T he HIV transactivator T AT binds to the
CDK-activating kinase and activates the
phosphorylation of the carboxy-terminal
domain of RNA polymerase II
T homas P. Cujec,1Hiroshi Okamoto,1Koh Fujinaga,1Jon Meyer,1Holly Chamberlin,2
David O. Morgan,2and B. Matija Peterlin1,3
1Howard Hughes Medical Institute, Departments of Medicine, Microbiology, and Immunology;2Department of Physiology,
University of California at San Francisco, San Franscisco, California USA
T he human immunodeficiency virus encodes the transcriptional transactivator T at, which binds to the
transactivation response (T AR) RNA stem–loop in the viral long terminal repeat (LT R) and increases rates of
elongation rather than initiation of transcription by RNA polymerase II (Pol II). In this study, we demonstrate
that T at binds directly to the cyclin-dependent kinase 7 (CDK7), which leads to productive interactions
between T at and the CDK-activating kinase (CAK) complex and between T at and T FIIH. T at activates the
phosphorylation of the carboxy-terminal domain (CT D) of Pol II by CAK in vitro. T he ability of CAK to
phosphorylate the CT D can be inhibited specifically by a CDK7 pseudosubstrate peptide that also inhibits
transcriptional activation by T at in vitro and in vivo. We conclude that the phosphorylation of the CT D by
CAK is essential for T at transactivation. Our data identify a cellular protein that interacts with the activation
domain of T at, demonstrate that this interaction is critical for the function of T at, and provide a mechanism
by which T at increases the processivity of Pol II.
[Key Words: HIV; transactivator T at; phosphorylation; CT D; Pol II; CAK; CDK7]
Received July 8, 1997; revised version accepted August 19, 1997.
T he human immunodeficiency virus (HIV) encodes a
highly conserved transcriptional transactivator T at, that
is expressed early in the viral life cycle and is essential
for viral replication and progression to disease (Cullen
1993; Jones and Peterlin 1994). T at binds to the transac-
tivation response (T AR) RNA stem–loop located from
positions +1 to +60 in the viral 5? long terminal repeat
(LT R). Interactions between T at and T AR are absolutely
required for the increased processivity of RNA polymer-
ase II (Pol II) and the production of full-length viral tran-
scripts (Kao et al. 1987; Laspia et al. 1989; Marciniak and
Sharp 1991; Kato et al. 1992). T at is unique because it is
the only eukaryotic transcription elongation factor
known to function via RNA (Madore and Cullen 1995).
Although the mechanism by which T at increases tran-
scription elongation rates is unknown, common regula-
tory themes must exist between viral and cellular genes
because T at can relieve Pol II pausing when artificially
targeted to the c-myc promoter (Wright et al. 1994).
T at can be divided into two functional domains. T he
activation domain contains 48 amino-terminal amino
acids and interacts with the cellular transcriptional ma-
chinery. A 10-amino-acid basic domain is required for
the binding of T at to T AR (Jones and Peterlin 1994). Cel-
lular proteins are clearly required for the function of T at
(Carroll et al. 1992; Madore and Cullen 1993). At least
one protein, encoded by the human chromosome 12, is
required for the efficient binding of T at to T AR (Hart et
al. 1989; Alonso et al. 1992). Numerous other proteins
have been postulated to interact with the activation do-
main of T at. T hese include general transcription factors
(GT Fs) such as the core Pol II (Mavankal et al. 1996), the
T AT A box-binding protein (T BP) (Kashanchi et al. 1994;
Veschambre et al. 1995), the T BP-associated factor
T AFII55 (Chiang and Roeder 1995) and T FIIH (Blau et al.
1996; Parada and Roeder 1996). Upstream DNA-bound
activators such as Sp1 have also been identified as pos-
sible co-activators of T at (Jeang et al. 1993). In addition,
a wide variety of other proteins that interact with T at
but whose role in transcription is somewhat unclear
have also been identified (Nelbock et al. 1990; Desai et
al. 1991; Zhou and Sharp 1996). Recently, T at has been
demonstrated to interact with the Pol II holoenzyme
(Cujec et al. 1997). T his large megadalton complex con-
sists of core Pol II, a subset of general transcription fac-
tors (T FIIE, T FIIF, T FIIH), human SRBs (suppressors of
mutations in RNA polymerase B), which confer the abil-
E-MAIL MAT IJA@IT SA.UCSF.EDU; FAX (415) 502-5081.
GENES & DEV ELOPMENT 11:2645–2657 © 1997 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/97 $5.00 2645
ity of the Pol II holoenzyme to respond to activators and
proteins involved in chromatin remodeling (SWI/SNF)
and DNA repair (Kim et al. 1994; Ossipow et al. 1995;
Chao et al. 1996; Maldonado et al. 1996; Wilson et al.
T FIIH contains nine polypeptides [ERCC3, ERCC2,
p62, p54, p44, CDK7 (MO15), cyclin H, MAT 1, and p34]
(Drapkin and Reinberg 1994; Hoeijmakers et al. 1996). It
contains a kinase activity that can phosphorylate the
carboxy-terminal domain (CT D) of Pol II (Feaver et al.
1991; Lu et al. 1992). T he kinase activity resides in the
cyclin-dependent kinase 7 (CDK7) subunit (Feaver et al.
1994; Roy et al. 1994; Serizawaet al. 1995; Shiekhattar et
al. 1995). In association with cyclin H, CDK7 forms the
CDK-activating kinase (CAK) complex that phosphory-
lates CDKs involved in the regulation of the cell cycle.
Association of MAT 1 with theCAK dimer stabilizes the
complex and allows for the activation of CAK indepen-
dently of the phosphorylation of CDK7 on the threonine
at position 170 (Fisher and Morgan 1994; Fisher et al.
1995). Moreover, the CAK trimer is much more efficient
at phosphorylating the CT D than the CAK dimer (Ros-
signol et al. 1997; Yankulov and Bentley 1997). T he tri-
partite CAK can exist in three distinct complexes in
cells. T he majority is present as free CAK. However,
CAK can also exist as a CAK–ERCC2 complex as well as
in association with the core T FIIH (ERCC3, p62, p54,
p44, and p34) (Drapkin et al. 1996; Reardon et al. 1996).
T he association of CAK with T FIIH confers kinase ac-
tivity to T FIIH and renders it transcriptionally compe-
tent. Interestingly, theyeast homolog of CDK7, Kin28, is
found only in a complex with T FIIH and is devoid of
CAK activity (Cismowski et al. 1995). Instead, CAK ac-
tivity resides in a novel protein called Civ1 or CAK1p
(Kladis et al. 1996; T huret et al. 1996).
T he eukaryotic Pol II is unique among polymerases in
that it contains multiple heptapeptide repeats of the se-
quence YSPT SPS, which together comprise the CT D
(Dahmus 1994, 1995). A large number of kinases capable
of phosphorylating the CT D in vitro have been identi-
fied. However, the functional relevance of these kinases
remains unclear. T o date, CDK7/cyclin H (T FIIH) and
CDK8/cyclin C (human homologs of the yeast SRB10/
SRB11) are the major kinases associated with transcrip-
tion factors that can phosphorylate the CT D (Liao et al.
1995; Serizawa et al. 1995; Shiekhattar et al. 1995).
Pol II enters into the assembling transcription com-
plex with its CT D unphosphorylated (IIA form). How-
ever, the CT D of elongating polymerases is highly phos-
phorylated (IIo form) primarily on its serine and threo-
nine residues (Laybourn and Dahmus 1989, 1990). T his
observation led to the suggestion that the phosphoryla-
tion of the CT D is important for promoter clearance and
for the processivity of Pol II. Numerous additional ob-
servations support this contention: (1) the CT D of poly-
merases paused on the Drosophila hsp 70 promoter be-
fore heat shock activation are hypophosphorylated (IIa),
whereas those of actively elongating polymerases are hy-
perphosphorylated (IIo) (O’Brien et al. 1994); (2) inhibi-
tors of CT D kinases inhibit promoter clearanceandelon-
gation of Pol II in vitro (Yankulov et al. 1995, 1996); (3)
the kinase activity of T FIIH is required for the clearance
of the DHFR promoter but not for the initiation of its
transcription (Akoulitchev et al. 1995); and (4) mutations
in the yeast Pol II CT D, the yeast homolog of CDK7 (Kin
28), or SRB2, a subunit of the Pol II holoenzyme, each
inhibit the processivity of Pol II in vivo (Akhtar et al.
1996). T he identification of CT D-binding proteins with
homology to serine/arginine-rich (SR) proteins suggest
that the phosphorylation of the CT D might also provide
a mechanism for coupling transcription and pre-mRNA
processing (Yuryev et al. 1996).
Recently, we and others demonstrated that the CT D is
absolutely required for the production of long transcripts
from the HIV LT R in vitro and in vivo (Chun and Jeang
1996; Okamoto et al. 1996; Parada and Roeder 1996;
Yang et al. 1996). In contrast, basal transcription and the
production of short transcripts from the HIV LT R is in-
dependent of the CT D. T ogether these results suggested
an important role for the CT D in the function of T at.
Given the critical role of the CT D for the function of
T at, and the fact that T FIIH is part of the Pol II holoen-
zyme that interacts with T at, we examined the possibil-
ity that T at might interact with T FIIH. In this study, we
report that T at co-immunoprecipitates with T FIIH/CAK
complexes in vivo. Furthermore, T at binds to recombi-
nant CAK in vitro and this association is mediated by
direct interactions between T at and CDK7. Functional
relevance for this association is provided by experiments
that demonstrate that T at potentiates the ability of
T FIIH and CAK to phosphorylate the CT D of Pol II. Fi-
nally, using specific pseudosubstrate peptides, which in-
hibit the activity of the CDK7 kinase, we demonstrate
that the phosphorylation of the CT D by CDK7 is re-
quired for T at transactivation in vitro and in vivo.
