In this study, we systematically compared the phosphorylation of differ-ent recombinant CTD substrates by recombinant CDK7/ CycH/MAT1, CDK8/CycC and CDK9/CycT1 kinases.. A possible cause
Trang 1Three cyclin-dependent kinases preferentially phosphorylate different parts of the C-terminal domain of the large subunit of RNA
polymerase II
Reena Pinhero, Peter Liaw, Kimberly Bertens and Krassimir Yankulov
Department of Molecular Biology and Genetics, University of Guelph, Ontario, Canada
The C-terminal domain (CTD) of the largest subunit of
RNA polymerase II plays critical roles in the initiation,
elongation and processing of primary transcripts These
activities are at least partially regulated by the
phosphory-lation of the CTD by three cyclin-dependent protein kinases
(CDKs), namely CDK7, CDK8 and CDK9 In this study,
we systematically compared the phosphorylation of
differ-ent recombinant CTD substrates by recombinant CDK7/
CycH/MAT1, CDK8/CycC and CDK9/CycT1 kinases We
showed that CDK7, CDK8 and CDK9 produce different
patterns of phosphorylation of the CTD CDK7/CycH/
MAT1 generates mostly hyperphosphorylated full-length
and truncated CTD peptides, while CDK8/CycC and
CDK9/CycT1 generate predominantly hypophosphoryl-ated peptides Total activity towards different parts of the CTD also differs between the three kinases; however, these differences did not correlate with their ability to hyper-phosphorylate the substrates The last 10 repeats of the CTD can act as a suppressor of the activity of the kinases in the context of longer peptides Our results indicate that the three kinases possess different biochemical properties that could reflect their actions in vivo
Keywords: carboxy-terminal domain; cyclin-dependent kin-ase; phosphorylation; RNA pol II
The C-terminus of the largest subunit of the eukaryotic
RNA polymerase II consists of multiple repeats of a
YSPTSPS consensus heptapeptide sequence [1,2] This part
of the polypeptide is referred to as CTD (C-terminal
domain) In higher eukaryotes, the CTD consists of 52
heptapeptide repeats [1–3] The N-terminal portion of the
CTD contains mainly perfect YSPTSPS repeats; however,
the repeats in the C-terminal portion significantly deviate
from the consensus [2–5], probably reflecting a more
specialized function of this part of the polypeptide It has
been demonstrated that the N-terminal half of the CTD
supports RNA synthesis and capping of the primary
transcript [6–8], whereas the C-terminal half supports
splicing and 3¢ processing of the transcripts [6] The
importance of the C-terminal ISPDDSDEEN sequence of
the CTD in the regulation of transcript processing has also
been shown [9] The CTD is phosphorylated at multiple
sites, which leads to the production of two forms of RNA polymerase II in vivo: a hypophosphorylated form called IIa, and a hyperphosphorylated form called IIo [1,2,4,5]
It is well established that phosphorylation of the CTD regulates the transition of RNA polymerase II from initiation to elongation, the capping of primary transcripts and the efficiency of pol II elongation [1,2,10] CTD phosphorylation has also been implicated in the coscriptional splicing and polyadenylation of nascent tran-scripts [1,2,10] However, little is known about how the phosphorylation of different parts of the CTD contributes
to these functions
At least three protein kinases are involved directly in the phosphorylation of the CTD and in the regulation of different stages of mRNA synthesis [1,2] Cyclin dependent kinase (CDK)7, in conjunction with cyclin (Cyc)H and MAT1, forms a tripartite complex known as CAK (CDK-activating kinase); however, a less abundant bipartite form (CDK7/CycH) has also been observed [11] At the same time, CDK7/CycH/MAT1 has been identified as a compo-nent of the general pol II transcription factor, TFIIH [1,2], and of large protein complexes containing RNA polymerase
II and general pol II transcription factors that are referred
to as pol II holoenzyme complexes [12] Another protein kinase, CDK8/CycC, has also been found in the pol II holoenzyme [13] and in other MED/SRB containing complexes such as TRAP/SMCC and NAT [14–17] TRAP/SMCC and NAT both phosphorylate the CTD and repress activated, but not basal, transcription [17] Another study indicates that NAT and TRAP/SMCC phosphorylate CycH of the TFIIH complex via its CDK8 kinase activity and inhibit TFIIH protein kinase activity [18] Studies in Saccharomyces cerevisiae suggest that the
Correspondence to K Yankulov, Department of Molecular Biology
and Genetics, University of Guelph, Guelph, Ontario, Canada,
N1G 2W1 Fax: + 519 837 4120, Tel.: + 519 824 4120 ext 56466,
E-mail: yankulov@uoguelph.ca
Abbreviations: CAK, CDK activating kinase; CDK, cyclin dependent
kinase; CTD, C-terminal domain; Cyc, cyclin; MED, mediator;
GST, glutathione S-transferase; MBP, myelin basic protein;
MOI, multiplicity of infection; NAT, negative regulator of activated
transcription; P-TEFb, positive transcriptional elongation factor b;
SMCC, SRB/MED containing complex; SRB, suppressor of RNA
polymerase B; TRAP, thyroid hormone receptor associated
protein complex.
