当前位置:首页 >> 理学 >>

2004The Crystal Structure of Human CDK7 and Its Protein Recognition Properties


Structure, Vol. 12, 2067–2079, November, 2004, ?2004 Elsevier Ltd. All rights reserved.

DOI 10.1016/j.str.2004.08.013

The Crystal Structure of Human CDK7 and Its Protein Recognition Properties
Graziano Lolli, Edward D. Lowe, Nick R. Brown, and Louise N. Johnson* Laboratory of Molecular Biophysics Department of Biochemistry University of Oxford Rex Richards Building Oxford, OX1 3QU United Kingdom ylation has not been identified, but in vitro active phospho-CDK2/cyclin A will promote phosphorylation and CDK7/cyclin H assembly (Fisher et al., 1995). CDK7 also plays a central role in the regulation of the initiation phase of messenger RNA synthesis by RNA pol II. CDK7 is part of the general transcription factor TFIIH (Feaver et al., 1994; Roy et al., 1994; Schiekhattar et al., 1995; Serizawa et al., 1995) and in this complex phosphorylates RNA polymerase II (RNA pol II) large subunit C-terminal domain (CTD) (Goodrich and Tjian, 1994; Lu et al., 1992; Oelgeschlager, 2002; Palancade and Bensaude, 2003; Shilatifard et al., 2003). The CTD contains a tandem array of 52 repeats (in humans) with the consensus sequence YSPTSPS (using the single letter code). CDK7 preferentially phosphorylates Ser5 in this sequence. Phosphorylation of the CTD facilitates promoter clearance, initiation of transcription (Dahmus, 1996), and recognition by RNA processing enzymes. In order to form a stable dimeric active complex with cyclin H, CDK7 must be phosphorylated on a conserved threonine (T170 in the human protein) in the activation loop (Fisher et al., 1995; Martinez et al., 1997). Unlike other CDKs, CDK7 has an additional phosphorylation site in the activation segment (Ser164 in the human protein). Phosphorylation at this site enhances activity and cyclin binding (Martinez et al., 1997). The trimeric CDK7/ cyclin H/MAT1 complex is active in the absence of phosphorylation but phosphorylation stabilizes the complex and stimulates activity toward the CTD of RNA Pol II without affecting activity on CDK2 (Larochelle et al., 2001). Additional factors also regulate CDK7 substrate selectivity in TFIIH (Akoulitchev et al., 2000; Akoulitchev and Reinberg, 1998; Chen et al., 2003; Nishiwaki et al., 2000; Serizawa, 1998). Unlike other cell cycle CDKs, cyclin H levels and CDK7 activity do not vary throughout the cell cycle. For several CDKs, the phosphatases PP2C (Cheng et al., 2000) and the kinase-associated phosphatase (KAP) (Poon and Hunter, 1995) have been identified that recognize and dephosphorylate the phospho site on the activation segment leading to inactive CDK. The crystal structure of KAP in complex with phospho-CDK2 (pCDK2) (Song et al., 2001) showed that the major protein interface between the two proteins is formed by the C-terminal helix of KAP and the C-terminal lobe of CDK2, regions that are remote from both the KAP catalytic site and the activation segment carrying the pThr160 residue of CDK2. At the KAP catalytic site, the CDK2 activation segment interacted almost entirely through the phospho group and the structure required a movement of the activation segment in which it became unfolded and drawn away from the kinase. Sequence comparisons showed that the KAP remote recognition site on CDK2 was not conserved in CDK7 and suggested that CDK7 would not be a substrate for KAP. As a first step toward understanding the recognition properties of CDK7, we report the crystal structure of CDK7 phosphorylated on Thr170 and in complex with ATP. The kinase is in an inactive conformation similar

