regulating initiation or promoter clearance.
A role of TFIIH kinase in stimulating elongation is consistent with our observation that this kinase binds to the VP16 activation domain which is a potent stimulator of pol II elongation (Yankulov et al., 1994). This activation domain was previously shown to make direct contacts with subunits of the core TFIIH (Xiao et al., 1994). It is not known, however, whether TFIIH kinase activity is directly stimulated by acidic activators.
TFIIH kinase phosphorylates pol II CTD, TATA-binding protein, and the large subunits of TFIIF and TFIIE in vitro (Ohkuma and Roeder, 1994; Fig. 2B). The relative importance of these phosphorylations for transcriptional elongation is not established. The modification of a known elongation factor, TFIIF, could stimulate processivity by stabilizing its interaction with pol II which is quite labile (Price et al., 1989). A role for the CTD in elongation is suggested by the observation that pol II is hypophosphorylated when it enters the preinitiation complex and when it pauses shortly after initiation, whereas the actively elongating form is hyperphosphorylated (Lu et al., 1991;O'Brien et al., 1994;Payne et al., 1989;Weeks et al., 1993). The timing of CTD phosphorylation therefore appears to coincide with the time when DRB is effective, during or immediately following initiation (Cisek and Corden, 1989;Kephart et al., 1992). Furthermore, DRB inhibits CTD phosphorylation in vivo (Dubois et al., 1994a(Dubois et al., , 1994b. These data are consistent with the idea that CTD phosphorylation by TFIIH is the DRBsensitive modification Roberts and Bentley, 1992;Bentley, 1995), which stimulates elongation by pol II.