Flavopiridol inhibits P-TEFb and blocks HIV-1 replication.

Flavopiridol (L86-8275, HMR1275) is a cyclin-dependent kinase (Cdk) inhibitor that is in clinical trials as a cancer treatment because of its antiproliferative properties. We found that the flavonoid potently inhibited transcription by RNA polymerase II in vitro by blocking the transition into productive elongation, a step controlled by P-TEFb. The ability of P-TEFb to phosphorylate the carboxyl-terminal domain of the large subunit of RNA polymerase II was inhibited by flavopiridol with a K(i) of 3 nm. Interestingly, the drug was not competitive with ATP. P-TEFb composed of Cdk9 and cyclin T1 is a required cellular cofactor for the human immunodeficiency virus (HIV-1) transactivator, Tat. Consistent with its ability to inhibit P-TEFb, flavopiridol blocked Tat transactivation of the viral promoter in vitro. Furthermore, flavopiridol blocked HIV-1 replication in both single-round and viral spread assays with an IC(50) of less than 10 nm.

Flavopiridol is a potential anti-cancer therapeutic agent currently being tested in phase I and II clinical trials (1). It has been proposed to target Cdks 1 controlling the cell cycle (2,3) and induce apoptosis in various types of cancerous cells (1). Biochemical evidence indicated that flavopiridol most strongly inhibited Cdk1, Cdk2, and Cdk4 and was less potent on other Cdks tested including Cdk7 (1). A structure of Cdk2 complexed with flavopiridol revealed that the drug docks in the ATP binding site (4), which helps to explain why inhibition of Cdk2 is competitive with ATP. Two previous studies demonstrated an effect of flavopiridol on transcription. Mammalian cells treated with the compound showed decreased transcription of the gene encoding cyclin D1 (5), and high levels of flavopiridol affected levels of 63 different mRNAs in Saccharomyces cerevisiae (6). The transcriptional inhibition observed in these studies could have been direct or a consequence of altered progression through the cell cycle.
The present study focuses on P-TEFb, a protein kinase composed of Cdk9 and a cyclin subunit derived from one of three different genes (7,8). P-TEFb controls the elongation phase of transcription by RNA polymerase II (9) and is a cellular cofactor required for activation of transcription of the HIV-1 genome by the viral transactivator Tat (10,11). Tat forms a triple complex with P-TEFb containing Cdk9 and cyclin T1 and the nascent transcript from the HIV-1 promoter (TAR) (12)(13)(14). This recruitment of P-TEFb activates transcription by causing an increase in the number of RNA polymerase II molecules that synthesize full-length mRNAs (10,15). Our results demonstrated that flavopiridol inhibited P-TEFb causing inhibition of transcription, especially from the HIV-1 promoter in the presence of Tat. The properties of the inhibitor toward P-TEFb and on Tat transactivation suggest that the drug should be examined as a potential HIV-1 therapy.

EXPERIMENTAL PROCEDURES
Transcription and Kinase Assays-Pulse-chase transcription assays were carried out as described previously (16,17) except for modifications indicated in the text. Kinase assays were as described in Marshall et al. (18) using Drosophila RNA polymerase II as the substrate. A 10 mM stock of flavopiridol (Aventis Inc.) in Me 2 SO was stored at Ϫ80°C. The stock was diluted to 0.1 mM in Me 2 SO, and a set of serial dilutions in 4% Me 2 SO was used to give the indicated concentration of flavopiridol. The final concentration of Me 2 SO in transcription or kinase assays was less than 1%.
Single-round HIV-1 Infection Assays-50 l of high titer viral stock generated from transfecting 293T cells with the HIV-1 HXB2 provirus was added to 10 4 Sx22-1 cells that contain one copy of the HIV-1 long terminal repeat linked to the ␤-galactosidase reporter gene (19). Cells were incubated for 5 h to allow entry and integration of HIV-1, washed, and grown for an additional 36 h before being fixed and stained for ␤-galactosidase (19). Flavopiridol was added to the culture 12 h prior to the infection, and cells were grown in its presence for the duration of the assay.