(Received 18 November 2003, revised 9 January 2004,
accepted 19 January 2004)
Trang 2CDK8 homolog, Srb10p, phosphorylates the CTD prior to
the formation of an initiation complex at promoters, which
results in the repression of pol II transcription [19] The third
CTD kinase, CDK9, in complex with one of several
homologous CycT molecules, has been initially identified as
P-TEFb (positive transcription elongation factor-b) [20,21]
Independently, CDK9/CycT1 has been isolated as the
HIV-tat-associated kinase, TAK It has been reported that the
P-TEFb kinase activity operates in a CAK-independent
manner [22] Unlike CDK7/CycH/MAT1 and CDK8/
CycC, which are recruited to promoters prior to
transcrip-tion initiatranscrip-tion, P-TEFb is recruited to the elongating
polymerase at a later stage of the transcription reaction
[23–26] The in vitro effects of P-TEFb on elongation cannot
be replaced by TFIIH, thus suggesting that these complexes
perform non-redundant functions [24]
Several studies have attempted to directly compare the
phosphorylation of the CTD or synthetic CTD
deriva-tives by CDK7/CycH/MAT1, CDK8/CycC and CDK9/
CycT1 [25,27–31] CDK7/CycH/MAT1 and CDK8/CycC
preferentially phosphorylate the S5 residue in the
YSPTSPS repeat [25,27–29,31] CDK7/CycH/MAT1 might
also phosphorylate some S2 residues in the less conserved
C-terminal portion of the CTD [32] CDK9/CycT1 seems
to preferentially phosphorylate the S2 residue of the
YSPTSPS repeat on longer CTD substrates [21,25];
however, it can also phosphorylate S5 on short peptide
substrates [28,31] In addition, CDK9/CycT1 can shift its
preference from S2 to S5 in the presence of the HIV-tat
protein [25]
There is a greater uncertainty as to how these kinases
phosphorylate full-length CTD and parts derived from it
One study demonstrates that the C-terminal portion of the
CTD is phosphorylated more efficiently by CDK7/CycH/
MAT1 than by CDK8/CycC [27] This effect is attributed
to the frequent presence of K in position 7 of the heptad
repeats in the C-terminal part of the CTD Indeed, synthetic
(YSPTSPK)4peptides are preferentially phosphorylated by
CDK7/CycH/MAT1 as compared to CDK8/CycC [27]
Another study, using immunoprecipitated CDK7, CDK8
and CDK9, indicates that the three kinases phosphorylate
equally well the N terminus of the CTD (repeats 1–29), but
only CDK7 is able to produce the hyperphosphorylated IIo
form of this substrate [28] The C terminus (repeats 30–52) is
efficiently phosphorylated by CDK7 only, but the
hyper-phosphorylated IIo form was not produced [28] The
authors conclude that the hyperphosphorylation, and the
production of the IIo form of pol II and the CTD, is a result
of phosphorylation of the first half of the CTD [28] On
full-length CTD, the immunoprecipitated CDK7 has a much
higher activity relative to CDK8 and CDK9 Surprisingly,
both CDK7 and CDK9 produced the full-length
hyper-phosphorylated IIo form [28]
In this study, we systematically compared the
phosphory-lation pattern of recombinant CTD substrates by
recom-binant CDK7/CycH/MAT1, CDK8/CycC and CDK9/
CycT1 kinases We showed that the three kinases do not
dramatically differ in their activity towards the CTD in vitro;
however, they displayed different abilities to
hyperphospho-rylate CTD substrates Only CDK7/CycH/MAT1 was able
to efficiently hyperphosphorylate the full-length CTD and
produce the IIo form of this substrate The N- and
C-terminal portions of the CTD were differentially phos-phorylated by CDK7/CycH/MAT1, CDK8/CycC and CDK9/CycT1 Finally, we showed data suggesting that certain CTD repeats in the context of larger polypeptides can suppress the activities of these kinases
Materials and methods
Expression vectors The baculoviruses for the expression of CDK7, MAT1, CycC and His6-CDK9/CycT1 were as described previously [21,33,34] The baculovirus for the expression of His-tagged CycH was produced by subcloning the human CycH into pBlueBac (Invitrogen) and transfecting Sf9 cells according
to the instructions of the manufacturer The baculovirus containing His6-CDK8 was produced by subcloning the human CDK8 into pFASTBACHta and using the BAC-to-BAC recombination system (Life Technologies) The plasmids for the expression of glutathione S-transferase (GST)-CTD(1–52), CTD(1–15, S5>A) and GST-CDK2 were as described previously [35] Plasmids for the expression of CTD(1–15), CTD(1–25), GST-CTD(27–39), GST-CTD(27–42) and GST-CTD(27–52) were as described previously [6] The plasmid for the expression of GST-CTD(42–52) was prepared by subclon-ing a PCR fragment, encompasssubclon-ing repeats 42–52, into pGEX2T (Amersham) GST-CTD(1–52), GST-CTD(27– 52) and GST-CTD(42–52) also contained the C-terminus ISPDDSDEEN peptide that is positioned next to the 52 heptad repeat in vertebrate RPB1
Expression and purification of recombinant kinases Recombinant kinases were expressed by infecting 0.5–1 L of Sf9 cells (1.5–2· 106 cells per mL) with combinations of individual baculoviruses at a multiplicity of infection (MOI) of 5 for 48 h The cells were harvested by centri-fugation (275 g, 5 min) at 4C and lysed in lysis buffer [10 mM Tris/HCl, pH 7.5, 10 mM NaCl, 2 mM 2-merca-ptoethanol, 0.5 mM EDTA, 10 mM 2-glycerophosphate, 0.5 mMsodium vanadate, 2 mMNaF, 2 lgÆmL)1leupeptin,
2 lgÆmL)1 aprotonin, 2 lgÆmL)1 pepstatin, 0.2% (v/v) Nonidet P-40, 50 lgÆmL)1phenylmethanesulfonyl fluoride]
by 10 strokes with the Dounce homogenizer The proteins were extracted by adding NaCl to a final concentration of 0.4Mand then rocking for 30 min The extract was clarified
by spinning (75 000 g, 30 min) at 4C in an SW50.1 rotor (Beckman) and mixed with 1 mL of Ni2+ nitrilotriacetic acid–agarose beads (Qiagen) that had been equilibrated with 10 mM Tris/HCl, pH 7.6, containing 0.5M NaCl,
5 mMimidazole, 50 lgÆmL)1phenylmethanesulfonyl fluor-ide, and 10% (v/v) glycerol The beads were washed in the equilibration buffer and transferred to a column Proteins were eluted in batch by buffers containing 15–400 mM imidazole, 10 mM Tris/HCl, pH 7.6, 0.1M NaCl, 50 lgÆmL)1 phenylmethanesulfonyl fluoride and 10% (v/v) glycerol The fractions containing the recombin-ant protein kinases were pooled and the buffer was exchanged in PD10 columns (Bio-Rad) to 25 mMsodium Hepes, pH 7.6, 0.1 mM EDTA, 1 mM dithiothreitol, 5% (v/v) glycerol The proteins were then loaded onto a 5 mL