Summary CDK7, a member of the cyclin-dependent protein kinase family, regulates the activities of other CDKs through phosphorylation on their activation segment and hence contributes to control of the eukaryotic cell cycle. CDK7 also assists in the regulation of transcription as part of the transcription factor TFIIH complex. For maximum activity and stability, CDK7 requires phosphorylation, association with cyclin H, and association with a third protein, MAT1. We have determined the crystal structure of human CDK7 in complex with ? ATP at 3 A resolution. The kinase is in the inactive conformation, similar to that observed for inactive CDK2. The activation segment is phosphorylated at Thr170 and is in a defined conformation that differs from that in phospho-CDK2 and phospho-CDK2/cyclin A. The functional properties of the enzyme against CDK2 and CTD as substrates are characterized through kinase assays. Experiments confirm that CDK7 is not a substrate for kinase-associated phosphatase. Introduction The cyclin-dependent protein kinase 7 (CDK7) plays key roles in the regulation of the eukaryotic cell cycle and in the control of transcription. These dual roles are in turn regulated by the phosphorylation state of the enzyme and the association with regulatory subunits. Like most other CDKs, CDK7 requires association with an activatory cyclin, cyclin H, and phosphorylation on the activation loop for activity. Assembly and activity are augmented by a third protein, the RING finger protein MAT1 (Devault et al., 1995; Fisher et al., 1995; Tassan et al., 1995). CDK7 CDK activation (CAK) activity is essential for cell cycle progression in vivo (Harper and Elledge, 1998; Larochelle et al., 1998; Wallenfang and Seydoux, 2002). CDK7/cyclin H or CDK7/cyclin H/MAT1 will phosphorylate CDK1, CDK2, CDK4, CDK5, and CDK6 (Kaldis et al., 1998; Rosales et al., 2003; Solomon et al., 1992) on their respective threonines in the activation segment leading to activation of these kinases. Levels of CDK7 in cells are low (Fisher and Morgan, 1994) and CDK7 location is predominantly nuclear (Tassan et al., 1994). The in vivo activating kinase for human CDK7 phosphor*Correspondence: louise.johnson@biop.ox.ac.uk

Structure 2068

Table 1. Summary of CDK7 Data Collection and Refinement Statistics ? 3.3 A, ID29 ? Space group and unit cell (A) ? Resolution range (A) (last shell) Rsym Observations Unique observations Mean(I)/ (I) Completeness Multiplicity Refinement statistics Rfactor Rfree ? Rmsd bond lengths (A) Rmsd bond angles ( ) P21: a 66.33, b 193.18, c 77.42, 95.12 23.3–3.3 (3.48–3.3) 0.105 (0.225) 90970 (12322) 28078 (4102) 10.4 (4.8) 96.4 (96.5) 3.2 (3.0) ? 3.0 A, ID13 P21: a 65.50, b 191.63, c 75.79, 94.40 30.0–3.0 (3.16–3.0) 0.165 (0.493) 54240 (7970) 29716 (4471) 6.1 (1.5) 80.0 (82.3) 1.8 (1.8) Combined Data

30.0–3.0 (3.16–3.0) 0.143 (0.569) 147115 (8046) 36804 (4531) 8.5 (1.2) 94.9 (80.3) 4.0 (1.8) 0.219 0.296 0.017 1.970

to that of inactive CDK2 but the activation segments of the two CDKs differ. We describe the similarities and differences between CDK7 and CDK2 and other protein kinases with respect to their ATP binding sites and their protein recognition sites, including that for KAP. We correlate the structural observations with kinase activity measurements. Results The Structure of CDK7 The structure of CDK7 was solved by molecular replacement using the structure of inactive CDK2 as a search object (Table 1). The CDK7 structure exhibits a typical kinase fold comprising the N-terminal lobe (residues 13–96), with mostly sheet and one helix ( C), and a C-terminal lobe (residues 97–311) comprised mostly of helices (Figure 1A). The fold can be traced from residue ? 13 to 311 in the 3.0 A resolution electron density, whose quality is enhanced by the 4-fold noncrystallographic symmetry averaging. There is one break between residues 44–55, corresponding to the link between the 3 strand and the C helix. The N-terminal residues 1–12 and the C-terminal residues 312–346, that contain a putative nuclear localization sequence, are not located in the electron density. ATP is bound between the two lobes and was clearly identifiable in all four copies of the CDK7 structure. The conformation of CDK7 is similar to the inactive conformation of CDK2 in the absence of cyclin A (De Bondt et al., 1993). The C-helix, which contains the sequence NRTALRE (residues 56–62) in CDK7 corresponding to the PSTAIRE sequence motif in CDK2, is rotated about its axis and shifted toward the exterior of the protein compared with its conformation in the active CDK2/cyclin A complex (Jeffrey et al., 1995; Russo et al., 1996a). All side chains in the region 56–62 (NRTALRE) are exposed. As a result, the interaction between the glutamate from the C helix (Glu62) and a lysine (Lys41), which stabilizes the lysine residue for interaction with the triphosphate of ATP in active protein kinases, is not made. The activation segment (Johnson et al., 1996) is well localized in the electron density but has higher B factors than the rest of the molecule (aver? age B factors for main chain atoms for chain A: 42 A2; average main chain B factors for activation segment ? residues 155-182 in chain A: 58 A2).