Multiple-round Viral Spread Assay-In the presence of 10 g of Polybrene, HIV-1 NL4 -3 viral particles (containing 2 ng of p24 gag ) were added to 5 ϫ 10 4 Jurkat cells. After 5 h virus was removed by extensive washing. Flavopiridol was added at indicated concentrations from 1.5 to 25 nM. On days 2, 5, 8, 11, and 15, supernatants were collected and cells were incubated with fresh medium and flavopiridol. Reverse transcriptase activity was measured on 10 l of the supernatant. With the high multiplicity of infection used most Jurkat cells were killed after 10 days of infection, and viral titers returned to base-line levels.

RESULTS
To investigate the effects of flavopiridol on transcription an in vitro assay with a template containing the cytomegalovirus (CMV) promoter and HeLa nuclear extract was used. The inclusion of increasing concentrations of flavopiridol in the reaction resulted in a dramatic inhibition of the appearance of the 660-nucleotide run-off transcript (Fig. 1A). The radioactivity in run-off transcripts was quantitated, and the IC 50 for the inhibitory effect was determined to be 34 nM. To determine if flavopiridol inhibited initiation or elongation a pulse-chase assay was used to separate the two processes. Flavopiridol was added into the reactions during the formation of preinitiation complexes (preincubation), initiation (pulse), or elongation (chase) (Fig. 1B). Initiation was not inhibited by the drug as indicated by the uniform production of short transcripts during the pulse (Fig. 1B, lanes marked P). However, the addition of flavopiridol at any step resulted in decreased levels of run-off transcripts during the subsequent chase (Fig. 1B, lanes marked C). Under the influence of flavopiridol the elongation complexes produced shorter, incomplete transcripts indicating that flavopiridol affected the elongation stage of transcription. Almost identical results were obtained when 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole (DRB) was used instead of flavopiridol (Fig. 1C). DRB inhibits P-TEFb, a cyclin-dependent kinase that controls the number of polymerases that enter into productive elongation (20). This similarity suggests that flavopiridol might in-hibit P-TEFb function, but an alternative explanation, that the drug might slow the rate of elongation by RNA polymerase II, was also possible.
To determine if flavopiridol directly inhibited RNA polymerase II, early elongation complexes were isolated under stringent washing conditions that eliminate all known factors able to affect the elongation properties of the polymerase (17). The polymerases in these complexes are able to efficiently, albeit slowly, elongate their nascent transcripts as indicated by the 1-, 2-, 3-, 4-, and 5-min chase time points (Fig. 1D). Flavopiridol had no effect on the elongation properties of RNA polymerase II because transcripts produced in a 5-min chase with increasing amounts of flavopiridol were all the same length (Fig. 1D). Flavopiridol did not have any affect on transcription in vitro by RNA polymerase I or III (data not shown). The data in Fig. 1 strongly suggest that flavopiridol affects the action of P-TEFb.
We next examined the effect of flavopiridol on the ability of P-TEFb to phosphorylate the carboxyl-terminal domain (CTD) of RNA polymerase II. The kinase assay (18) utilized recombinant P-TEFb, composed of human Cdk9 and cyclin T1 (8), and purified RNA polymerase II. P-TEFb was dramatically inhibited by flavopiridol in a standard assay using 10 M ATP ( Fig.  2A). The IC 50 was calculated to be 6 nM. Assays were also carried out using 30, 100, and 300 M ATP. We were surprised to find that the IC 50 did not vary significantly (6 -10 nM) at the different concentrations of ATP (Fig. 2, A-D). The data from Fig. 2 were fit to equations derived for competitive, noncompetitive, and uncompetitive inhibition (21). The data fit uncompetitive inhibition best and gave an apparent K i of 3 nM. This is 4 orders of magnitude below the K m for ATP of 36 M. Even though the kinetic analysis suggested that the drug was not competitive with ATP, a competitive mechanism is possible if very tight binding of the inhibitor essentially inactivates the enzyme. Regardless of the mechanism, these results indicate that flavopiridol is the most effective P-TEFb inhibitor described so far.