Trang 3Econo-Pac Mono S cartridge (Bio-Rad) and eluted with a
linear 0.08–0.5MNaCl gradient in 25 mMsodium Hepes,
pH 7.6, 0.1 mM EDTA, 1 mM dithiothreitol, 5% (v/v)
glycerol The fractions containing recombinant protein
kinases were identified by SDS/PAGE followed by silver
staining, pooled and stored at)80 C The identity of the
recombinant proteins was confirmed by Western blot with
antibodies against CDK7, CycH, MAT1, CDK8, CycC
and CDK9
Expression and purification of recombinant substrates
All GST-CTD fusion proteins and GST-CDK2 were
expressed in BL21 cells using 0.5 mM isopropyl
thio-b-D-galactoside (IPTG) for 3 h at 30C Cells were lysed by
sonication in TEN buffer (20 mMTris/HCl, pH 7.5, 5 mM
EDTA, 200 mM NaCl, 1 lgÆmL)1 aprotonin, 1 lgÆmL)1
leupeptin, 1 lgÆmL)1 pepstatin, 2 mM benzamidine and
1 mM phenylmethanesulfonyl fluoride) Triton-X-100 was
added to 1% (v/v) and the extract was rocked for 20 min at
4C and then spun at 12 100 g in a JA20 rotor (Beckman)
at 4C The supernatant was loaded onto glutathione–
sepharose 4B beads (Amersham) The bound proteins were
eluted with 15 mM glutathione, 50 mMKCl, 20 mMTris/
HCl, pH 8.0, 15% (v/v) glycerol, and stored at )80 C
Highly purified myelin basic protein (MBP) from bovine
brain was a gift from G Harauz (University of Guelph)
Kinase assay
Kinase reactions were performed in a 20 lL volume
containing 50 mM KCl, 20 mM Tris/HCl, pH 8.0, 7 mM
MgCl2, 5 mM 2-glycerophosphate, 100 lgÆmL)1 BSA,
10 lM ATP, 2 lCi (7.4· 104 Bq) [32P]ATP[cP] (ICN),
40 lgÆmL)1recombinant substrate and 100–400 ngÆmL)1
purified kinase, or the same volume of control fractions
from uninfected Sf9 cells The kinase reactions were
incubated for 30 min at 30C, stopped by the addition of
SDS/PAGE loading buffer, and analyzed on SDS/PAGE
gels and by autoradiography The separation of substrates
from kinases after the kinase reaction was carried out as
follows The kinase reaction was stopped by adding 200 lL
of ice-cold STOP buffer [10 mM sodium EDTA, pH 8,
50 mMKCl, 0.2% (v/v) Nonidet P-40] and incubated with
20 lL of glutathione–sepharose 4B beads The suspension
was rocked for 20 min, the beads were washed three times
in STOP buffer containing 200 mMNaCl, and the bound
proteins were eluted by boiling in SDS/PAGE loading
buffer
Quantification of levels of phosphorylation
Levels of phosphorylation were measured by scanning
exposed films on a Kodak DS 440CF image station using
the KODAK 1D image analysis software Relative signals
along each lane in the gels were evaluated by using the grid
option of the data analysis software Quantification was
conducted only with subsaturated films and only if the
grids did not show saturation (flat) signals Signals in each
segment of the grid were corrected in MicrosoftEXCELby
subtracting the corresponding signals from the identical
segment in the grid from a sample without a substrate
Intensity curves were prepared in MicrosoftEXCEL Total phosphorylation of each substrate was calculated as the sum of signals in all segments corresponding to the substrate bands Relative phosphorylation of individual substrates was calculated by measuring the signals from different substrates on the same X-ray film and normal-izing them to a postulated value of 1 for the intensity of the phosphorylation of the GST-CTD(1–52) substrate Aver-age relative phosphorylation was calculated from these values
The incorporation of ATP in GST-CTD(1–52) and MBP (pmols of ATP min)1Æmg)1 of protein) was determined according to the previously published procedure [36]
Results
Expression, purification and characterization
of recombinant CDK7/CycH/MAT1, CDK8/CycC and CDK9/CycT1
Earlier studies have provided important information on the substrate preferences of CDK7/CycH/MAT1, CDK8/CycC and CDK9/CycT1 However, a comprehensive description
of their properties is far from complete We therefore attempted a more systematic comparison of the activities of these kinases towards different substrates To minimize variations resulting from different sources of material or purification procedures, we prepared the three recombinant kinases following the same expression/purification scheme Briefly, CycH, CDK8 and CDK9 were cloned in baculo-virus vectors as N-terminally 6-Histidine tagged proteins CDK7, MAT1, CycC and CycT1 were expressed as untagged proteins Sf9 cells were infected with combina-tions of CDK7/His6-CycH/MAT1, His6-CDK8/CycC and His6-CDK9/CycT1 baculoviruses The kinases were subse-quently purified by immobilized metal-affinity chromato-graphy (IMAC) using Ni2+nitrilotriacetic acid–agarose and then by ion-exchange chromatography on MonoS beads This procedure purified the three kinases to near-homogeneity, as determined by silver staining (Fig 1A) The identities of the CDK7, CycH, MAT1, CDK8, CycC, and CDK9 bands in Fig 1A were confirmed by Western blot (data not shown) All of these preparations displayed strong kinase activities towards the GST-CTD(1–52) and MBP (Figs 1B, 2 and 3) Typically, different preparations
of CDK7/CycH/MAT1 and CDK9/CycT1 transferred between 0.3 and 1.6 nmols of ATP min)1Æmg)1of protein with both substrates The CDK8/CycC preparations showed somewhat lower specific activities, of 0.09–0.12 nmols of ATP min)1Æmg)1of protein Importantly, when the Ni2+nitrilotriacetic acid–agarose/MonoS fractions that correspond to the fractions with kinase complexes were isolated from uninfected cells, none showed detectable kinase activity towards these substrates (results not shown)
We concluded that most, if not all, of the CTD- and MBP-kinase activity in these preparations belonged to the expressed kinases
Next, we tested whether the three kinases would show substrate specificities that were reported by other groups Kinase reactions were performed with MBP, GST-CDK2(K33>R) and GST-CTD15(S5>A) MBP is a common non-physiological kinase substrate that is rich in
Trang 4serine/threonine (17%) and lysine/arginine residues (19%).