Tat interacts with TFIIH in vivo
We demonstrated recently that T at interacts with the
Pol II holoenzyme (Cujec et al. 1997). Given that T FIIH
is present in the Pol II holoenzyme (Ossipow et al. 1995;
Maldonado et al. 1996) and the essential role that the
phosphorylation of theCT D by T FIIH is believed to have
in the processivity of Pol II, we examined whether T at
can associate with T FIIH independently of the Pol II ho-
loenzyme. T o determine whether T at interacts with
T FIIH in vivo, hemagglutinin (HA)-tagged wild-type and
mutant T at proteins were expressed in COS cells. Im-
munoprecipitations were done with an antibody that is
specific for ERCC3 (?ERCC3), the largest subunit of
T FIIH (Drapkin et al. 1996). Because T FIIH is part of the
Pol II holoenzyme, coimmunoprecipitations were done
under stringent conditions (0.5 M NaCl and 1% T riton
X-100), which were expected to dissociate T FIIH from
the Pol II holoenzyme. T o confirm that only T FIIH, and
not the entire Pol II holoenzyme, was immunoprecipi-
tated in these experiments, ?ERCC3 immunoprecipi-
tates werealso probed with antibodies against thelargest
Cujec et al.
2646GENES & DEV ELOPMENT
subunit of the core Pol II (RPB1) (T hompson et al. 1989).
As shown in Figure 1A, ?ERCC3 antibodies efficiently
precipitated ERCC3 and CDK7 subunits of T FIIH, but
failed to precipitate Pol II (Fig. 1A, lane 3). As a control,
antibodies against the transcriptional activator CIIT A
(C) (Steimle et al. 1993) failed to coimmunoprecipitate
any of the transcription factors (Fig. 1A, lane 2). T o de-
termine if T at could bind to T FIIH in vivo, ?ERCC3
immunoprecipitates were probed with ?HA antibodies.
As shown in Figure 1B, although equivalent amounts of
wild-type (T at) and mutant (C30G and K41A) T at pro-
teins were expressed in COS cells (Fig. 1B, left three
lanes), only the wild-type T at was coimmunoprecipi-
tated by the ?ERCC3 antibodies (Fig. 1B, right three
lanes). Again, antibodies against the transcriptional ac-
tivator CIIT A (C) failedto coimmunoprecipitateT at (Fig.
1B). Similar results were obtained when using ?CDK7
antibodies (Fig. 1C). T ogether, these results demonstrate
that T at can interact with T FIIH in the absence of both
the Pol II holoenzyme and T AR in vivo.
Tat Binds to CDK7 in vitro
Although wecould detect readily an interaction between
T at and purified T FIIH in vitro (Parada and Roeder 1996;
data not shown), repeated attempts to detect specific in-
teractions between T at and ERCC3, ERCC2, or p62 sub-
units of T FIIH wereunsuccessful. Of theseproteins, only
p62 bound to T at and all our mutant T at proteins (data
not shown). T herefore, we tested whether T at could bind
to recombinant CAK in vitro (Fig. 2). Recombinant wild-
type and mutant T at proteins were attached to strepta-
vidin agarosebeads by virtueof streptavidin-binding pep-
tides at their 3? termini (Cujec et al. 1997). CDK7, cyclin
H, and MAT 1 were coexpressed in insect cells by bacu-
lovirus infection. Extracts were purified initially over a
HiT rap chelating column and then an anion exchange
column (HiT rap Q). Relevant fractions were purified fur-
ther by gel filtration and concentrated on a second Hi-
T rap Q column (see Materials and Methods). T he result-
ing CAK preparations were >99% pure and were subse-
quently used for crystallization studies (data not shown).
CAK was incubated with streptavidin–agarose beads
alone (C) or with streptavidin–agarose beads containing
wild-type (T at) or mutant (mT at = K41A) T at proteins
bound to them. After washing, the presence of CAK was
monitored by Western blotting using ?cyclin H antibod-
ies. As shown in Figure 2, wild-type T at (Fig. 2, lane 4),
but not mutant T at (Fig. 2, lane 5), bound to CAK very
efficiently, as at least half of the input CAK was retained
on T at beads (Fig. 2, cf. lanes 2 and 4 with 5). Western
blotting with ?CDK7 and ?MAT 1 antibodies revealed
the presence of the other CAK subunits as well (data not
shown). T at proteins containing other debilitating mu-
tations in their activation domains (C22G or C30G)
(Kuppuswamy et al. 1989; Madore and Cullen 1993; Cu-
jec et al. 1997) also failed to bind recombinant CAK in
vitro (data not shown).
holoenzyme in vivo. (A) T otal cell lysates (T CL) from COS cells
were immunoprecipitated with ?CIIT A (lane 2) or ?ERCC3 an-
tibodies (lane 3). One-third of the lysate was used as the input
control (lane 1). Samples were separated by SDS-PAGE, trans-
ferred to membranes, and probed with the antibodies indicated
on the left. (B) COS cells expressed HA-tagged wild-type (T at) or
mutant (C30G or K41A) T at proteins. Cell lysates were immu-
noprecipitated with either the ?CIIT A (C) antibody or the
?ERCC3 antibody (?ERCC3). Equal amounts of thelysates were
loaded as input controls. Samples were processed as in A and
probed with the ?HA antibody. (C) Nuclear extracts (NE) or
lysates from cells expressing HA-tagged wild-type T at (T at) or
mutant T at (mT at = K41A) were immunoprecipitated with the
?CDK7 antibody. One-half of these cell lysates were loaded as
input controls (lanes 1,2). Samples were processed and blots
probed with ?HA antibodies as described above.
T at associates with T FIIH in the absence of the Pol II
combinant CAK trimer (50 ng) was incubated with the wild-
type (T at) or mutant (mT at = K41A) T at proteins (1 µg) bound to
streptavidin–agarose beads or with streptavidin–agarose beads
alone as the control (C). Nuclear extract (10 µg, lane 1) or CAK
(50 ng, lane 2) were loaded as input controls. After washing, the
streptavidin beads and controls were subjected to SDS-PAGE,
transferred to membranes, and probed with the ?cyclin H anti-
T at associates with recombinant CAK in vitro. Re-
T at binds to and activates CAK
GENES & DEV ELOPMENT2647
We next tested the ability of each of the individual
CAK subunits to bind to T at. CDK7, cyclin H, and MAT
1 were labeled with [35S]methionine by in vitro transla-
tion and CAK dimers or trimers were allowed to form.
Complexes were immunoprecipitated with ?CDK7 an-
tibodies bound to protein-A–Sepharose beads. Immuno-
precipitated complexes were then incubated with [?-
32P]AT P-labeled T at and the ability of CDK7 beads to
retain T at was evaluated. As demonstrated in Figure 3,
equal amounts of T at were retained on these beads re-
gardless of whether CDK7 was present alone (Fig. 3, lane
6), as a dimer with cyclin H (Fig. 3, lane 9), or as the CAK
trimer (Fig. 3, lane 12). Although equivalent amounts of
wild-type (T ) or mutant (mT ) T at proteins were used in
these experiments (Fig. 3, lanes 13,14), the mutant T at
did not bind to any CDK7 complexes. As expected, ad-
dition of MAT 1 to the CDK7–cyclin H dimers stabilized
the CAK complex (Fisher et al. 1995) and increased the
amount of CAK immunoprecipitated by ?CDK7 anti-
bodies (Fig. 3, cf. lane 7 with 10). T hese results suggest
that the reticulocyte lysate used for in vitro translation
contained minimal amounts of endogenous cyclin H or
MAT 1 proteins that could associate with the introduced
CDK7 monomer or CDK7–cyclin H dimer and form the
tripartite CAK complex. Similar experiments failed to
reveal specific interactions between T at and cyclin H or
MAT 1 in the absence of CDK7 (data not shown). T o-
gether these results demonstrate that T at can interact
with CDK7 directly and that this interaction leads to the
association of T at with the higher order CAK and T FIIH
Tat increases the ability of CAK to phosphorylate
Having demonstrated a strong affinity between T at and
CAK, we wished to determine whether T at affected the
ability of CAK to phosphorylate the CT D of Pol II. Wild-
type or mutant CAK /T FIIH complexes were immuno-
precipitated from lysates of cells that stably expressed
HA-tagged wild-type CDK7 or its kinase-deficient vari-
ant (D155A). Although this mutation (D155A) abolishes
the kinase activity of CDK7, it does not affect its ability
to bind to cyclin H and MAT 1 (D.O. Morgan, unpubl.).
lengths of time with recombinant CT D and wild-type or
mutant T at proteins (Fig. 4A). T he phosphorylation of
the CT D by casein kinase (CK) was used as a marker for
the electrophoretic mobility of the CT D (Fig. 4A, lane 1).
Because casein kinase only phosphorylates the CT D at
one site (Dahmus 1996), only the hypophosphorylated
form of the CT D is apparent (Fig. 4, lane 1). Importantly,
this control demonstrates that only the hypophosphory-
lated form of the CT D (CT Da) is present in our protein
preparations. Wild-type but not mutant (mT at) T at pro-
teins increased the ability of immunoprecipitated CAK /
T FIIH complexes to phosphorylate the hypophosphory-
lated (CT Da) and hyperphosphorylated (CT Do) forms of
the CT D. T he effect of T at was evident after 20 min and
increased throughout the duration of the experiment
(Fig. 4A, cf. lanes 2–5 with 6–9). Interestingly, T at didnot
affect the ability of CAK /T FIIH complexes to autophos-
phorylate cyclin H. As a control for the specificity of our
complexes (M) were also immunoprecipitated. As ex-
pected, these complexes failed to phosphorylate the
CT D and cyclin H (Fig. 4A, lane 10).