Because P-TEFb comprised of Cdk9 and cyclin T1 is required for activation of the HIV-1 promoter by Tat, we examined the effect of flavopiridol on the ability of Tat to activate transcription from the HIV-1 promoter in vitro. Tat stimulated the appearance of a 694-nucleotide run-off transcript from the HIV-1 promoter about 5-fold (Fig. 3A). As has been found before, the P-TEFb inhibitor DRB blocked the formation of long run-off transcripts (Fig. 3A). When increasing concentrations of flavopiridol were included in Tat transactivation reactions the amount of run-off transcription was reduced to background levels. Quantitation of the results indicated that the IC 50 was 7 nM (Fig. 3B). These results are consistent with the ability of flavopiridol to inhibit the kinase activity of P-TEFb.
Flores et al. (22) demonstrated that reducing the activity of P-TEFb via the expression of a dominant negative Cdk9 protein or treatment with P-TEFb inhibitors reduced HIV-1 replication in cells. Therefore, effects of flavopiridol on the single round of infection by HIV-1 HXB2 in Sx22-1 indicator cells and viral replication of HIV-1 NL4 -3 in Jurkat cells were examined (Fig. 4). Sx22-1 cells are HeLa cells that contain one copy of the HIV-1 promoter linked to the ␤-galactosidase reporter gene and can be efficiently infected by the HIV-1 HXB2 strain. Following the infection by HIV-1, the production of Tat leads to the expression of ␤-galactosidase, which is detected by the blue staining of Sx22-1 cells with 5-bromo-4-chloro-3-indolyl ␤-Dgalactopyranoside (X-gal) (19). The addition of flavopiridol reduced the number of blue cells to background levels and exhibited an IC 50 of 8 nM (Fig. 4A). Of note, the Sx22-1 cells remained viable even at the highest concentrations of flavopiridol (100 nM) as determined by trypan blue exclusion (data not shown). These findings were extended with a viral spread assay using Jurkat cells infected with the HIV-1 NL4 -3 virus (Fig. 4B). Using high multiplicities of infection, these cells produce maximal viral titers 5 days after the initial infection after which the cells begin to die. Reverse transcriptase assays again demonstrated that flavopiridol reduced the production of virus in a dose-dependent fashion. As was found in the single-round assay, a dramatic block of HIV-1 replication occurred at a concentration of flavopiridol between 6 and 12.5 nM (Fig. 4B). Cells that did not replicate HIV-1 remained viable at the highest concentra-tion of flavopiridol used (25 nM) (data not shown). We conclude that flavopiridol blocks HIV-1 replication at low nanomolar concentrations in human cells. DISCUSSION The results presented here demonstrate that flavopiridol inhibits P-TEFb more potently than the previously suspected targets Cdk1 and Cdk4. Our data indicate that the K i for flavopiridol against P-TEFb (3 nM) is significantly lower than the K i against either Cdk1 (41 nM) (2) or Cdk4 (65 nM) (3). Because flavopiridol is not competitive with ATP on P-TEFb but is competitive with ATP on Cdk1 and Cdk4, there should be an even greater difference in IC 50 between P-TEFbs compared with Cdk1 or Cdk4 at the high ATP levels found in vivo. Because flavopiridol inhibits P-TEFb much more potently than other Cdks, it is possible that the antiproliferative effects of the drug might be due to inhibition of P-TEFb and the subsequent negative effect on transcription.