GST-CDK2(K33>R) is a catalytically inactive CDK2
molecule [37] CDK2 is believed to be a physiological
sub-strate of CDK7/CycH/MAT1 [11] GST-CTD15(S5>A)
contains 15 synthetic consensus YSPTAPS repeats [37]
As expected, all three kinases showed significant activity
towards the generic MBP substrate (Fig 1B, lanes 2, 6 and
10), transferring between 0.0001 and 0.001 pmols of ATP
per pmol of MBP per min (data not shown) Only CDK7/
(K33>R) substrate (Fig 1B, lane 3) In agreement with
previous studies [21,25,38], only CDK9/CycT1
phosphor-ylated the GST-CTD15(S5>A) substrate (Fig 1B, lane
12), thus stressing the specificity of CDK7/CycH/MAT1
and CDK8/CycC for S5 of the YSPTSPS consensus and the preference of CDK9/CycT1 for S2 None of the substrates was phosphorylated in the absence of a kinase (Fig 1B, lanes 13–16) We also noticed phosphorylated bands in the CDK8/CycC and CDK9/CycT1 samples that had the mobility of CDK8 (Fig 1B, lanes 5–8) or CDK9, respect-ively (Fig 1B, lanes 9–12) These bands probably represen-ted the autophosphorylation of CDK8 and CDK9 that was reported previously [21,27,38] In summary, we established that our recombinant kinases had properties that were similar or identical to the ones reported in previous studies for their native counterparts We concluded that further comparison of the recombinant kinases was justified
Fig 1 Characteristics of the recombinant CDK7/CycH/MAT1, CDK8/CycC and CDK9/CycT1 kinases (A) Samples from pooled Mono S chromatography fractions containing the three kinases were separated by SDS/PAGE (10% gel) and silver stained The position of each individual recombinant polypeptide is shown on the left The CycH/CDK7 band corresponds to a doublet of CDK7 and His 6 -CycH ( 40 kDa) MAT1 is
36 kDa The CDK8 corresponds to a molecular mass of 53 kDa and CycC 36 kDa; CDK9 is 43 kDa and CycT1, 81 kDa (B) Phos-phorylation of the myelin basic protein (MBP), glutathione S-transferase (GST)-CDK2 and GST-C-terminal domain (CTD)(S5>A) 15 substrates
by CDK7/CycH/MAT1, CDK8/CycC and CDK9/CycT1 Kinase reactions were performed with the combinations of kinase and substrate as indicated above each lane The mobility of the substrate polypeptides are indicated on the left The mobility of 60 and 20 kDa molecular mass markers are indicated on the right.
Trang 5CDK7/CycH/MAT1 hyperphosphorylates CTD Next, we normalized the activity of the three kinases using MBP and compared their activity towards the full-length CTD substrate [GST-CTD(1–52)] (Fig 2A) In these and all subsequent reactions, we used at least a 100-fold molar excess of CTD substrates vs kinase We did not notice major differences in the preference of the three kinases
Fig 2 CDK7/CycH/MAT1, but not CDK8/CycC and CDK9/CycT1
produce a hyperphosphorylated GST-CTD (A) Kinase reactions were
performed with the combinations of kinase and substrate, as indicated
above each lane and as described in the Materials and methods (B)
Kinase reactions were performed with a fixed amount (800 ng) of
GST-CTD(1–52) and serial 1 : 3 dilutions of the kinases, as indicated
above each panel of lanes The mobility of the hypophosphorylated
IIa and the hyperphosphorylated
GST-CTD(1–52)-IIo bands is indicated on the left (C) Kinase reactions were performed
with GST-CTD(1–52) and the kinases as indicated above each lane.
The samples in the input lanes were loaded without further
manipu-lations The samples in the GST pull-down lanes were incubated with
GSH-Separose 4B and the bound proteins were eluted from the
washed beads The mobility of the hypophosphorylated GST-CTD(1–
52)-IIa and the hyperphosphorylated GST-CTD(1–52)-IIo bands are
indicated on the left The mobility of the 90 kDa molecular mass
markers is indicated on the right.
Fig 3 Differential phosphorylation of parts of the C-terminal domain (CTD) by CDK7/CycH/MAT1, CDK8/CycC and CDK9/CycT1 (A) Kinase reactions were performed with the combinations of kinase and substrate, as indicated above each lane The positions of the unphos-phorylated substrate polypeptides (IIa) were derived from Coomassie stained gels without any kinase added (data not shown) and are marked by the asterisk in each lane The bars above each lane represent relative levels of phosphorylation of the substrates The signals of phosphorylation of the GST-CTD(1–52) were equalized between the three different kinases (lanes 2, 9, 16) and the signals of phosphory-lation of the truncated CTD substrates were plotted relative to GST-CTD(1–52) The figure is representative of at least three independent kinase assays with each substrate/kinase combination (B) The average ratios of phosphorylation of individual substrates relative to the GST-CTD(1–52) substrate were calculated and plotted The bars represent
at least three independent parallel experiments with each substrate and the three kinases.