The results from CDK7 mass spectrometry show two peaks with molecular mass of 39,217 and 39,141 corresponding to approximately 30% bisphosphorylated CDK7 (39,061 160) and 70% monophosphorylated CDK7 (39,061 80). The results were identical for two separate preparations. In the crystal structure, we find only the monophosphorylated form. Thr170 is phosphorylated and Ser164 is not phosphorylated (Figure 1B). The pThr170 makes one hydrogen bond through its phosphate group to the main chain nitrogen of Gln22 from the glycine loop between 1/ 2 and is otherwise exposed to the solvent (Figure 1C). The phospho-threonine approaches the phosphate of ATP (the distance between the phosphorous of ATP and the phospho? threonine phosphorus atoms is 6.7 A), but the side chain of Gln22 partly intervenes between the two phosphates. The structure suggests that autophosphorylation on Thr170 is unlikely in agreement with results in solution (Martinez et al., 1997; Poon et al., 1994). Ser164 is exposed on the surface of the protein. There is no additional density surrounding this side chain to indicate phosphorylation (Figure 1B). Ser164 is contained in a consensus sequence for CDK2 phosphorylation (SPNR in single letter code) but has not been 100% phosphorylated in insect cells. Comparison of CDK7 and CDK2 Structures The structures of CDK7 and CDK2 are similar with an ? rms deviation in C positions for 251 residues of 1.25 A. CDK7 has 44% sequence identity to CDK2 (Figure 2). Major changes between CDK7 and CDK2 occur in three regions: (i) the activation segment (CDK7 residues 155– 174); (ii) the region where three parts of chain pack together comprising D (CDK7 residues 101–112), the 7/ 8 loop (CDK7 residues 146–149), and the C-terminal region (CDK7 residues 297–311); and (iii) the region of the L14 loop (residues 247-251) which is part of the KAP recognition site and Cks1 regulatory protein binding site in CDK2 (Figure 3A). Changes in these regions have possible functional significance. In addition there are also smaller changes at the N-terminal region and in the loop regions between 1/ 2, 2/ 3 and 4/ 5 (Figure 3A). The activation segment of CDK7 adopts a different conformation to that observed for CDK2 and phosphoCDK2 (pCDK2) (Figure 3B). In pCDK2 (CDK2 phosphory-

Crystal Structure of Human CDK7 2069

Figure 1. The Structure of CDK7/ATP Complex (A) A schematic diagram of CDK7 with secondary structural elements labeled. ATP is bound between the N- and C-terminal lobes. There is a break in electron density between the end of 3 and the start of C. The activation segment is in a darker color. (B) The activation segment is phosphorylated on Thr170 and not on Ser164. The final SigmaA weighted 2Fo Fc electron density map contoured at 0.94 for residues 164–172. (C) The activation segment of CDK7 in residues 163–171 showing the contacts of the pThr170 with the main chain nitrogen of Gln22 from the glycine loop between 1/ 2. This diagram and other figures were produced with AESOP (M.E.M. Noble, personal communication).

lated on T160 but without the association of cyclin A) (Brown et al., 1999b) the structure is similar to inactive nonphospho CDK2 with the only difference that the activation segment is disordered. In CDK7, at the start of the activation segment, the side chain of aspartate Asp155 of the DFG motif is turned away from the catalytic site forming a hydrogen bond with Lys41, in a similar conformation to that observed for inactive CDK2. In this position the aspartate is unable to participate in contacts to the magnesium ion that interacts with the and phosphates of ATP in the activated form of CDKs. CDK7 and inactive CDK2 activation segments differ from this point (Asp155 in CDK7, Asp145 in CDK2) until they come back into register at residue Arg176 in CDK7 (Leu166 in CDK2) corresponding to the middle of the P 1 loop. As in CDK2, there is a short two-turn helix (residues 158–165), which docks against the C helix keeping the latter in the inactive conformation through interactions such as the packing of Phe156 against Val65. The activation segment takes a more open route in CDK7 than in CDK2 (Figure 3B), resulting in the dis? tance of 3.8 A between C atoms of pThr170 (CDK7) and Thr160 (CDK2). On association with cyclin A and

phosphorylation, the activation segment of CDK2 undergoes a dramatic change in conformation (Figure 3B). The separation of C pThr170 (CDK7) to C pThr160 in ? pCDK2/cyclin A is 23.7 A. The second major change is at the C terminus where residues 297–309 in CDK7 differ from the corresponding residues in CDK2. The CDK7 proline-rich C-terminal region makes more intimate contacts with the rest of the protein than with the corresponding region in CDK2. The interactions involve the end of D and the D/ E loop (residues 101–112, a region where there is little similarity in sequence between CDK7 and CDK2), and the region of the 7/ 8 loop residues146–149 (Figure 3A). The C-terminal region threads between these two loops and appears to displace the 7/ 8 loop. The end of the hinge region at 6 is also displaced and the D helix in CDK7 lacks the final turn as seen in CDK2. The shifts allow favorable contacts: for example, Pro308 interacts with Val108 ( D/ E loop), Arg309 ion pairs with Glu147 ( 7/ 8 loop), and Pro310 near the C terminus interacts with Val100 and Ile101 ( D helix). Protein kinases contain several protein interacting sites on their surfaces. The MAP kinase family members