Because approximately 10% of CAK is part of T FIIH,
and the rest exists as a free complex (Drapkin and Rein-
berg 1994; Fisher and Morgan 1994; Fisher et al. 1995),
immunoprecipitations with ?CDK7 antibodies would be
expected to yield both CAK and some T FIIH. T o test
directly whether T at could affect the ability of CAK to
phosphorylate the CT D, recombinant CAK was incu-
bated with the CT D in the presence or absence of wild-
type or mutant T at proteins. As demonstrated in Figure
4B, wild-type T at (T at) was much more effective at
stimulating the ability of CAK to phosphorylate the
CT D than the mutant T at protein (mT at) (Fig. 4B, cf.
lanes 3 and 4 with 2, 5, and 6). T he wild-type T at also
increased the ability of CAK to phosphorylate highly pu-
rified preparations of Pol II (Fig. 4C). In neither case did
T at affect the phosphorylation of cyclin H. T he CDK7
monomer was incapable of phosphorylating the CT D
(datanot shown). By demonstrating that T at can increase
the kinase activity of CDK7, we provide functional rel-
evance for the binding results presented in the previous
T o test for the specificity of these effects of T at, we
CAK /T FIIH
CDK7, cyclin H, and MAT 1 (p36) were labeled with
translation in vitro. CDK7 alone (lanes 4–6), in combination
with cyclin H (lanes 7–9), or as a mixture with cyclin H and
MAT 1 (lanes 10–12), was immunoprecipitated with ?CDK7
antibodies bound to protein-A–Sepharose beads. Immunopre-
cipitates were divided into three fractions. Whereas one-third
served as a control for immunoprecipitations (lanes 4,7,10), the
remaining two-thirds were incubated with wild-type (T ) or mu-
tant (mT = K41A) T at proteins labeled at their 3? termini with
[?-32P]AT P using the catalytic subunit of cAMP-dependent
heart muscle kinase. After washing, the beads were subjected to
SDS-PAGE. Gels were dried and visualized by autoradiography.
One-fifth of the wild-type and mutant T at proteins used in the
binding reactions were loaded as the input controls (lanes
T at associates with recombinant CDK7 in vitro.
Cujec et al.
2648GENES & DEV ELOPMENT
next examined whether T at affected the ability of CAK
to phosphorylate cyclinA?171/CDK2HA
CyclinA?171/CDK2HA complexes were bound to pro-
tein-A–Sepharose beads and incubated with CAK in the
presence of wild-type or mutant T at proteins. Because
the ability of cyclinA?171/CDK2HA to phosphorylate
histone H1 is dependent on the phosphorylation of
CDK2 by CAK, thephosphorylation of histoneH1 can be
used as a measure of CAK activity (Fisher and Morgan
1994). As shown in Figure 4D, neither the wild-type
(T at) nor mutant T at (mT at) affected the ability of CAK
to phosphorylate cyclinA?171/CDK2HA. As expected,
H1 was not phosphorylated if either cyclinA?171/
CDK2HA or CAK were omitted from the reaction. Fur-
thermore, thesecontrols demonstratethat our T at prepa-
ration did not contain a kinase activity capable of acti-
A CDK2 mutant peptide inhibits the phosphorylation
of the CTD by CAK
T he activity of some kinases can be inhibited by excess
amounts of substratepeptides that contain amutation in
the amino acid phosphorylated by the kinase of interest
(Poteet-Smith et al. 1997). Because CAK phosphorylates
the threonine of CDK2 at position 160 (Fisher and Mor-
gan 1994; Makela et al. 1994), wetested whether a CDK2
peptide (amino acids 149–170) having a T → A mutation
(mC2p) could inhibit phosphorylation of the CT D by
CAK. Increasing amounts of the CDK2 mutant peptide
(mC2p) were added to kinase reactions containing re-
combinant CAK (0.02 µM) in the presence or absence of
T at. As demonstrated in Figure 5A, a 65-fold molar ex-
to phosphorylate the CT D of Pol II. (A)
Wild-type HA-tagged CDK7 (CDK7), or a
kinase-deficient mutant of CDK7 (M) were
immunoprecipitated from stably express-
ing HeLa cells. GST –CT D protein (25 ng)
was incubated with casein kinase (CK, 25
ng), or CDK7 immunoprecipitates in the
presence of wild-type (T at) or mutant
(mT at = K41A) T at proteins and ?-AT P.
Kinase reaction products were subjected to
SDS-PAGE, andgels werevisualizedby au-
toradiography after drying. Positions of the
hypophosphorylated (CT Da) and hyper-
sions, as well as cyclin H are indicated
on the left. (B) GST –CT D was incubated
with recombinant CAK (50 ng) in the
presence of wild-type (T at) or mutant
(mT at = K41A) T at proteins in a kinase re-
action. Products were processed as de-
scribed above. (C) Purified Pol II was incu-
bated with CAK in the presence of wild-
T at increases theability of CAK
GST –CT D fu-
type (T at) or mutant (mT at = K41A) T at proteins in a kinase reaction. Products were processed as described above. (D) CyclinA?171/
CDK2HA complexes bound to protein-A–Sepharose beads were incubated with CAK (50 ng) in the presence of wild-type (T at) or
mutant (mT at = K41A) T at proteins. After washing the beads were incubated with histone H1 and kinase reactions allowed to proceed
for 1 hr. Products were processed as described above. Position of labeled histone H1 (H1) is indicated.
tion of the CT D by recombinant CAK. (A) Increasing concen-
trations of a mutant CDK2 peptide(mC2p) wereadded to kinase
reactions containing GST –CT D, and CAK in the presence (+) or
absence (−) of T at. Casein kinase was used as a control for the
mobility of the labeled GST –CT D (lane 1). Kinase reactions
were processed as described in Fig. 4. Positions of the hypophos-
phorylated (CT Da) and hyperphosphorylated (CT Do) GST –CT D
fusions, as well as cyclin H are indicated on the left. (B) T he
randomized CDK2 peptide (rC2p) was added to kinase reactions
as described in A. T he labeled GST –CT D fusion protein is pre-
A CDK2 mutant peptide inhibits the phosphoryla-
T at binds to and activates CAK
GENES & DEV ELOPMENT2649
cess of mC2p (1.3 µM) inhibited the kinase activity of
CAK by 50%, whereas a 650-fold excess of the peptide
(13 µM) abolished almost all the activity of CAK. Inter-
estingly, mCp inhibited both T at-dependent and T at-in-
dependent activity of CAK, as well as the phosphoryla-
tion of cyclin H. A randomized peptide (rC2p) having the
same net charge and solubility as mC2p had no effect on
the kinase activity of CAK (Fig. 5B). T he mutant peptide
(mC2p) did not affect the ability of CDK8/cyclin C to
phosphorylate the CT D, or the activity of cyclinA?171/
CDK2HA (data not shown).
The mutant peptide inhibits Tat transactivation
Having demonstrated that T at binds to CDK7, and that
this interaction increases the ability of CAK to phos-
phorylate the CT D, we next examined more closely the
roleof CAK in T at transactivation. Wetook advantageof
the fact that mC2p could inhibit the CDK7-mediated
kinase activity of CAK to examine the effect of this pep-
tide on the function of T at in vitro. As shown in Figure
6A, the addition of recombinant T at to transcription re-
actions containing the wild-type HIV LT R as the tem-
plate stimulated transcription 10-fold compared with
basal levels (Fig. 6A, lanes 1,2). Increasing concentra-
tions of mC2p selectively inhibited T at transactivation.
At a concentration of 65 µM, mC2p inhibited T at trans-
activation to basal levels (Fig. 6A, lanes 9,10), whereas at
130 µM, transcription was abolished completely (Fig. 6A,
lane 12). As a control, the randomized rC2p peptide did
not affect T at transactivation (Fig. 6B). In contrast, tran-
scription in the absence of T at was relatively unaffected
by mC2p or rC2p, even at high concentrations of the
peptide (130 µM) (Fig. 6A and B, cf. lane 1 with 11). T he
slight increase in basal transcription at low concentra-
tions of mC2p or rC2p (Fig. 6, cf. lanes 5, 7, or 9 with 1
in A and B) suggests that these peptides might also have
a nonspecific stabilizing effect on basal transcription. It
is surprising that T at transactivation in vitro is less sen-
sitiveto mC2pconcentrations than CAK kinaseactivity.
Perhaps high levels of CAK in nuclear extracts, or the
nonspecific binding of the peptide to other proteins in
these extracts, can explain the relatively high concentra-
tions of mC2p required to affect T at activity in the in
vitro transcription reactions. Furthermore, it is also pos-
sible that CAK is less accessible to the peptide when
complexed with T FIIH.
T o control for the possibility that mC2p might affect
other transcription factors nonspecifically, we tested the
effect of the peptide on transcription from adenovirus
major late (AdML) and DHFR promoters. T hese promot-
ers represent important controls because it has been
demonstrated that transcription from the AdML pro-
moter is independent of CDK7, whereas transcription
from the DHFR promoter is absolutely dependent on the
kinase activity of CDK7 (Akoulitchev et al. 1995). As
shown in Figure 6C, increasing concentrations of mC2p
had no effect on the transcription from the AdML pro-
moter. In contrast, mC2p inhibited the transcription
from the DHFR promoter by more than 50% at a con-
centration of 65 µM and completely at a concentration of
130 µM (Fig. 6D). T he randomized peptide had no effect
on transcription from the DHFR promoter (Fig. 6E).