Trang 6towards MBP or GST-CTD(1–52) Therefore, in contrast
to a previous report [28], we do not support the idea that
there was a markedly higher CTD activity in CDK7/CycH/
MAT1 as compared to CDK8/CycC and CDK9/CycT1
(Fig 2A, lanes 2, 5, 8) However, there was a substantial
difference in the mobility of the phosphorylated
GST-CTD(1–52) species that were generated by the three kinases
Whereas CDK8/CycC and CDK9/CycT1 produced mostly
the higher mobility (hypophosphorylated) IIa form, CDK7/
CycH/MAT1 produced equal amounts of both the
higher-mobility IIa and lower-higher-mobility (hyperphosphorylated) IIo
forms (Fig 2A, compare lanes 2, 5 and 8) Most of the
GST-CTD(1–52) retained the mobility of the
unphosphory-lated/hypophosphorylated band, as determined by
Coo-massie staining of the gels after the kinase reactions (data
not shown) In addition, in the reactions with GST-CTD(1–
52), the three kinases transferred between 0.002 and 0.008
pmols of ATP per pmol of GST-CTD(1–52) per min (data
not shown) Thus, assuming only one phosphorylation per
CTD molecule, a maximum of 6–20% of the GST-CTD
molecules could be phosphorylated over the course of the
reaction It is therefore unlikely that the observed generation
of the IIo band was a consequence of limiting substrate
leading to high levels of phosphorylation Nevertheless, to
further test the possibility of limiting substrate, we titrated
the kinases, thus reducing the kinase/substrate ratios As
indicated in Fig 2B, titration of the kinases over a 24-fold
range did not significantly alter the pattern of
phosphory-lation of the GST-CTD(1–52) substrate Similarly,
extend-ing the incubation time of the kinase reactions did not
produce a different pattern of phosphorylation of the
GST-CTD (data not shown) Hence, the differential pattern of
CTD phosphorylation does not appear to be solely a
function of the level of kinase activity
A possible cause for the differential mobility of the
GST-CTD substrate phosphorylated by the three kinases could
be the contamination of the kinase reactions with the
peptidy-prolyl isomerase, Pin1/Ess1 [39–42] Pin1/Ess1 is
a known modifier of the CTD structure that has been
implicated in pol II transcription and RNA processing
[39–42] To test the possibility of Pin1/Ess1 involvement, we
performed reactions with GST-CTD and the three kinases
in the presence of the Pin1/Ess1 inhibitor, juglone [41,42]
Because we found no effect of juglone at concentrations up
to 30 lM(data not shown), we believe it unlikely that the
effects observed occurred as a result of Pin/Ess1
contami-nation
In all preparations of CDK8/CycC and CDK9/CycT1 we
noticed the appearance of phosphorylated bands with
similar mobility to the GST-CTD(1–52)-IIo band (Fig 2B,
lanes 5–12) These bands could potentially obscure the
detection of the GST-CTD(1–52)-IIo form in the kinase
reactions with CDK8/CycC and CDK9/CycT1 In order to
circumvent this potential problem, we pulled out the
GST-CTD(1–52) substrate molecules after completion of the
kinase reactions and analyzed them separately Briefly,
kinase reactions were performed as usual and terminated by
the addition of EDTA Glutathione–sepharose 4B beads
(Amersham) were used to pull out GST-CTD(1–52) and
elute it in SDS sample-loading buffer In Fig 2C (lanes
10–12), we clearly show that under conditions of
non-limiting substrate, only CDK7/CycH/MAT1 could produce
substantial amounts of the hyperphosphorylated GST-CTD(1–52) IIo form
CDK7/CycH/MAT1, CDK8/CycC and CDK9/CycT1 have preferences towards different parts of the CTD
In a set of subsequent experiments, we analyzed the activity
of the three kinases towards different parts of the CTD CTD heptad repeats 1–15, 1–25, 27–39, 27–42, 27–52 and 42–52 (see Fig 6) were expressed as GST fusion proteins Kinase assays were performed exactly as with the full length GST-CTD(1–52) substrate In these analyses, we first determined the relative levels of phosphorylation (IIa + IIo signals) of each of these substrates by the three kinases The CTD repeats 27–39, 27–42 and 27–52 were definitely much better substrates for CDK7/CycH/MAT1 than repeats 1–15 and 1–25 (Fig 3A, compare lanes 3 and 4 with lanes 5, 6 and 7) The best substrate for CDK7/CycH/ MAT1 appeared to be repeats 27–42 (Fig 3A, lane 6) Noticeably, repeats 1–25 and 27–52 were less favored substrates than the shorter substrates represented by repeats 1–15 and 27–42, respectively (Fig 3A, compare lanes 3 and
4 with lanes 6 and 7) CDK8/CycC phosphorylated repeats 1–25, 27–39 and 27–42 well (Fig 3A, lanes 11–13), repeats 1–15 less well (Fig 3A, lane 10) and repeats 27–52 very poorly (Fig 3A, lane 14) CDK9/CycT1 followed a pattern
of total (IIo + IIa) phosphorylation that was similar to that of CDK8/CycC However, the phosphorylation of repeats 27–52 was comparable to that of repeats 1–15 (Fig 3A, lanes 15–21)
A comparison between the phosphorylation of individual CTD substrates by the three kinases is shown in Fig 3B CDK7/CycH/MAT1, CDK8/CycC and CDK9/CycT1 phosphorylated repeats 1–15 and 27–39 at comparable levels CDK7/CycH/MAT1 consistently showed slightly higher activity; however, the difference in total phosphory-lation (IIo + IIa) of these substrates in several independent experiments was not greater than twofold (Fig 3B, graphs
a, c) Repeats 1–25 were phosphorylated well by CDK8/ CycC and CDK9/CycT1 and only moderately by CDK7/ CycH/MAT1 (Fig 3B, graph b) In contrast, repeats 27–42 and 27–52 were much better phosphorylated by CDK7/
(Fig 3B, graphs d, e) In the case of repeats 27–42, these differences were a result of the remarkably higher activity of CDK7/CycH/MAT1, while, in the case of repeats 27–52, the differences were caused by the modest-to-poor activity
of CDK8/CycC and CDK9/CycT1 (Fig 3A) In summary,
we showed that the three kinases did not phosphorylate different parts of the CTD equally Importantly, we outlined regions that enhance or suppress the activity of each kinase
Generation of the IIo form by different parts of the CTD
We noticed substantial variations in the generation of the hyperphosphorylated IIo form of different CTD substrates
by the three kinases We decided to assess these variations
by calculating the percentage of the signal in the IIo substrate bands In order to do so, we measured the intensity of the radioactive signal along each lane of the gel and then subtracted the corresponding signals from the
Trang 7lanes with samples that contained no substrate After
preparing a graph of the intensity of the signal from the
substrates only, we calculated the percentage of signal in the
IIo and IIa bands Thus, assuming that the lower mobility
bands corresponded to the hyperphosphorylated forms of
the substrate, we evaluated the levels of production of
hyperphosphorylated GST-CTD substrates by each kinase
In order to obtain measurable signals for all substrates and a
comprehensive picture of the generation of the IIo form
along the CTD, we performed kinase assays with the
GST-CTD(1–15) and GST-CTD(27–52) substrates with higher
amounts of the CDK8/CycC and CDK9/CycT1 (Fig 4A)
The results from these experiments and the calculations
are presented in Fig 4B As in the case of the full length
CTD (repeats 1–52), CDK7/CycH/MAT1 was very efficient
in generating slowly migrating bands with all but the
CTD(1–25) substrate (Fig 3A, lanes 2–7, and Fig 4B) We
estimated that > 50% of the signal in these reactions was derived from the IIo-band of the substrates (Fig 4B) In sharp contrast, CDK8/CycC did not generate considerable signals in the IIo band with repeats 1–25, 27–39 and 27–42, despite the similar levels of phosphorylation with CDK7/ CycH/MAT1 (Fig 3A, lanes 11–13 and Fig 4B) CDK9/ CycT1 produced a slightly higher percentage of signal from the IIo bands in these three substrates, but still the pattern of phosphorylation was similar to that observed with CDK8/ CycC (Fig 3A, lanes 18–20 and Fig 4B) Surprisingly, CDK8/CycC and CDK9/CycT1 generated an ample per-centage of signal in the IIo form of the CTD(1–15) and CTD(27–52) substrates, while total phosphorylation (IIo + IIa) was lower as compared to the other substrates (Fig 3A, lanes 10, 14, 17, 21 and Fig 4B) In the case of CDK9/CycT1, the GST-CTD(27–52) substrate generated 42% signal in the IIo band, which is comparable to that
Fig 4 Generation of hyperphosphorylated GST-CTD substrates (A) Kinase reactions were performed with the combinations of kinase and substrate, as indicated above each lane The positions of the unphosphorylated substrate polypeptides (IIa) were derived from Coomassie stained gels without any kinase added (data not shown) and are marked by the asterisk in each lane In order to obtain measurable signals, the kinase activity in lanes 4–9 was increased threefold as compared to the experiments in Fig 3 (B) The intensity of the phosphorylation signal was determined along each lane in the gels of Fig 3A The signals from lanes 2–7, lanes 9–13 and lanes 16–21 of the gels in Fig 3A were plotted after subtracting the signals from lanes 1, 8 and 15, respectively The graph representing the phosphorylation of GST-CTD(27–52) by CDK8/CycC was derived from lane 6 of (A), after subtracting the signal from lane 4 The asterisk indicates the position of the unphos-phorylated/hypophosphorylated (IIa) sub-strate The percentage of the signal in the IIo band is shown above each peak.