Structure 2070

Figure 2. Sequence Alignment of CDK7 and CDK2 Showing the Assignment of Secondary Structural Elements in CDK7 Beta strands are shown as arrows and helices as shaded boxes. Short stretches of helix or 310 helix are shown as open boxes.

interact specifically with docking site sequences in substrates (Sharrocks et al., 2000). These interactions may confer pathway specificity and substrate selection and provide an additional recognition site that is separate from the catalytic substrate recognition site. The structure of p38 in complex with docking peptides from two such interacting proteins, one from a transcription factor MEF2A and the other from an activating kinase MKK3b (Chang et al., 2002), show these peptides interacting at a site that is equivalent to that occupied by the C-terminal region of CDK7 (i.e., the pocket between 7/ 8 loop and the D/ E loop [Figure 3C]). In particular in CDK7 proline residues 308 and 310 occupy the equivalent positions to the nonpolar residues in the sequence φa-X-φb that is the hallmark of these p38-docking peptides. Although the C-terminal region of CDK2 also occupies this pocket, the nonpolar interactions are not manifest to such a dramatic extent as in CDK7. The third region of major difference is in the CDK insert region L14 where there is a deletion of one residue in CDK7 with respect to CDK2 (Figures 2 and 3A). This region forms part of the recognition site of pCDK2 for binding KAP (Song et al., 2001) (Figure 4A). In the pCDK2/KAP complex, Lys237 of CDK2 makes an ion pair with Glu191 of KAP (Figure 4B). In CDK7, the corresponding residue to Lys237 is Val247 (Figure 4C). The small changes in sequence and conformation of this region suggest that KAP would not be able to recognize pCDK7 and would not be able to dephosphorylate CDK7. This hypothesis is further explored below. The ATP Binding Site ATP bound to CDK7 has a slightly different conformation and contacts than those of ATP bound to CDK2 (Figure

5A). The differences arise from the overall differences in protein conformation, partly promoted by the difference in conformation of the activation segments, and the lack of magnesium ions in the CDK7 structure. Overall, the number of van der Waals contacts made between ATP and CDK7 and between ATP and the inactive form of CDK2 are similar but the residues involved are different. ATP in CDK7 makes more extensive contacts to the glycine loop than in CDK2 but fewer contacts to the catalytic residues and the metal binding residues. The adenine of ATP makes two hydrogen bonds to the hinge region between the two lobes, as observed in other kinase structures, and also contacts the hinge region residue Met94 (Figure 5B). The ribose in CDK7 is closer toward the glycine loop than in CDK2, while the triphosphate moieties are very different especially for the phosphates. In CDK7 two of the phosphate oxygens hydrogen bond to main chain nitrogens of Phe23 and Ala24 at the 1/ 2 turn making a well-localized arrangement with the glycine loop. A third phosphate oxygen hydrogen bonds to Lys41. The difference in conformation of the activation segment between CDK7 and CDK2 allows the side chain of Ser161 to contribute a hydrogen bond to the phosphate in CDK7. This residue occupies the position of Tyr15 in CDK2 reflecting differences in the activation segments and glycine loops. In CDK7 Phe162, near the start of the activation segment, stacks against Phe23, the residue corresponding to Tyr15 in CDK2. The absence of metal ion in the CDK7 structure appears to have encouraged the phosphate to seek hydrogen bonds with the glycine-loop main chain nitrogens rather than with the catalytic residues.

Crystal Structure of Human CDK7 2071

Figure 3. Comparison of the Structures of CDK7 and CDK2 (A) A schematic diagram of CDK7 (in brown) and CDK2 (in cyan). There are significant changes in three major regions: the activation segment; the C-terminal region and contacts to the 7/ 8 loop and the end of the D helix; and the L14 loop, the site of interaction of KAP in CDK2. (B) Comparison of the activation segments from CDK7 (brown), CDK2 (cyan), and activated pCDK2/cyclin A (green). The region from 155 (CDK7) DFG to 182 (CDK7) APE is shown. (C) The C-terminal region of CDK7 (brown) with the p38 structure (green) (PDB ID 1LEW) superimposed. The binding of the MEF2A docking peptide to p38 is shown in magenta. This overlaps the C-terminal region of CDK7.