T hese results, together with studies on the inhibition of
the kinase activity, suggest that the phosphorylation of
its T at transactivation. (A) In vitro tran-
scription reactions were done with linear-
ized DNA templates containing the wild-
type HIV LT R sequences in the presence
(+) or absence (−) of T at and increasing con-
centrations of the CDK2 mutant peptide
(mC2p). Runoff transcripts (750 nucleo-
tides) were resolved on a 5% polyacryl-
amide/urea sequencing gel and exposed to
x-ray film after drying. (B) Increasing con-
centrations of the randomized CDK2 pep-
tide(rC2p) wereaddedto in vitro transcrip-
tion reactions containing the wild-type
HIV LT R sequences in the presence (+) or
absence (−) of T at. Runoff transcripts were
processed as described in A. (C) Increasing
concentrations of the mutant CDK2 pep-
tide (mC2p) were added to in vitro tran-
scription reactions containing linearized
DNA templates containing the AdML promoter. Lane 7 was identical to lane 1 except that reactions were done in the presence of
?-amanitin (?A = 2 µg/ml). Runoff transcripts (390 nucleotides) were processed as described in A. (D) Increasing concentrations of the
mutant CDK2 peptide (mC2p) were added to in vitro transcription reactions containing linearized DNA templates containing the
DHFR promoter as described in the experimental procedures. Lane 7 was identical to lane 1 except that reactions were done in the
presence of ?-amanitin (?A = 2 µg/ml). Runoff transcripts (390 nucleotides) were processed as described in A. T he doublet corresponds
to alternate transcription start sites. (E) Increasing concentrations of the randomized CDK2 peptide (rC2p) were added to in vitro
transcription reactions as described in D.
A CDK2 mutant peptide inhib-
Cujec et al.
2650GENES & DEV ELOPMENT
the CT D by CDK7 is required for T at transactivation in
CDK7 is required for Tat transactivation in vivo
T o confirm that the kinase activity of CDK7 is required
for T at transactivation in vivo, increasing amounts of
the mC2p peptide were co-electroporated into COS
cells, in the presence or absence of T at (see Materials
(pHIV?KBCAT ) consisted of the HIV LT R containing se-
quences encoding T AR, the initiator, the T AT A box, and
three Sp1-binding sites. Because transcription from the
HIV LT R can be activated through NF-?B-binding sites
during electroporation, the reporter construct lacked
these sites (T ong-Starksen et al. 1987). Levels of specific
transcripts were determined by the RNase protection as-
say. As documented extensively by our laboratory (Kao
et al. 1987; Selby et al. 1989; Lu et al. 1993) and those of
others (Ratnasabapathy et al. 1990; Sheldon et al. 1993),
transcription from the HIV LT R in the absence of T at
gives rise to primarily short, nonpolyadenylated tran-
scripts ∼55–59 nucleotides in length. On the other hand,
in the presence of T AT , transcription is highly proces-
sive resulting in full-length polyadenylated transcripts
that protect an 80-nucleotide-long RNA probe in our
RNase protection assays. Consequently, although the
levels of short transcripts are a measure of transcription
initiation rates, the amount of long transcripts can be
used to estimate the efficiency of transcription elonga-
tion. Importantly, because T at does not affect transcrip-
tion initiation rates, short transcripts also serve as useful
internal controls for transfection efficiency and subse-
quent RNA manipulations. As presented in Figure 7A,
the production of long transcripts from the HIV LT R was
reduced dramatically in the presence of low concentra-
tions of mC2p (5 µM), and abolished completely at a pep-
tide concentration of 10 µM. In contrast, the production
of short transcripts in the presence or absence of T at was
unaffected by mC2p, even at a peptide concentration of
20 µM. Although overall transcript levels appearedhigher
in the absence of the randomized peptide (Fig. 7B), the
ratio of long to short transcripts remained the same re-
gardless of the peptide concentration. Similarly, increas-
ing concentrations of rC2p had no effect on transcription
in the absence of T at (data not shown).
T o examine further the role of CDK7 on the activity of
T at in vivo, we overexpressed wild-type or a kinase-de-
ficient mutant (D155A) of CDK7 in COS cells. RNase
protection assays were used to quantify levels of tran-
scripts from the HIV LT R or from an enhancerless pro-
moter consisting of only a T AT A box and four Sp1-bind-
ing sites (4XSp1). T his promoter (4XSp1) was shown pre-
viously to beindependent of theCT D (Gerber et al. 1995)
and consequently should not require the kinase activity
of CDK7. T ransfection of HA-tagged T at increased dra-
matically the ratio of long transcripts (LT ) to short tran-
scripts (ST ) (Fig. 8, top panel). As expected, T at did not
affect transcription from the 4XSp1 promoter (Fig. 8,
middle panel). T he cotransfection of the wild-type
T he reporter construct
CDK7–HA construct resulted in significant increase in
the ratio of long transcripts (LT ) to short transcripts (ST )
compared with cells expressing only the endogenous
CDK7. In contrast, the cotransfection of the kinase-de-
ficient mutant of CDK7 (Mut) inhibited the production
of long transcripts to levels below those observed in cells
expressing endogenous CDK7, but had relatively little
effect on the production of short transcripts. As ex-
pected, overexpression of wild-type or mutant CDK7
proteins had no effect on transcription from the 4XSp1
promoter. As additional controls, Western blotting re-
vealed that similar levels of T at–HA and CDK–HA pro-
teins wereexpressedin theseexperiments (Fig. 8, bottom
panels). T ogether with previous data, these results con-
firm that the transcriptional activation by T at in vivo is
dependent on the kinase activity of CDK7. In contrast,
the synthesis of short transcripts from the HIV promoter
is independent of the kinase activity of CDK7.
Our results indicate that T at binds to CDK7 directly and
tion in vivo. (A) Increasing concentrations of the mutant CDK2
peptide (mC2p) were cotransfected into COS cells with a re-
porter plasmid containing HIV LT R promoter sequences that
lacked NF-?B-binding sites (pHIV?KBCAT ) and either func-
tional (+)or nonfunctional (−)T at. RNaseprotection assays were
done with a probe 220 nucleotides long and hybridized to full-
length transcripts of 80 nucleotides (LT ) or prematurely termi-
nated transcripts (ST ) of 55–59 nucleotides (Okamoto et al.
1996). Protected fragments were resolved on a 11% polyacryl-
amide/urea sequencing gel and exposed to x-ray film after dry-
ing. (B) Increasing concentrations of the randomized CDK2 pep-
tide (rC2p) were cotransfected into COS cells with a reporter
plasmid containing the HIV LT R lacking NF-kB-binding sites
(pHIV?KBCAT ) and functional T at. RNase protection assays
were done as in A.
T he mutant CDK2 peptide inhibits T at transactiva-
T at binds to and activates CAK
GENES & DEV ELOPMENT2651
that this interaction mediates the association between
T at and CAK or T FIIH. Furthermore, T at stimulated the
ability of recombinant CAK, as well as immunoprecipi-
tated CAK /T FIIH complexes to phosphorylate the free
CT D and purified Pol II. Finally, whereas basal transcrip-
tion from the HIV LT R was independent of CDK7, the
phosphorylation of the CT D by CDK7 was absolutely
essential for T at transactivation in vitro and in vivo. T o-
gether, our results reveal a cellular target of the activa-
tion domain of T at, suggest that the interaction between
T at and its cellular counterpart is critical for the func-
tion of T at, and suggest a model by which T at increases
the processivity of Pol II.
T at was reported previously to bind to the p62 subunit
of T FIIH (Blau et al. 1996; ParadaandRoeder 1996). How-
ever in our hands, T at mutants used in this study (C30G,
K41A)boundto p62 as well as thewild-typeT at (datanot
shown). Instead, we found a strong and specific interac-
tion between T at and CDK7. T his observation was sup-
ported by the finding that T at bound to the recombinant
CAK in vitro with high affinity and specificity. Despite
this result, it is possible that additional contacts be-
tween T at and other subunits of T FIIH stabilize further
the binding of T at to T FIIH in vivo. Furthermore, our
finding that T at can bind to T FIIH in vivo is consistent
with our previous results demonstrating an interaction
between T at and the Pol II holoenzyme (Cujec et al.
Our initial experiments confirmed other studies that
demonstrated that T at can increase the ability of T FIIH
to phosphorylate the CT D of Pol II (Parada and Roeder
1996; Garcia-Martinez et al. 1997). We extend these find-
ings by demonstrating that T at can also increasetheabil-
ity of CAK to phosphorylate the CT D and Pol II. Al-
though CAK was more efficient at phosphorylating free
CT D than Pol II (compare the ratio of substrate phos-
phorylation with cyclin H phosphorylation in Figs. 4A
and B), the effect of T at was evident in both cases. It is
possible that when associated with Pol II, the CT D is
relatively inaccessibleto CAK. Following theinteraction
of Pol II with T FIIH/T FIIE, SRBs, or T AR and T AR-bind-
ing proteins, the CT D of Pol II might become more ac-
cessible to phosphorylation. Although Gaynor and col-
leagues (Garcia-Martinez et al. 1997) reported recently
that T at has only a low affinity for CAK and that it fails
to potentiate its kinase activity, these experiments were
done using crude fractions from gel filtration columns
and the authors themselves suggested that inhibitors in
their fractions could mask potential interactions be-
tween T at and CAK.
T he CT D of Pol II has an important role in Pol II tran-
scription (Dahmus 1995), in premRNA
(Yuryev et al. 1996) and polyadenylation (McCracken et
al. 1997). T he phosphorylation of the CT D by T FIIH has
an important role in the processivity of Pol II (Mal-
donado and Reinberg 1995). Evidence supporting this hy-
pothesis comes from the observation that the CT Ds of
polymerases paused on the Drosophila hsp 70 promoter
are relatively unphosphorylated, whereas those of ac-
tively elongating polymerases are high phosphorylated
(O’Brien et al. 1994), and the fact that the inhibition of
T FIIH abolishes transcriptional elongation (Yankulov et
al. 1995; Akhtar et al. 1996). T herefore, it is remarkable
that our results demonstrating that T at potentiates the
kinase activity of CAK illuminates further the pivotal
role of T FIIH in phosphorylating the CT D and ensuring
the processivity of Pol II. T he mechanism by which
phosphorylation of the CT D allows for increased proces-
sivity of Pol II is unknown. However, onemight envision
that the phosphorylation of the CT D results in confor-
mational changes in Pol II, which in turn facilitate the
escape of the DNA-bound polymerase from the as-
sembled pre-initiation complex, increase the activity of
its catalytic subunit, and allow for the binding of new
transcription elongation factors.