Trang 8produced by CDK7/CycH/MAT1 (Fig 4B) In conclusion,
we observed differential ability of the three kinases to
hyperphosphorylate different parts of the CTD This ability
did not necessarily correlate to the levels of total
phos-phorylation of these parts
We also need to mention the mobility of the IIo forms of
the different substrates The IIo form of the N-terminal 1–15
repeats was only slightly retarded relative to the position of
the unphosphorylated polypeptide (Fig 3A, lanes 3, 10, 17,
Fig 4A) In comparison, the IIo forms of the C-terminal
repeats 27–39, 27–42 and 27–52 were dramatically retarded,
independently of the kinase that produced them (Figs 3A
and 4A) The magnitude of mobility shift was not dependent
on the number of repeats For example, both CTD(1–15)
and CTD(27–42) contain 15 heptad repeats, but their
mobility shift was substantially different (Fig 3A)
Phosphorylation of GST-CTD(42–52)
CDK8/CycC and CDK9/CycT1 phosphorylated
GST-CTD(27–52) very weakly as compared to GST-CTD(27–
42) (Fig 3A) These observations suggested that repeats
42–52 are a poor substrate for these kinases and that they
contributed to the overall decrease of the phosphorylation
of GST-CTD(27–52) We tested this possibility by
perform-ing kinase assays with a GST-CTD(42–52) substrate In
Fig 5, we show that all three kinases poorly phosphorylated
repeats 42–52 as compared to the full length CTD (Fig 5,
compare lanes 2, 6 and 10 to 4, 8 and 12, respectively)
CDK7/CycH/MAT1 phosphorylated repeats 42–52 better
than CDK8/CycC and CDK9/CycT1 (Fig 5, lanes 4, 8,
12), but the overall signal was low Thus, we obtained a
separate set of data, which indicated that repeats 42–52 are a poor substrate for the three kinases and that they can influence the phosphorylation of the C-terminal portion of the CTD
Discussion
In this study we performed a systematic comparison of the activity of CDK7/CycH/MAT1, CDK8/CycC and CDK9/ CycT1 towards recombinant CTD substrates We expressed and purified all three recombinant kinases from insect cells following the same purification strategy Our CTD sub-strates were portions of the natural mouse CTD We worked under the conditions of non-limiting substrates and evaluated the relative activities of the kinases and the levels
of production of hypophosphorylated (IIa) and hyperphos-phorylated (IIo) forms for each of the substrates This approach unveiled important differences between the three kinases that were not noticed in earlier studies [25,27–29,31] First, we demonstrated that the three recombinant kinases transferred approximately equal amounts of phos-phoryl groups to the full-length CTD substrate, yet they clearly produced different amounts of the hyperphosphory-lated IIo form (Fig 2A) CDK7/CycH/MAT1 generated approximately equal amounts of the hyperphosphorylated IIo and hypophosphorylated IIa forms, whereas CDK8/ CycC and CDK9/CycT1 produced predominantly the hypophosphorylated IIa form (Fig 2) Titration of the kinases (Fig 2b) and time-course experiments (data not shown) indicated that these specific patterns of phosphory-lation were independent of the kinase/substrate ratio Our observations strongly suggest that the three kinases act by different mechanisms on the pol II CTD substrate Because the CTD contains multiple serine residues on the same polypeptide, it can be phosphorylated in two modes
In the disruptive mode, the kinase–CTD complex will uncouple after the transfer of a phosphoryl group to the CTD, then the kinase will form a complex with another CTD molecule and phosphorylate it Under the conditions
of not-limiting substrate, a hypophosphorylated CTD (IIa) will be predominantly produced If the kinase forms a complex with the CTD substrate and phosphorylates multiple S residues, then it acts by a processive mechanism Under the conditions of not-limiting substrate, a hyper-phosphorylated (IIo) form of the CTD will be produced Another way of producing a hyperphosphorylated IIo form would be if (after the initial phosphorylation) the phos-phorylated molecules become higher affinity substrates, leading to a multiple phosphorylation in a disruptive mode Our results suggest that the CTD is phosphorylated in the disruptive mode by CDK8/CycC and CDK9/CycT1 However, CDK7/CycH/MAT1 operates by both processive and disruptive modes Previous studies have shown that CDK7/CycH/MAT1 acts by the disruptive mechanism on short (YSPTSPS)2 substrates [43] and that longer CTD substrates are much better phosphorylated [27] Taken together, these and our observations suggest that the processive mechanism by CDK7/CycH/MAT1 needs more than two YSPTSPS repeats
Second, we demonstrated that different parts of the CTD are differentially phosphorylated by CDK7/CycH/MAT1
We showed that this kinase phosphorylates GST-CTD(27–
Fig 5 Phosphorylation of GST-CTD(42–52) Kinase reactions were
performed with the combinations of kinase and substrate, as indicated
above each lane The position of the unphosphorylated GST-CTD(42–
52) was derived from Coomassie stained gels without any kinase added
(not shown) and is marked by asterisks.