Kinase Assays and Substrate Specificity In order to determine the activation state of the crystalline CDK7 and to probe the influence of cyclin H and MAT1 on activity and specificity, we compared the relative activities of CDK7, CDK7/cyclin H, and CDK7/cyclin H/MAT228 against CDK2 and CTD as substrates. MAT228 (residues 228–309) is the minimal region of MAT1 that is able to stimulate CDK7/cyclin H activity (Busso et al., 2000). CDK7 has no activity against CDK2 but does have a weak basal activity against the CTD (Figure 6A, lane 1 and Figure 6B). CDK7/cyclin H complex is active with greater activity against CDK2 than against CTD by a

factor 2.5 (Figure 6A, lane 2, and Figure 6B) in agreement with (Watanabe et al., 2000). The trimeric CDK7/ cyclin H/MAT228 complex is more active against the CTD substrate than against CDK2 (Figure 6A, lane 3, and Figure 6B). Under our conditions, CTD phosphorylation is increased by 4 times over that observed for CDK7/cyclin H while the CDK2 CAK activity is not significantly affected by the presence of MAT228. In the presence of ATP, both CDK7/cyclin H and CDK7/cyclin H/MAT228 will autophosphorylate cyclin H (unpublished data). Phosphorylation of cyclin H results in a reduction in the activity of the dimeric CDK7/cyclin H complex against both CDK2 and CTD (Figure 6C)

Structure 2072

Figure 4. Comparison of the KAP Binding Site between pCDK2 and pCDK7 (A) Overall schematic diagram with CDK7 (brown) superimposed on the pCDK2/KAP complex (pCDK2 is in cyan and KAP in blue). (B) Details of the pCDK2/KAP binding site. pCDK2 is in cyan and KAP in magenta. There is an ionic interaction between Lys237 (pCDK2) and Glu191 (KAP). (C) Details of the changes at this site in CDK7 (in gold) with the key KAP residues superimposed. Val247 (which replaces Lys237 of CDK2) cannot interact favorably with Glu191. For further details see text.

but no change in activity of the trimeric CDK7/cyclin H/MAT228 complex against either CDK2 or CTD (Figure 6C). It is possible that cyclin H phosphorylation destabilizes the dimeric complex (thus reducing its activity), but the presence of MAT1 provides additional stability in the trimeric complex and thus is able to counteract the effects of phosphorylation of cyclin H. Thus, MAT1 may have an additional role in regulating CDK7 activity by allowing the activity to escape regulation by cyclin H phosphorylation.

CDK7/cyclin H or CDK7/cyclin H/MAT228 complexes produced by coinfection in insect cells, were found by mass spectrometry to have CDK7 60% bisphosphorylated and 40% monophosphorylated (data not shown). To analyze the influence of CDK7 phosphorylation state on activity, we compared the CDK2 CAK activities before and after incubation with pThr160-CDK2/cyclin A and ATP (Figure 6D). The CDK2 reaction appears to go to completion because subsequent incubations of these preparations with pCDK2/cyclin A and ATP led to no

Crystal Structure of Human CDK7 2073

CDK7/cyclin H and CDK7/cyclin H/MAT228 complexes (Figure 6D). There is an increase in activity of 2.8 and 3.7 times for the additionally phosphorylated CDK7/ cyclin H and CDK7/cyclin H/MAT228 complexes, respectively. The lower activity of pCDK7/cyclin H with respect to pCDK7/cyclin H/MAT228 can most likely be ascribed to the autophosphorylation on cyclin H (described above) that causes a 35% reduction in activity of CDK7/cyclin H but no reduction for CDK7/cyclin H/MAT228 (Figure 6C). Activity against the CTD could not be tested under these conditions because pThr160CDK2/cyclin A also phosphorylates CTD (Palancade and Bensaude, 2003).

Figure 5. Comparison of pCDK7/ATP and CDK2/ATP Interactions (A) Comparison of ATP bound to CDK7 (carbon atoms in green) and ATP bound to CDK2 (carbon atoms in yellow). (B) Details of the interactions between CDK7 and ATP. (C) Details of the interactions between CDK2 and ATP. For further details see text.