Previous attempts to demonstrate that the kinase ac-
tivity of T FIIH is required for T at transactivation have
centered on the use of kinase inhibitors (H8), AT P ana-
logs (DRB), or immunodepletion experiments (Herrmann
and Rice 1995; Parada and Roeder 1996). However, these
in vivo. COS cells were cotransfected with HIVSCAT and
4XSp1?glob reporter plasmids, and plasmids encoding HA-
tagged functional (+) or nonfunctional (−) T at, either alone or in
combination with wild-type HA-tagged CDK7 (WT ) or a kinase-
deficient HA-tagged CDK7 (Mut). RNaseprotection assays were
done using a HIV LT R probe that hybridized to full-length LT R–
CAT transcripts of 80 nucleotides (LT ) or prematurely termi-
nated transcripts (ST ) of 55–59 nucleotides (top panel) and a 216
nucleotide ?-globin probe that hybridized to transcripts of 179
nucleotides (4XSp1) (Okamoto et al. 1996). (Bottom two panels)
Lysates from transfected cells were subjected to SDS-PAGE,
transferred to membranes and probed with ?HA antibodies to
detect T at–HA and CDK7–HA protein levels.
Over-expression of CDK7 affects T at transactivation
Cujec et al.
2652GENES & DEV ELOPMENT
approaches were limited by the absence of specific in-
hibitors to CDK7, and the possibility that the immuno-
depletion of T FIIH removed additional proteins or activi-
ties (3? → 5? helicaseof ERCC3) required for thefunction
of T at. Although pseudosubstrates have been used pre-
viously to inhibit kinases such as protein kinase C (Po-
teet-Smith et al. 1997), we believe ours is the first report
of a peptide being used to inhibit the enzymatic activity
of a GT F. T he use of a mutant CDK2 peptide to inhibit
specifically thekinaseactivity of CDK7 was validated by
thefollowing: (1) theinhibition of theactivity of CAK by
mC2p in in vitro kinase assays; (2) the inhibition of tran-
scription from the CDK7-dependent DHFR promoter; (3)
the observation that mC2p did not affect the transcrip-
tion from the CDK7-independent AdML promoter; and
(4) theobservation that mC2phadno effect on thekinase
activity of either CDK8/cyclin C
CDK2HA. Consequently, we used this peptide to dem-
onstrate that phosphorylation of the CT D by CDK7 is
required for the function of T at in vitro and in vivo.
Overexpression of the wild-type and kinase-deficient
CDK7 proteins in COS cells confirmed that CDK7 is
essential for the production of long transcripts from the
HIV LT R. T ogether these results extend our previous
observations demonstrating that the CT D is required for
T at transactivation (Okamoto et al. 1996), and firmly
establish that the HIV LT R is an enhancer-dependent
Based on these results we propose a model for how the
function of T at increases the processivity of Pol II (Fig.
9). T at enters into the transcription complex by virtue of
its association with the Pol II holoenzyme before dock-
ing of the complex onto the DNA template. Pol II then
copies promoter proximal sequences and T AR is synthe-
Following its interaction with T AR, T at is reposi-
tioned and/or modified such that it increases the ability
of CDK7 to phosphorylate the CT D. T at might function
in a catalytic manner by modifying T FIIH, it may repo-
sition T FIIH in closer proximity of the CT D, or it might
increase the length of time that T FIIH remains in con-
tact with the CT D. In this regard it is intriguing that
T FIIH remains associated with Pol II for only a relative
short period of time following promoter clearance by Pol
II (Zawel et al. 1995) and that core Pol II has been re-
ported to interact with both T at (Mavankal et al. 1996)
and T AR (Wu-Baer et al. 1995). Although our data is
entirely consistent with the one-step recruitment of T at
to the transcription complex via its interaction with the
Pol II holoenzyme, it does not preclude the possibility
that additional T at is also recruited by nascent T AR
(Keen et al. 1996). In this manner, other CT D kinases
that are not part of the Pol II holoenzyme might be
brought into the complex (Herrmann and Rice 1995;
Yang et al. 1996). Synergistic effects of multiple kinases
might ensure efficient phosphorylation of the Pol II CT D
and optimal transcription elongation. T he CT D of Pol II
remains unphosphorylated despite the fact that T FIIH is
present in the Pol II holoenzyme (Ossipow et al. 1995;
Chao et al. 1996; Maldonado et al. 1996). Presumably
conformational changes in T FIIH, Pol II, and perhaps
other transcription factors are required before the kinase
activity of T FIIH can be activated (Laybourn and Dah-
mus 1990; Peterson et al. 1992). T heinability of T FIIH to
phosphorylate the CT D before docking of the polymer-
ase onto the DNA template would not be affected by the
presence of T at. T he mechanism by which T at increases
theprocessivity of Pol II establishes important principles
of transcriptional elongation, and suggests a mechanism
by which pause sites in other promoters (c-myc) might
be circumvented (Wright et al. 1994). Future work on
template commitment, the ability of T AR decoys to de-
which T at increases the processivity of Pol
II. As depicted above, transcription from the
HIV LT R can be divided into four stages: (1)
assembly of the transcription complex onto
the DNA template; (2) initiation of tran-
scription; (3) promoter clearance; and (4)
elongation of Pol II along theDNA template.
In this schematic, the DNA template is de-
picted by a thick black line, the unphos-
phorylated form of the core Pol II (IIA) by an
open oval with its unphosphorylated CT D
as a thin curved line, and the nine subunits
T FIIH as striped forms with the CDK7 as a
solid circle. T at is designated as an open
A model of the mechanism by
circle and the transcription initiation site with an arrow. During assembly of the initiation complex, T at, T FIIH and other components
of the Pol II holoenzyme (omitted for the sake of simplicity) associate with the core Pol II onto the DNA template. T ranscription
initiates with the hydrolysis of AT P between the ? and ? phosphates and the synthesis of the first phosphodiester bond. Nascent RNA
(line with a large dot at its 5? terminus) is synthesized following the promoter clearance. T he CT D might be partially phosphorylated
at this point and is depicted by a curved line with a partially thickened section. Following synthesis of T AR, T at binds to the bulge
region and is repositioned or modified such that it can increase the ability of T FIIH to phosphorylate the CT D. T he highly phos-
phorylated form of Pol II (IIo), depicted as a shaded oval with its CT D as a thick curved line, is therefore rendered highly processive
and can now efficiently transcribe the entire viral genome.
T at binds to and activates CAK
GENES & DEV ELOPMENT2653
plete T at at different stages of transcription, and the
mapping of surfaces on CDK7 that interact with T at will
reveal further mechanistic details by which interactions
between T at and T AR stimulate the ability of T FIIH to
phosphorylate the CT D of Pol II.
Material and methods
T he constructs pCMV–T AT HA(T at) and pCMV–T AT (C30G)-
HA containing wild-type or mutant T at (C30G) fused to the
influenza virus HA epitope tag (3?) were described previously
(Cujec et al. 1997). A second HA-tagged mutant of T at (K41A)
was constructed by PCR-mediated mutagenesis. Briefly, PCR
primers K41A (CAT T GCT ACGCGT GT T T CACAAGAgccG-
GCT T AGGC, lowercase letters denote mutation) and T AT 3
(CAGT CT GAGT AGT T CGAAGAGT AG) were used to amplify
a112-bpfragment of T at that was then clonedintotheAflIII and
HindIII sites of pCMV–T AT HA. Both mutations (C30G, K41A)
are in the activation domain of T at and render T at inactive
without affecting its RNA-binding ability or protein expression
levels (Kuppuswamy et al. 1989; Fig. 1). T he T at constructs (5
µg) were transfected into COS-7 cells by liptofectin (10 µl) ac-
cording to the manufacturer’s recommendations (GIBCO BRL,
Gaithersburg, MD). Approximately 36 hr after transfection the
cells were lysed [50 mM HEPES–KOH at pH 7.8, 0.5 M NaCl 1%
T riton X-100, 10 mM EDT A, 5 mM dithiothreitol (DT T ), 0.1 mM
phenylmethylsulfonyl fluoride (PMSF), 20 µg/ml of aprotinin
10 µg/ml of leupeptin], and the supernatants immunoprecipi-
tated with the indicated antibodies. Immunoprecipitates bound
to protein-A–Sepharose beads were washed three times in CAK-
binding buffer (lysis buffer + 10% glycerol). Washed beads were
subjected to gradient (5%–15%) SDS-PAGE, transferred to Im-
mobilon-NC membranes (Millipore, Bedford, MA) and reacted
with the indicated antibodies. T he proteins were visualized by
enhanced chemiluminescence detection (Amersham, Arlington
Heights, IL). T he anti-HA antibody was purchased from Boe-
hringer Mannheim (Indianapolis, IN), the CDK7 antibody from
Santa Cruz Biotechnology (Santa Cruz, CA), and the cyclin H
antibody from Upstate Biotechnology (Lake Placid, NY) and the
Pol II antibody (?RPB 1) from Promega (Madison, WI). T he
CIIT A antibody was from our laboratory (Steimle et al. 1993).