Trang 942) significantly better than GST-CTD(27–39) (Fig 3A).
This minor extension of three heptad repeats creates a
cluster of YSPTSPK repeats that leads to very high levels
of phosphorylation of the rest of the molecule by CDK7/
CycH/MAT1 (Fig 3A) Our observation is agreement
with results of a previous study, which showed that a
(YSPTSPK)4peptide is a better substrate for CDK7/CycH/
MAT1 than (YSPTSPS)4 [27] We therefore propose that
the region encompassing repeats 38–42 is the major site of
CTD phosphorylation by CDK7/CycH/MAT1 At the
same time, we showed that CDK7/CycH/MAT1 weakly
phosphorylates GST-CTD(1–25) relative to the shorter
GST-CTD(1–15) substrate (Fig 3A) This difference
bet-ween the two substrates applied only to CDK7/CycH/
MAT1 For CDK8/CycC and CDK9/CycT1, the better of
the two substrates was GST-CTD(1–25) (Fig 3A) Repeats
16–25 contained two YSPTSPN repeats (Fig 6) It has been
shown that (YSPTSPN)4 peptides are a less favored
substrate of CDK7/CycH/MAT1 than (YSPTSPS)4 [27]
We therefore propose that these repeats act as a suppressor
of CDK7/CycH/MAT1 in the context of longer CTD
substrates
Third, the last 10 C-terminal repeats (42–52) are a very
poor substrate for all three kinases (Fig 5) The presence of
these repeats in the GST-CTD(27–52) substrate has a
significantly negative effect on the activity of CDK8/CycC and CDK9/CycT1 and a moderately negative effect on the activity of CDK7/CycH/MAT1 (Fig 3A) These results are consistent with the idea that repeats 42–52 could act as a kinase suppressor in the context of the full-length CTD The moderate effect on the activity of CDK7/CycH/MAT1 (Fig 3A) could be attributed to the potent positive influence
of the YSPTSPK repeats in 37–42 It is noteworthy that the 42–52 domain contains YSPTSPK repeats that alternate with an equal number of YSPTSPT repeats (see Fig 6) In addition, both GST-CTD(27–52) and GST-CTD(42–52) contain the C-terminal ISPDDSDEEN sequence that is missing from GST-CTD(27–42) The importance of this peculiar alternating of the seventh amino acid in the C-terminal repeats and the ISPDDSDEEN sequence remains to be established
Fourth, we showed that the production of the hyper-phosphorylated CTD substrates is not necessarily a result
of their total phosphorylation For example, total phos-phorylation of GST-CTD(27–39) was approximately equal between the three kinases (Fig 3B, column c) However, CDK7/CycH/MAT1 generated 62% of the signal in the IIo form, while CDK8/CycC and CDK9/CycT1 generated 2% and 5%, respectively (Fig 4B, column d) At the same time, on the longer GST-CTD(27–52) substrate, CDK9/CycT1 produced 42% in the IIo form, yet total phosphorylation was very low (Figs 3A and 4B) The structure of the CTD provides little explanation for the basis of these differences Nonetheless, it is clear that the ability of the kinases to produce hyperphosphorylated substrates is not related to their overall activity towards them
On a minor note, we noticed clear differences in the extent
of retardation of the IIo band in SDS/PAGE between the C-terminal and the N-terminal parts of the CTD (Figs 3A, 4A and 5) The first 15 CTD repeats only slightly change their mobility upon hyperphosphorylation, independently
of the phosphorylating kinase (Figs 3A and 4A), while the other CTD substrates display a dramatic retardation (Figs 3A, 4A and 5) We therefore suggest that the phosphorylation of the C-terminus of the CTD is respon-sible for the generation of the IIo form of pol II in vivo An earlier study had reached the opposite conclusion [28] This discrepancy might stem from the different substrates used Furthermore, we used recombinant kinases, while the other group used immunoprecipitated CDK7 that might contain other kinase activities
Some of our conclusions and observations do not completely agree with separate pieces of evidence reported
by other groups Some of the differences can be explained
by the fact that these studies used short synthetic CTD heptad peptides [25,27–29,31], while we used longer regions
of the natural mouse CTD For example, synthetic peptides were used to address the preference towards S2 or S5 and the effect of the seventh amino acid in YSPTSPS [27–29,31], but these substrates might have a limited use in assessing the preferences in the context of the natural CTD Indeed, CDK9/CycT1 seems to phosphorylate well S5 of the CTD heptad consensus on short synthetic peptides [28,31], but it definitely prefers S2 on full length CTD [25] In addition, in some of these studies, high (1 : 3) or unknown enzyme/ substrate ratios were used, thus posing the risk of masking
Fig 6 A model depicting the possible action of the three kinases on the
C-terminal domain (CTD) The amino acid sequence of the mouse
CTD is shown at the bottom In the diagram, the heptad repeats, with
N at position 7 of the YSPTSPS consensus, are shown as solid
rec-tangles Heptad repeats with K at position 7 are shown as halftone
rectangles Heptad repeats with T at position 7 are shown as striated
rectangles Bent arrows indicate processive phosphorylation Straight
arrows indicate disruptive phosphorylation The three kinases seem to
employ the processive mode of phosphorylation at the N-terminus of
the CTD The YSPTSPN repeats between the 20th and 30th repeats
act as a suppressor of CDK7/CycH/MAT1 and could possibly prevent
the spreading of the phosphorylation by this kinase into the
N-ter-minal portion The C-terN-ter-minal subdomain is phosphorylated by the
processive CDK7/CycH/MAT1 kinase via the cluster of YSPTSPKs
around the 40th repeat At the same time, CDK9/CycT1 is a weak
processive kinase in this region.