further incorporation of phosphate and the phosphoCDK7 ran as a single band under gel electrophoresis in contrast to the two bands seen when both bis- and mono-phospho forms were present. After preincubation with pThr160-CDK2/cyclin A CDK7 showed weak activity, quantifiable as approximately 2% of the CDK7/cyclin H activity. Martinez et al. (1997) found that Xenopus CDK7 and CDK7/cyclin H can be activated by CDK2/ cyclin A or CDK1/cyclin B in reticulocyte-translated lysates with pCDK7 achieving approximately one-third of the activity of the CDK7/cyclin H complex. From our results, we conclude that in human phospho-CDK7, in contrast to Xenopus CDK7, does not exhibit appreciable CDK2 CAK activity. The effects of phosphorylation are more evident in the

KAP Dephosphorylates CDK2 but Not CDK7 The structure of the complex between KAP and pCDK2 (Song et al., 2001) showed that recognition at the catalytic site was achieved solely through the phosphate group but that the major determinant of specificity was ? located about 20 A from the catalytic site where the C-terminal helix of KAP interacted with residues from the G helix and the L14 loop of CDK2. The L14 region differs in CDK7 (Figure 4) resulting from sequence changes that involve the sequence DYV(247)-T in CDK7 (where – denotes a gap) and DYK(237)PS in CDK2 (Figure 2). In order to test if these features destroy the docking site and abolish KAP activity, we tested KAP activity against CDK7 and constructed KAP-interacting mutants (KM) for CDK2 and CDK7. KM-CDK2 has the sequence Val237- in place of Lys237Pro238; KM-CDK7 has the sequence Lys247Pro in place of Val247-. These mutations create two chimeras: CDK2 with CDK7-L14 loop and CDK7 with CDK2-L14 loop. The profile of KAP activity against pThr160-CDK2, KM-pCDK2, CDK7, and KM-CDK7 is showed in Figure 7A. As expected, KAP is active against pCDK2 and inactive against CDK7. Dephosphorylation of pCDK2 results in a band shift in which dephosphorylated CDK2 migrates slower than the phosphorylated isoform (Figure 7A, lanes 2 and 1, respectively). No bandshifts can be detected after KAP treatment of CDK7 either for the monophosphorylated (upper band) or bisphosphorylated (lower band) forms of CDK7 (Figure 7A, lanes 5 and 6). The L14 loop mutations in KM-pCDK2 abolish KAP activity (no bandshift detectable: Figure 7A, lanes 3 and 4). This result highlights the importance of the docking site for the regulation of KAP activity on pCDK2. However, KAP is not active against KM-CDK7 (Figure 7A, lanes 7 and 8). Evidently the generation of a correct KAP docking site on CDK7 was not sufficient to make it a substrate. This may be because other effects are also important. For example, the rigidity of the pCDK7 activation segment compared with the disordered activation segment in pCDK2 may make it less amenable for conformational change to reach the KAP catalytic site. We note that the relative intensity of the bands in Figure 7, show that KM-CDK7 is approximately 70% bisphosphorylated and 30% monophosphorylated (Figure 7A, lane 7), while CDK7 is 30% bisphosphorylated and 70% monophosphorylated (Figure 7A, lane 5). In order to verify that these mutations did not affect the overall folding and the other binding sites of CDK2

Structure 2074

Figure 6. Kinase Activities of CDK7, CDK7/ Cyclin H, and CDK7/Cyclin H/MAT228 (A) Autoradiograph of CDK2 and CTD phosphorylated by CDK7 (lane 1), CDK7/cyclin H (lane 2), and CDK7/cyclin H/MAT228 (lane 3). The reactions were performed as described in Experimental Procedures. (B) Relative CDK2 and CTD phosphorylation activities for CDK7 CDK7/cyclin H and CDK7/ cyclin H/MAT228 from results shown in (A). Cyclin H in the dimeric complex generates CAK activity on CDK2 making CDK2 the preferred substrate for CDK7. MAT228 in the trimeric complex increases CTD phosphorylation without affecting CAK activity. (C) Autophosphorylation on cyclin H decreases CDK7/cyclin H activities on CDK2 and CTD without affecting CDK7/cyclin H/ MAT228 activities. (D) Effect of CDK7 phosphorylation by pThr160-CDK2/cyclin A on CAK activity. Minimal basal activity is induced in CDK7 (change too small to be observed in this figure). CDK7/ cyclin H and CDK7/cyclin H/MAT228 complexes showed an increase in activity of 2.8 and 3.7 times after preincubation with pThr160-CDK2/cyclin A.