The lack of competition with ATP we found for P-TEFb inhibition by flavopiridol suggests that the drug binds relatively tightly to P-TEFb. Because a crystal structure indicates that the drug docks in the ATP binding site of Cdk2 (4), it is likely that the ATP binding site of P-TEFb is also targeted. It is also possible that flavopiridol binds to another site and this is supported by the kinetic data that fit an uncompetitive model best. A structure of P-TEFb complexed with flavopiridol would be useful in resolving this issue.

FIG. 3. Effect of flavopiridol on Tat transactivation in vitro.
Continuous labeling transcription assays using the HIV-1 promoter were performed as described earlier (10). A, autoradiograph of run-off transcripts analyzed by denaturing PAGE. Tat (20 ng/reaction), DRB (50 M), and flavopiridol were added to the indicated reactions. B, run-off transcripts were quantitated, and an IC 50 was determined as described in Fig. 1. The assay shown is representative of more than 5 assays performed with different preparations of Tat. The involvement of P-TEFb in the regulation of transcription from the HIV-1 promoter is well documented, but it is not clear why replication of the virus is more sensitive to reduction of P-TEFb activity than other viral or cellular promoters. One study used both an in vitro and in vivo assay to screen a library of over 100,000 compounds for molecules that blocked Tat transactivation of the HIV-1 promoter but not expression from the CMV promoter. Every compound identified was found to inhibit P-TEFb (23). Three of these inhibitors were shown to block HIV-1, but not human T-cell lymphotrophic virus, type 1 replication at concentrations that were on average about 14 times lower than concentrations that caused cytotoxic effects (22). Significantly, expression of a dominant negative Cdk9 transgene at levels equal to the wild-type Cdk9 dramatically reduced the ability of the cells to support HIV-1 replication (22). HIV-1 replication in T cells requires activation of the cells, which is coupled to up-regulation of cyclin T1 levels (24). Initiation from the HIV-1 promoter in T cells before activation is not accompanied by efficient elongation to make viral mRNAs. After T cell activation, when the levels of P-TEFb containing cyclin T1 have increased, viral mRNAs are produced. This tight control of expression in vivo suggests that the HIV-1 promoter is strongly controlled by the action of negative factors that increase the dependence on P-TEFb. Two factors, negative elongation factor and DRB sensitivity-inducing factor, play a general role in elongation control (25)(26)(27), and DRB sensitivityinducing factor has been shown to play a role in Tat transactivation (28). It is also possible that a CTD phosphatase might be especially active at the HIV-1 promoter. Presumably, the phosphatase would have a negative effect on elongation by reversing the P-TEFb-dependent phosphorylation of the CTD. Support for this idea comes from the finding that the phosphatase is inhibited by Tat (29,30). Studies aimed at elucidating the mechanism of enhanced sensitivity of transcription from the HIV-1 promoter compared with other viral and cellular promoters are needed. These future studies should emphasize in vivo assays with virus or stably integrated viral promoters (22), because less than a 5-fold difference was seen between the HIV-1 and CMV promoter in vitro (Figs. 1 and 3) and in transient transfection assays (not shown).
Because P-TEFb is a key factor in HIV-1 infection and flavopiridol blocks HIV-1 propagation in cultured cells, we suggest that flavopiridol should be evaluated as a potential AIDS therapy. Currently the drug must be administered parenterally, and at its maximal tolerated dose in cancer patients (200 -400 nM), it caused diarrhea and a proinflammatory syndrome (31). However, selection of lower dose levels that achieve the nanomolar drug levels predicted from these experiments to be potentially useful (10 -20 nM) might alleviate these problems. The concept of using a P-TEFb inhibitor, however, may be broadly applicable because P-TEFb is also required for HIV-2, equine infectious anemia virus, simian immunodeficiency virus, and bovine immunodeficiency virus (11). Moreover, treat-ment with current drugs results in the selection and propagation of resistant viral strains. Because P-TEFb is a cellular factor it is unlikely that resistant strains will arise.