Purification of recombinant CAK trimer
Recombinant baculoviruses encoding human CDK7 (Fisher and
Morgan 1994), MAT 1 (Fisher et al. 1995), and an amino-termi-
nally 6-Histidine-tagged version of cyclin H (Kim et al. 1996)
were constructed as described. Sf9 insect cells (4 × 109) were
coinfected with all three baculoviruses (multiplicity of infec-
tion for each virus: 5 PFU/cell), incubated 2 days at 28°C, and
harvested by centrifugation. Cells were resuspended in 120 ml
of buffer A (20 mM Na phosphate, 25 mM NaCl, 1 mM PMSF, 1
µg/ml of leupeptin, 2 µg/ml of aprotinin, 1 mM DT T at pH 7.4).
NaCl concentration was raised to 300 mM and the lysate was
clarified by centrifugation and loaded over tandem 5-ml Phar-
macia HiT rap chelating columns loaded with CoCl2and pre-
equilibrated with buffer B (300 mM NaCl, 20 mM Na phosphate,
10% glycerol at pH 7.2). T he column was washed with buffer
B + 1 mM DT T and eluted with a linear gradient of 0–200 mM
imidazole in buffer B + 1 mM DT T . Fractions containing the
CAK trimer were pooled, diluted fourfold in buffer C (20 mM
HEPES–NaOH, 1 mM EDT A, 10% glycerol, 1 mM DT T at pH
7.4), and loaded on a 5-ml Pharmacia HiT rap Q column pre-
equilibrated with buffer C. T he column was washed with buffer
C and eluted with a linear gradient of 25–1000 mM NaCl in
buffer C. Fractions containing the CAK trimer were pooled and
subjected to gel filtration on a 125-ml Pharmacia Superdex 200
column pre-equilibrated in buffer C + 150 mM NaCl. Peak frac-
tions were concentrated by ion exchange on a 1-ml Pharmacia
HiT rap Q column. T he pure CAK trimer (2 mg/ml; >99% ho-
mogeneous) was stored at −80°C.
In vitro binding assays
For the CAK-binding assays, wild-type T at (ST at) or mutant T at
(T atK41A) proteins containing at their 3? ends both a phos-
phorylation site for the cAMP-dependent heart muscle kinase
and a streptavidin binding peptide, were expressed in bacteria
and bound to streptavidin–agarosebeads as described previously
(Cujec et al. 1997). Equilibrated T at streptavidin-agarose beads
were incubated with 250 µg of nuclear extract or 50 ng of puri-
fied CAK as described above. Pelleted beads were washed four
times in CAK-binding buffer (see above), and processed as de-
scribed in the immunoprecipitation protocol. For the CDK7-
binding assays, T at was eluted from the streptavidin–agarose
beads with 0.8 M NaCl and 2 mM biotin and dialyzed into buffer
D (25 mM HEPES–KOH at pH 7.6, 0.1 M KCl, 20% glycerol, 10
mM DT T , 0.1 mM EDT A). Approximately 50 ng of eluted T at
was labeled using 10 units of catalytic subunit of cAMP-depen-
dent heart muscle kinase (Sigma P-2645) and 20 µCi of [?-
32P]AT P. Kinase reactions were performed in 50 µl of 0.1 M T ris
at pH 7.5, 5 mM DT T , 0.5 M NaCl, 60 mM MgCl2. CDK7, cyclin
H, and MAT 1 proteins were labeled with L-[35S]methionine
(>1000Ci/mmole; Amersham) and the Promega T NT protein
expression system. CAK complexes were formed in association
buffer (20 mM HEPES–KOH at pH 7.6, 50 mM KCl, 10 mM DT T ,
5 mM EDT A, and 10% glycerol) for 1 hr at 4°C and then immu-
noprecipitated with ?CDK7 antibodies attached to protein-A–
Sepharose beads. After binding, the beads were washed three
times in the association buffer and then two times in CAK-
binding buffer (see above). Labeled T at (0.5 ng) was added to the
CDK7 complexes and incubated for 1 hr at 25°C. Beads were
washed again with binding buffer and then with PBS before
loading onto SDS-PAGE (5%–20% gradient). After drying, the
gels were visualized by autoradiography.
GST –CT D fusion proteins were expressed in Escherichia coli
using pGCT D (a generous gift of W. Dynan) as described (Peter-
son et al. 1992). Fusion proteins were eluted from glutathione–
Sepharose beads with 15 mM glutathione and purified by gel
filtration on a S-300 column (Pharmacia, Piscataway NJ). Ap-
proximately 25 ng of the eluted GST –CT D fusion was used in
each kinase reaction. Purified preparations of Pol II (a generous
gift of C. Kane) were obtained as described (Hodo and Blatti
1977; Kerppola and Kane 1990) except that a Mono S column
was used instead of aDEAE–5PW column in thefinal step of the
purification. CAK /T FIIH complexes were immunoprecipitated
from HeLa-cells that stably expressed HA-tagged wild-type
CDK7 or a kinase-deficient mutant (D155A) under the control
of a tetracycline-repressible promoter (Jin et al. 1996). Cells
were lysed (50 mM HEPES–KOH at pH 7.6, 150 mM NaCl, 5 mM
EDT A, 0.1% T riton X-100, 5 mM DT T , 0.2 mM PMSF, 1 mM
NaF, 0.1 mM NaVO410 µg/ml of aprotinin, 1 µg/ml leupeptin)
and immunoprecipitations done as described above. T ypically,
wild-type CAK /T FIIH complexes were immunoprecipitated
from four 150-mm culture dishes after four days of growth in
the absence of tetracycline (10 mg/ml). Immunoprecipitated
Cujec et al.
2654 GENES & DEV ELOPMENT
beads were washed three times with lysis buffer and then two
times with CT D–kinase buffer (20 mM T ris at pH 7.6, 50 mM
KCl, 5 mM MgCl2, 2.5 mM MnCl2, 10 mM DT T ). Reactions were
supplemented with 10 µCi of [?-32P]AT P and 50 µM of unlabeled
AT P in a final reaction volume of 50 µl. Reactions were incu-
bated for 1 hr at 30°C. In some experiments, recombinant CAK
(50 ng) or casein kinase II (25 ng) (Upstate Biotechnology) was
used as the source of kinase activity. Peptide concentrations
were determined by the ESL Protein Assay (Boehringer Mann-
heim) system. T he sequences of mC2p and rC2p are ARAF-
GVPVRT YaHEVVT LWYRA (lowercase letter denotes residue
mutated from threonine) and HART VGVWYRAEYARFVT -
PaVV, respectively. Histone H1 kinase assays were done as re-
ported previously (Fisher and Morgan 1994).
In vitro transcription reactions
Run-off transcription reactions from thewild-typeHIV LT R and
AdML promoters linearizedwith NcoI andHindIII, respectively,
were carried out as described (Okamoto et al. 1996; Cujec et al.
1997). T ranscription reactions containing the DHFR promoter
(500 ng) fused to a G-less cassette (linearized at NcoI) were
supplemented with 3 mM (NH4)2SO4, 2% PEG 8000, 50 units of
T 1 RNase(Boehringer Mannheim), 500 units of RNaseinhibitor
(Boehringer Mannheim) and contained 5 µM instead of 40 µM of
unlabeled UT P. Phosphocreatine, poly[d(I-C)], and poly[r(I-C)]
were omitted from the reactions.
RNase protection assays
For the peptide inhibition studies 1 × 107COS-7 cells were elec-
troporated (500 µl) (Bio-Rad, Hercules, CA) at 210 V, 960 mF,
using 2 µg of reporter DNA (pHIV?KBCAT ) containing HIV
LT R sequences lacking NF-kB-binding sites, 2 µg of effector
DNA (pSVT AT or pSVT AT ZX) and varying concentrations of
mC2p or rC2p (1 µg/µl of solution). For the CDK7 overexpres-
sion studies, COS-7 cells were transfected with lipofectin using
2 µg of reporter DNA (HIVSCAT and4XSp1), and2 µg of effector
DNA (pSVT AT or pSVT AT ZX) as described above. Vector alone
or plasmids encoding HA-tagged wild-type CDK7 (SR?–CDK7–
HA) or mutant HA-tagged CDK7 [SR?–CDK7(D155A)–HA]
were cotransfected as indicated. Cells were incubated in the
Opti-MEM medium for 5 hr and harvested 48 hr after transfec-
tion. T wenty micrograms of RNA was used for the RNase pro-
tection assays. T o make the rabbit ?-globin or the HIV LT R
CAT probe, Sp6?T S or pGEMI/WT vectors werelinearizedwith
EcoRI andtranscribedwith Sp6 or T 7 polymerases, respectively,
to produce [?-32P]UT P-labeled RNA probes. Assays were per-
formed as described (Okamoto et al. 1996), the protected frag-
ments were separated on 11% polyacrylamide/urea sequencing
gels and processed as outlined above.
We thank Michael Armanini for excellent secretarial assistance
and members of our laboratories for comments on the manu-
script. We thank Sasha Akoulitchev, Ron Drapkin, and Danny
Reinberg for antibodies, T FIIH preparations, plasmids, and help-
ful suggestions. We are grateful to Ken Sakurabayashi and W.
Dynan for the pGCT D construct, and to Caroline Kane and
Rodney Weilbaecher for the purified core Pol II preparations.
T .P.C. was funded by a Fellowship from the University-wide
T askforce on AIDS.
T he publication costs of this article were defrayed in part by
payment of page charges. T his article must therefore be hereby
marked ‘‘advertisement’’ in accordance with 18 USC section
1734 solely to indicate this fact.
Akhtar, A., G. Faye, and D.L. Bentley. 1996. Distinct activated
and non-activated RNA polymerase II complexes in yeast.
EMBO J. 15: 4654–4664.
Akoulitchev, S., T .P. Makela, R.A. Weinberg, and D. Reinberg.
1995. Requirement for T FIIH kinaseactivity in transcription
by RNA polymerase II. Nature 377: 557–560.