Trang 10the differences in the kinase activity because of limiting
substrates
The data in this report are summarized in the model
presented in Fig 6 All three kinases phosphorylate equally
well the N-terminal repeats of the CTD Even though to
a different extent, all three kinases seem to employ the
processive mode of phosphorylation in this region The
YSPTSPN repeats between the 20th and 30th repeats act
as a suppressor of CDK7/CycH/MAT1 and may prevent
the spreading of phosphorylation by this kinase into the
C-terminal portion Thus, the CTD seems to be separated
into two subdomains The C-terminal subdomain is mainly
phosphorylated by the processive CDK7/CycH/MAT1
kinase via a focal point in the cluster of YSPTSPKs around
the 40th repeat At the same time, CDK9/CycT1 is a weak
processive kinase in this region As indicated previously
[44,45], partially phosphorylated CTD is a better substrate
for CDK9/CycT1 than unphosphorylated CTD It is
therefore possible that the CDK9/CycT1 activity could
significantly increase upon initial phosphorylation of the
CTD by another kinase It is also possible that the potent
phosphorylation of repeats 38–42 by CDK7/CycH/MAT1
could spread partially into the last 10 repeats, thus triggering
higher levels of processive activity by CDK9/CycT1 Such
an idea is in agreement with the concept that TFIIH, which
contains CDK7/CycH/MAT1, acts early in the
transcrip-tion process [2,46] P-TEFb, which contains CDK9/CycT1,
acts after TFIIH [2,46]
In vivo, the function of the CTD is influenced by other
modifications, including other phosphorylations,
glycosyla-tion and proline isomerizaglycosyla-tion [1,2,5] All these
modifica-tions and the corresponding enzymes could have additional
effects on the substrate specificities of CDK7/CycH/MAT1,
CDK8/CycC and CDK9/CycT1 These effects are beyond
the scope of the current study
The proposed model indicates a probable pattern of
phosphorylation of the CTD by CDK7/CycH/MAT1,
CDK8/CycC and CDK9/CycT1 The physiological
signifi-cance of certain potential sites of phosphorylation has
already been investigated [1,2,10] However, future studies
are needed to link the described effects to the
phosphory-lation of these sites
Acknowledgements
We would like to thank D Morgan, E Lees and D Price for providing
baculoviruses and vectors for the expression of the recombinant
kinases; N Fong and D Bentley for vectors for the expression of
recombinant CTD substrates; C Hill, J Haines and G Harauz for
MBP; and L Holland and R Dziak for comments and advice This
study was supported by grants to K Y from the Natural Sciences and
Engineering Research Council of Canada (NSERC no 217548) and the
Ontario Genomics Institute (OGI no 043567) K B was supported by
an NSERC studentship.
References
1 Palancade, B & Bensaude, O (2003) Investigating RNA
poly-merase II carboxyl-terminal domain (CTD) phosphorylation Eur.
J Biochem 270, 3859–3870.
2 Kobor, M.S & Greenblatt, J (2002) Regulation of transcription
elongation by phosphorylation Biochim Biophys Acta 1577, 261–
275.
3 Dahmus, M.E (1996) Reversible phosphorylation of the C-ter-minal domain of RNA polymerase II J Biol Chem 271, 19009– 19012.
4 Dahmus, M.E (1994) The role of multisite phosphorylation in the regulation of RNA-polymerase-II activity Prog Nucleic Acid Res Mol Biol 48, 143–179.
5 Oelgeschlager, T (2002) Regulation of RNA polymerase II activity by CTD phosphorylation and cell cycle control J Cell Physiol 190, 160–169.
6 Fong, N & Bentley, D.L (2001) Capping, splicing, and 3¢ pro-cessing are independently stimulated by RNA polymerase II: different functions for different segments of the CTD Genes Dev.
15, 1783–1795.
7 Meininghaus, M., Chapman, R.D., Horndasch, M & Eick, D (2000) Conditional expression of RNA polymerase II in mam-malian cells Deletion of the carboxyl-terminal domain of the large subunit affects early steps in transcription J Biol Chem 275, 24375–24382.
8 Meininghaus, M & Eick, D (1999) Requirement of the carboxy-terminal domain of RNA polymerase II for the transcriptional activation of chromosomal c-fos and hsp70A genes FEBS Lett.
446, 173–176.
9 Fong, N., Bird, G., Vigneron, M & Bentley, D.L (2003) A 10 residue motif at the C-terminus of the RNA pol II CTD is required for transcription, splicing and 3¢ end processing EMBO J 22, 4274–4282.
10 Bentley, D (2002) The mRNA assembly line: transcription and processing machines in the same factory Curr Opin Cell Biol 14, 336–342.
11 Nigg, E.A (1996) Cyclin-dependent kinase-7 – at the cross-roads
of transcription, DNA-repair and cell-cycle control Curr Opin Cell Biol 8, 312–317.
12 Lee, T.I & Young, R.A (2000) Transcription of eukaryotic pro-tein-coding genes Annu Rev Genet 34, 77–137.
13 Liao, S.M., Zhang, J., Jeffery, D.A., Koleske, A.J., Thompson, C.M., Chao, D.M., Viljoen, M., van Vuuren, H.J &Young, R.A (1995) A kinase-cyclin pair in the RNA polymerase II holo-enzyme Nature 374, 193–196.
14 Rachez, C., Lemon, B.D., Suldan, Z., Bromleigh, V., Gamble, M., Naar, A.M., Erdjument-Bromage, H., Tempst, P & Freedman, L.P (1999) Ligand-dependent transcription activation by nuclear receptors requires the DRIP complex Nature 398, 824–828.
15 Malik, S., Gu, W., Wu, W., Qin, J & Roeder, R.G (2000) The USA-derived transcriptional coactivator PC2 is a submodule of TRAP/SMCC and acts synergistically with other PCs Mol Cell 5, 753–760.
16 Wang, G., Cantin, G.T., Stevens, J.L & Berk, A.J (2001) Char-acterization of mediator complexes from HeLa cell nuclear extract Mol Cell Biol 21, 4604–4613.
17 Sun, X., Zhang, Y., Cho, H., Rickert, P., Lees, E., Lane, W & Reinberg, D (1998) NAT, a human complex containing Srb polypeptides that functions as a negative regulator of activated transcription Mol Cell 2, 213–222.
18 Akoulitchev, S., Chuikov, S & Reinberg, D (2000) TFIIH is negatively regulated by cdk8-containing mediator complexes Nature 407, 102–106.
19 Hengartner, C.J., Myer, V.E., Liao, S.M., Wilson, C.J., Koh, S.S.
& Young, R.A (1998) Temporal regulation of RNA polymerase
II by Srb10 and Kin28 cyclin-dependent kinases Mol Cell 2, 43–53.
20 Marshal, N & Price, D (1995) Purification of P-ETFb, a tran-scription factor required for transition into productive elongation.
J Biol Chem 270, 12335–12338.
21 Peng, J., Zhu, Y., Milton, J.T & Price, D.H (1998) Identification
of multiple cyclin subunits of human P-TEFb Genes Dev 12, 755–762.