and CDK7, we compared activity of wild-type and mutant proteins (Figure 7B). Phosphorylation of CDK7 by pThr160CDK2 or KM-pThr160CDK2 are similar (Figure 7B, lanes 1 and 2) and the activatory effects of cyclin A on pCDK2 or KM-pCDK2 for phosphorylation of CDK7 are also similar (Figure 7B, lanes 3 and 4). These results show that KM-pCDK2 has been phosphorylated by Civ-1 during expression to give the active enzyme and that mutation of pCDK2 L14 loop does not affect interactions with either cyclin A or with the substrate CDK7. Both CDK7 and KM-CDK7 are able to phosphorylate CTD (Figure 7B, lanes 5 and 6). The higher activity of KMCDK7 is ascribed to its different phosphorylation state in respect to CDK7 (described above). We conclude that mutation of CDK7 L14 loop does not affect activity against the CTD. Discussion The crystal structure of CDK7 shows that the overall fold of the kinase is similar to that of inactive CDK2, as expected from their sequence similarity. Major differences occur in the activation segment, the C-terminal region and the L14 loop, which is part of the CDK2 recognition site for KAP. Mass spectrometry indicated a 30:70 ratio of bisphosphorylated and monophosphorylated forms of CDK7 produced by expression from baculoviral vectors in insect cells but the crystal structure contains only the monophosphorylated form in which Thr170 is phosphorylated. Evidently crystallization has selected the monophosphorylated form from the mixture. In contrast to pThr160CDK2 where the phosphorylated activation segment is disordered, CDK7 shows an ordered activation segment. The conformation of the activation segment is different from that of CDK2 and from that of pCDK2/cyclin A but is closer to that of the inactive CDK2. The CDK7 preparation prepared for

crystallization showed no activity against CDK2 but did exhibit basal activity against the CTD from RNA polymerase. The lack of conventional activity against CDK2 is consistent with the inactive conformation of the kinase observed in the crystal structure. Further phosphorylation of CDK7 by pCDK2/cyclin A in order to achieve full phosphorylation on both Thr170 and Ser164, led to no significant CAK activity against CDK2. Phosphorylation of the activation segment in CDK7 is insufficient to trigger the conformational changes that convert an inactive kinase to an active kinase. Association of CDK7 with cyclin H and with cyclin H/MAT228 leads to significant activity against both CDK2 and the CTD. In these complexes there is a higher amount of bisphosphorylated CDK7 than monophosphorylated CDK7 (ratio 60:40) produced by the insect cells. We find that the CDK7/cyclin H complex has a preference for CDK2 over CTD of about 2.5-fold at substrate concentrations of 7.6 M, as others have also observed (Watanabe et al., 2000) but that with the CDK7/cyclin H/MAT228 complex the substrate preference is shifted by a ratio of about 1.4 in favor of the CTD. Others have observed a more dramatic shift in substrate specificity on association with MAT1 (Larochelle et al., 1998; Yankulov and Bentley, 1997). The MAT1 construct of residues 228-309, which was used in our experiments, has been identified as the minimal construct need to provide activation of CDK7/cyclin H (Busso et al., 2000). However, the N-terminal RING finger domain is important for transcription activation and promotion of phosphorylation of the CTD while the middle region comprising a coiled-coil allows binding of CDK7/cyclin H/MAT1 through interactions with both XPD and XPB helicases (Busso et al., 2000). In contrast to CDK7 alone, further phosphorylation of CDK7/cyclin H and CDK7/cyclin H/MAT228 by pCDK2/cyclin A led to an enhancement of activity against CDK2 substrate of 3.7-fold, suggesting that phosphorylation on both Thr170 and Ser164

Crystal Structure of Human CDK7 2075

Figure 7. KAP Activity against CDK2, KM-CDK2, CDK7, and KMCDK7 and Kinase Assays (A) KAP activity. Dephosphorylation of CDK2 is detectable by bandshift (lanes 1 and 2) but not for KM-CDK2 (lanes 3 and 4); no changes are detectable in the relative intensity of monophosphorylated and diphosphorylated CDK7 (lanes 5 and 6) or for KM-CDK7 (lanes 7 and 8). The relative shift of bands between CDK7 and KM-CDK7 arise from the difference in phosphorylation state of CDK7 (discussed in text). (B) Comparison of kinase activities of CDK2 and CDK7 and their respective L14-loop mutants. CDK7 was used as substrate for CDK2; CTD was used as substrate for CDK7. (See text for further details.)