Alonso, A., D. Derse, and B.M. Peterlin. 1992. Human chromo-
some12 is requiredfor optimal interactions between T at and
T AR of human immunodeficiency virus type 1 in rodent
cells. J. Virol. 66: 4617–4621.
Blau, J., H. Xiao, S. McCracken, P. O’Hare, J. Greenblatt, and D.
Bentley. 1996. T hree functional classes of transcriptional ac-
tivation domains. Mol. Cell. Biol. 16: 2044–2055.
Carroll, R., B.M. Peterlin, and D. Derse. 1992. Inhibition of hu-
man immunodeficiency virus type 1 T at activity by coex-
pression of heterologous trans-activators. J. Virol. 66: 2000–
Chao, D.M., E.L. Gadbois, P.J. Murray, S.F. Anderson, M.S.
Sonu, J.D. Parvin, and R.A. Young. 1996. A mammalian SRB
protein associated with an RNA polymerase II holoenzyme.
Nature 380: 82–85.
Chiang, C.-M. and R.G. Roeder. 1995. Cloning of an intrinsic
human T FIID subunit that interacts with multipletranscrip-
tional activators. Nature 267: 531–536.
Chun, R.F., and K.T . Jeang. 1996. Requirements for RNA poly-
merase II carboxyl-terminal domain for activated transcrip-
tion of human retroviruses human T -cell lymphotropic virus
I and HIV-1. J. Biol. Chem. 271: 27888–27894.
Cismowski, M.J., G.M. Laff, M.J. Solomon, and S.I. Reed. 1995.
KIN28 encodes a C-terminal domain kinase that controls
mRNA transcription in Saccharomyces cerevisiae but lacks
cyclin-dependent kinase-activating kinase (CAK) activity.
Mol. Cell. Biol. 15: 2983–2992.
Cujec, T .P., H. Cho, E. Maldonado, J. Meyer, D. Reinberg, and
B.M. Peterlin. 1997. T he human immunodeficiency virus
transactivator T at interacts with the RNA polymerase II ho-
loenzyme. Mol. Cell. Biol. 17: 1817–1823.
Cullen, B.R. 1993. Does HIV-1 T at induce a change in viral
initiation rights? Cell 73: 417–420.
Dahmus, M.E. 1994. T he role of multisite phosphorylation in
the regulation of RNA polymerase II activity. Prog. Nucleic
Acid Res. Mol. Biol. 48: 143–179.
———. 1995. Phosphorylation of the C-terminal domain of
RNA polymerase II. Biochim. Biophys. Acta. 1261: 171–182.
———. 1996. Phosphorylation of mammalian RNA polymerase
II. Methods Enzymol. 273: 185–193.
Desai, K., P.M. Loewenstein, and M. Green. 1991. Isolation of a
cellular protein that binds to the human immunodeficiency
virus T at protein and can potentiate transactivation of the
viral promoter. Proc. Natl. Acad. Sci. 88: 8875–8879.
Drapkin, R. and D. Reinberg. 1994. T he multifunctional T FIIH
complex and transcriptional control. Trends Biochem. Sci.
Drapkin, R., G. LeRoy, H. Cho, S. Akoulitchev, and D. Rein-
berg. 1996. Human cyclin-dependent kinase-activating ki-
nase exists in three distinct complexes. Proc. Natl. Acad.
Sci. 93: 6488–6493.
Feaver, W.J., O. Gileadi, Y. Li, and R.D. Kornberg. 1991. CT D
kinase associated with yeast RNA polymerase II initiation
factor b. Cell 67: 1223–1230.
T at binds to and activates CAK
GENES & DEV ELOPMENT2655
Feaver, W.J., J.Q. Svejstrup, N.L. Henry, and R.D. Kornberg.
1994. Relationship of CDK-activating kinase and RNA poly-
merase II CT D kinase T FIIH/T FIIK. Cell 79: 1103–1109.
Fisher, R.P. and D.O. Morgan. 1994. A novel cyclin associates
with MO15/CDK7 to form the CDK-activating kinase. Cell
Fisher, R.P., P. Jin, H.M. Chamberlin, and D.O. Morgan. 1995.
Alternativemechanisms of CAK assembly requirean assem-
bly factor or an activating kinase. Cell 83: 47–57.
Garcia-Martinez, L.F., G. Mavankal, J.M. Neveu, W.S. Lane, D.
Ivanov, and R. B. Gaynor. 1997. Purification of a T at-associ-
ated kinase reveals a T FIIH complex that modulates HIV-1
transcription. EMBO J. 16: 2836–2850.
Gerber, H.P., M. Hagmann, K. Seipel, O. Georgiev, M.A. West,
Y. Litingtung, W. Schaffner, and J.L. Corden. 1995. RNA
polymerase II C-terminal domain required for enhancer-
driven transcription. Nature 374: 660–662.
Hart, C.E., C.Y. Ou, J.C. Galphin, J. Moore, L.T . Bacheler, J.J.
Wasmuth, S. Petteway Jr., and G. Schochetman. 1989. Hu-
man chromosome 12 is required for elevated HIV-1 expres-
sion in human-hamster hybrid cells. Science 246: 488–491.
Herrmann, C.H., and A.P. Rice. 1995. Lentivirus T at proteins
specifically associate with a cellular protein kinase, T AK,
that hyperphosphorylates the carboxyl-terminal domain of
the large subunit of RNA polymerase II: Candidate for a T at
cofactor. J. Virol. 69: 1612–1620.
Hodo III, H.G. and S.P. Blatti. 1977. Purification using polyeth-
ylenimine precipitation and low molecular weight subunit
analyses of calf thymus and wheat germ DNA-dependent
RNA polymerase II. Biochemistry 16: 2334–2343.
Hoeijmakers, J.H.J., J.-M. Egly, and W. Vermeulen. 1996. T FIIH:
A key component in multiple DNA transactions. Curr.
Opin. Genet. Dev. 6: 26–33.
Jeang, K.T ., R. Chun, N.H. Lin, A. Gatignol, C.G. Glabe, and H.
Fan. 1993. In vitro and in vivo binding of human immuno-
deficiency virus type 1 T at protein and Sp1 transcription
factor. J. Virol. 67: 6224–6233.
Jin, P., Y. Gu, and D.O. Morgan. 1996. Role of inhibitory CDC2
phosphorylation in radiation-induced G2 arrest in human
cells. J. Cell Biol. 134: 963–970.
Jones, K.A. and B.M. Peterlin. 1994. Control of RNA initiation
and elongation at the HIV-1 promoter. Annu. Rev. Biochem.
Kao, S. Y., A.F. Calman, P.A. Luciw, and B.M. Peterlin. 1987.
Anti-termination of transcription within the long terminal
repeat of HIV-1 by T at gene product. Nature 330: 489–493.
Kashanchi, F., G. Piras, M.F. Radonovich, J.F. Duvall, A. Fat-
taey, C.M. Chiang, R.G. Roeder, and J.N. Brady. 1994. Direct
interaction of human T FIID with the HIV-1 transactivator
T at. Nature 367: 295–299.
Kato, H., H. Sumimoto, P. Pognonec, C.H. Chen, C.A. Rosen,
and R.G. Roeder. 1992. HIV-1 T at acts as a processivity fac-
tor in vitro in conjunction with cellular elongation factors.
Genes & Dev. 6: 655–666.
Keen, N.J., M.J. Gait, and J. Karn. 1996. Human immunodefi-
ciency virus type-1 T at is an integral component of the ac-
tivated transcription-elongation complex. Proc. Natl. Acad.
Sci. 93: 2505–2510.
Kerppola, T .K. and C.M. Kane. 1990. Analysis of the signals for
transcription termination by purified RNA polymerase II.
Biochemistry 29: 269–278.
Kim, K.-K., H.M. Chamberlin, D.O. Morgan, and S.-H. Kim.
1996. T hree-dimensional structure of human cyclin H, a
positive regulator of the CDK-activating kinase. Nature
Struct. Biol. 3: 849–855.
Kim, Y.J., S. Bjorklund, Y. Li, M.H. Sayre, and R.D. Kornberg.
1994. A multiprotein mediator of transcriptional activation
and its interaction with the C-terminal repeat domain of
RNA polymerase II. Cell 77: 599–608.
Kladis, P., A. Sutton, and M. Solomon. 1996. T he CDK-activat-
ing kinase (CAK) from budding yeast. Cell 86: 553–564.
Kuppuswamy, M., T . Subramanian, A. Srinivasan, and G. Chin-
nadurai. 1989. Multiplefunctional domains of T at, thetrans-
activator of HIV-1, defined by mutational analysis. Nucleic
Acids Res. 17: 3551–3561.
Laspia, M.F., A.P. Rice, and M.B. Mathews. 1989. HIV-1 T at
protein increases transcriptional initiation and stabilizes
elongation. Cell 59: 283–292.
Laybourn, P.J., and M.E. Dahmus. 1989. T ranscription-depen-
dent structural changes in the C-terminal domain of mam-
malian RNA polymerase subunit IIa/o. J. Biol. Chem.
———. 1990. Phosphorylation of RNA polymerase IIa occurs
subsequent to interaction with the promoter and before the
initiation of transcription. J. Biol. Chem. 265: 13165–13173.
Liao, S.M., J. Zhang, D.A. Jeffery, A.J. Koleske, C.M. T hompson,
D.M. Chao, M. Viljoen, H.J. van Vuuren, and R.A. Young.
1995. A kinase-cyclin pair in the RNA polymerase II holo-
enzyme. Nature 374: 193–196.
Lu, H., L. Zawel, L. Fischer, J.-M. Egly, and D. Reinberg. 1992.
Human general transcription factor IIH phosphorylates the
C-terminal domain of RNA polymerase II. Nature 358: 641–
Lu, X., T .M. Welsh, and B.M. Peterlin. 1993. T he human im-
munodeficiency virus type 1 long terminal repeat specifies
two different transcription complexes, only one of which is
regulated by T at. J. Virol. 67: 1752–1760.
Madore, S.J., and B.R. Cullen. 1995. Functional similarities be-
tween HIV-1 T at and DNA sequence-specific transcriptional
activators. Virology 206: 1150–1154.
Madore, S.J. and B.R. Cullen. 1993. Genetic analysis of the co-
factor requirement for human immunodeficiency virus type
1 T at function. J. Virol. 67: 3703–3711.
Makela, T .P., J.P. T assan, E.A. Nigg, S. Frutiger, G.J. Hughes,
and R.A. Weinberg. 1994. A cyclin associated with the CDK-
activating kinase MO15. Nature 371: 254–257.
Maldonado, E. and D. Reinberg. 1995. News on initiation and
elongation of transcription by RNA polymerase II. Curr.
Opin. Cell Biol. 7: 352–361.
Maldonado, E., R. Shiekhattar, M. Sheldon, H. Cho, R. Drapkin,
J.A. Inostroza, P. Rickert, E. Lees, C.W. Anderson, S. Linn,
and D. Reinberg. 1996. A human RNA polymerase II com-
plex associated with SRB and DNA-repair proteins. Nature
Marciniak, R.A. and P.A. Sharp. 1991. HIV-1 T at protein pro-
motes formation of more processive elongation complexes.
EMBO J. 10: 4189–4196.
Mavankal, G., S.H. Ou, H. Oliver, D. Sigman, and R.B. Gaynor.
1996. Human immunodeficiency virus type 1 and 2 proteins
specifically interact with RNA polymerase II. Proc. Natl.
Acad. Sci. 93: 2089–2094.
McCracken, S., N. Fong, K. Yankulov, S. Ballantyne, G. Pan, J.
Greenblatt, S. D. Patterson, M. Wickens, and D.L. Bentley.
1997. T he C-terminal domain of RNA polymerase II couples
mRNA processing to transcription. Nature 385: 357–361.
Nelbock, P., P.J. Dillon, A. Perkins, and C.A. Rosen. 1990. A
cDNA for a protein that interacts with the human immuno-
deficiency virus T at transactivator. Science 248: 1650–1653.
O’Brien, T ., S. Hardin, A. Greenleaf, and J.T . Lis. 1994. Phos-
phorylation of RNA polymerase II C-terminal domain and
transcriptional elongation. Nature 370: 75–77.
Okamoto, H., C.T . Sheline, J. Corden, K.A. Jones, and B.M. Pe-
Cujec et al.
2656 GENES & DEV ELOPMENT
terlin. 1996. T rans-activation by human immunodeficiency Download full-text
virus T at protein requires the C-terminal domain of RNA
polymerase II. Proc. Natl. Acad.Sci. 93: 11575–11579.
Ossipow, V., J.P. T assan, E.A. Nigg, and U. Schibler. 1995. A
mammalian RNA polymerase II holoenzyme containing all
components required for promoter-specific transcription ini-
tiation. Cell 83: 137–146.
Parada, C.A. and R.G. Roeder. 1996. Enhanced processivity of
RNA polymerase II triggered by T at-induced phosphoryla-
tion of its carboxy-terminal domain. Nature 384: 375–378.
Peterson, S.R., A. Dvir, C.W. Anderson, and W.S. Dynan. 1992.
DNA binding provides a signal for phosphorylation of the
RNA polymerase II heptapeptide repeats. Genes & Dev.
Poteet-Smith, C.E., J.B. Shabb, S.H. Francis, and J.D. Corbin.
1997. Identification of critical determinants for autoinhibi-
tion in the pseudosubstrate region of type I alpha cAMP-
dependent protein kinase. J. Biol. Chem. 272: 379–388.
Ratnasabapathy, R., M. Sheldon, L. Johal, and N. Hernandez.
1990. T he HIV-1 long terminal repeat contains an unusual
element that induces the synthesis of short RNAs from vari-
ous mRNA and snRNA promoters. Genes & Dev. 4: 2061–
Reardon, J.T ., H. Ge, E. Gibbs, A. Sancar, J. Hurwitz, and Z.Q.
Pan. 1996. Isolation and characterization of two human tran-
scription factor IIH (T FIIH)-related complexes: ERCC2/CAK
and T FIIH. Proc. Natl. Acad. Sci. 93: 6482–6487.
Rossignol, M., I. Kolb-Cheynel, and J.-M. Egly. 1997. Substrate
specificity of thecdk-activating kinase(CAK) is altered upon
association with T FIIH. EMBO J. 16: 1628–1637.
Roy, R., J.P. Adamczewski, T . Seroz, W. Vermeulen, J.P. T assan,
L. Schaeffer, E.A. Nigg, J.H. Hoeijmakers, and J.M. Egly.
1994. T he MO15 cell cycle kinase is associated with the
T FIIH transcription-DNA repair factor. Cell 79: 1093–1101.
Selby, M.J., E.S. Bain, P.A. Luciw, and B.M. Peterlin. 1989.
Structure, sequence, and position of the stem-loop in T AR
determine transcriptional elongation by T at through the
HIV-1 long terminal repeat. Genes & Dev. 3: 547–558.
Serizawa, H., T .P. Makela, J.W. Conaway, R.C. Conaway, R.A.
Weinberg, and R.A. Young. 1995. Association of Cdk-acti-
vating kinase subunits with transcription factor T FIIH. Na-
ture 374: 280–282.
Sheldon, M., R. Ratnasabapathy, and N. Hernandez. 1993. Char-
acterization of the inducer of short transcripts, a human im-
munodeficiency virus type 1 transcriptional element that
activates the synthesis of short RNAs. Mol. Cell. Biol.
Shiekhattar, R., F. Mermelstein, R.P. Fisher, R. Drapkin, B.
Dynlacht, H.C. Wessling, D.O. Morgan, and D. Reinberg.
1995. Cdk-activating kinase complex is a component of hu-
man transcription factor T FIIH. Nature 374: 283–287.
Steimle, V., L.A. Otten, M. Zufferey, and B. Mach. 1993.
Complementation cloning of an MHC class II transactivator
mutated in hereditary MHC class II deficiency (or bare lym-
phocyte syndrome). Cell 75: 135–146.
T hompson, N.E., T .H. Steinberg, D.B. Aronson, and R.R. Bur-
gess. 1989. Inhibition of in vivo and in vitro transcription by
monoclonal antibodies prepared against wheat germ RNA
polymerase II that react with the heptapeptide repeat of eu-
karyotic RNA polymerase II. J. Biol. Chem. 264: 11511–
T huret, J.-Y., J.-G. Valay, G. Faye, and C. Mann. 1996. Civ
1(CAK in vivo), a novel cdk-activating kinase. Cell 86: 565–
T ong-Starksen, S.E., P.A. Luciw, and B.M. Peterlin. 1987. Hu-
man immunodeficiency virus long terminal repeat responds
to T cell activation signals. Proc. Natl. Acad. Sci. 84: 6845–
Veschambre, P., P. Simard, and P. Jalinot. 1995. Evidence for
functional interaction between the HIV-1 T at transactivator
and the T AT A box binding protein in vivo. J. Mol. Biol.
Wilson, C.J., D.M. Chao, A.N. Imbalzano, G.R. Schnitzler, R.E.
Kingston, and R.A. Young. 1996. RNA polymerase II holo-
enzyme contains SWI/SNF regulators involved in chromatin
remodeling. Cell 84: 235–244.
Wright, S., X. Lu, and B.M. Peterlin. 1994. Human immunode-
ficiency virus type 1 T at directs transcription through at-
tenuation sites within the mouse c-myc gene. J. Mol. Biol.
Wu-Baer, F., D. Sigman, and R.B. Gaynor. 1995. Specific binding
of RNA polymeraseII tothehuman immunodeficiency virus
trans-activating region RNA is regulated by cellular cofac-
tors and T at. Proc. Natl. Acad. Sci. 92: 7153–7157.
Yang, X., C.H. Herrmann, and A.P. Rice. 1996. T he human im-
munodeficiency virus T at proteins specifically associate
with T AK in vivo and require the carboxyl-terminal domain
of RNA polymerase II for function. J. Virol. 70: 4576–4584.
Yankulov, K.Y. and D.L. Bentley. 1997. Regulation of CDK7
substrate specificityby MAT 1 and T FIIH. EMBO
Yankulov, K., K. Yamashita, R. Roy, J.-M. Egly, and D. Bentley.
1995. T he transcriptional elongation inhibitor 5,6-dichlor-1-
?-D-ribofuranosylbenzimidazole inhibits transcription fac-
tor IIH-associated protein kinase. J. Biol. Chem. 270: 23922–
Yankulov, K.Y., M. Pandes, S. McCraken, D. Bouchard, and D.
Bentley. 1996. T FIIH functions in regulating transcriptional
elongation by RNA polymerase II in Xenopus oocytes. Mol.
Cell. Biol. 16: 3291–3299.
Yuryev, A., M. Patturajan, Y. Litingtung, R.V. Joshi, C. Gentile,
M. Gebara, and J.L. Corden. 1996. T he C-terminal domain of
the largest subunit of RNA polymerase II interacts with a
novel set of serine/arginine-rich proteins. Proc. Natl. Acad.
Sci. 93: 6975–6980.
Zawel, L., K.P. Kumar, and D. Reinberg. 1995. Recycling of the
general transcription factors during RNA polymerase II tran-
scription. Genes & Dev. 9: 1479–1490.
Zhou, Q. and P.A. Sharp. 1996. T at-SF1: Cofactor for stimula-
tion of transcriptional elongation by HIV-1 T at. Science
T at binds to and activates CAK
GENES & DEV ELOPMENT 2657