are required for maximal activity. The change in substrate specificity conferred by MAT1, which shifts the preference to the CTD over CDK2, raises intriguing questions that can only be answered by structural studies on the dimeric and trimeric complexes and their association with the substrates. Protein/protein interactions at sites remote from the catalytic site play key roles in many processes of protein kinase substrate recognition (e.g., MAPK family; CDK2; GSK; PLK1). The positions of these sites have been elucidated in a number of structural studies with peptides from the cognate substrate proteins (Bax et al., 2001; Chang et al., 2002; Cheng et al., 2003; Dajani et al., 2003; Elia et al., 2003; Lowe et al., 2002; Russo et al., 1996b) and demonstrate that different parts of the kinase surface or regulatory domains may be involved in such recognition. These studies have not yet shown how the docking sites and the catalytic sites communicate, if at all. The crystal structure of KAP in complex with pCDK2 represents one of the few examples where an enzyme in complex with its complete phospho-protein substrate is known. The structural results (Song et

al., 2001) explained why KAP will only dephosphorylate pCDK2 when it is in its folded state and not when it is denatured. There are changes in sequence in the KAP recognition site of CDK2 in CDK7, which suggested that this recognition site would be disrupted in CDK7. This has been confirmed by the present work, which has shown that CDK2 is a substrate for KAP but not CDK7. We used site directed mutagenesis to change the L14 loop of CDK2 to resemble that of CDK7 (resulting in KMCDK2). The changes in sequence of just two amino acids were sufficient to disrupt the recognition site and KAP showed no activity against the KM-CDK2. In contrast engineering the CDK2 sequence into the CDK7 L14 loop (KM-CDK7) did not give a protein substrate for KAP. We conclude that other factors, such as the greater rigidity of the phospho-activation segment, could also contribute to the resistance of CDK7 to dephosphorylation by KAP. Protein kinases are major targets for drugs. There are several compounds that are either on the market or in advanced clinical trials for treatment of cancer, diabetes and inflammatory diseases (Noble et al., 2004). One of the major problems is to develop a compound that is selective for a particular kinase or signaling pathway. Almost all drugs developed so far target the ATP binding site. We compared the ATP site in two homologous kinases CDK2 and CDK7 to see if there are features that might confer selectivity. The major differences between CDK7 and CDK2 ATP sites occur at the “CDK2 Lys89 pocket” or “hydrophobic pocket” (Fabbro et al., 2002). This pocket has been targeted successfully by CDK2 inhibitors (Davies et al., 2002; Davis et al., 2001; Gray et al., 1998) to provide potent and selective inhibitors. In CDK7 the residue corresponding to Lys89 at the bottom of the pocket is Val100 (Figures 5B and 5C). This change together with other changes in the region such as Thr96 (in place of Gln85 in CDK2), and the presence of Pro310 creates a pocket of greater nonpolar character in CDK7 than in CDK2. In CDK4, Lys89 of CDK2 is replaced by a threonine and this change has been exploited in the development of specific CDK4 inhibitors (Beattie et al., 2003; Breault et al., 2003; Honma et al., 2001a, 2001b; Soni et al., 2001) and a CDK4-like mutant of CDK2 (Ikuta et al., 2001). In CDK7 this hydrophobic pocket offers the possibility for designing selective inhibitors. The structure of CDK7 joins the structures of CDK2 (Brown et al., 1999a; De Bondt et al., 1993; Jeffrey et al., 2001; Russo et al., 1996a), CDK5 (Tarricone et al., 2001), and CDK6 (Brotherton et al., 1998; Russo et al., 1998; Schulze-Gahmen and Kim, 2002) as representative structures of the CDK family. Each of these exhibits different regulatory and substrate specificity properties. For CDK7 further work is required to understand the activation by regulatory domains and to understand why CDK2 can phosphorylate CDK7 and CDK7 can phosphorylate CDK2 but neither can autophosphorylate. The role of additional protein recognition sites in these processes should provide a fascinating explanation of substrate specificity as will understanding other processes such as the interactions of cyclin H with the U1 snRNA that provide a link between transcription and RNA processing (Kwek et al., 2002) and the interactions of CDK7

Structure 2076

with the Hepatitis C virus core protein that could link viral infection to cell cycle regulation (Ohkawa et al., 2004).
Experimental Procedures All reagents, unless specified otherwise, were purchased from Sigma-Aldrich (St. Louis, MO).

Expression, Purification, and Mass Spectrometry pBS-CDK7 vector, kindly provided by Prof. David Morgan (Dept. of Physiology, University of California, San Francisco, CA) and Prof. Robert P. Fisher (Memorial Sloan Kettering Cancer Center, New York), was used as a template for the subsequent PCR. CDK7 was cloned into the vector pVL1392 (Pharmingen BD, Franklin Lakes, NJ) modified with the insertion of GST and PreScission protease cleavage site. CDK7-expressing baculovirus was obtained by transfecting 2 106 Sf21 cells with 2 g of plasmid DNA and 0.5 g of linearized baculovirus DNA for 5 days using BaculoGold Transfection kit (Pharmingen BD). After two rounds of amplification, expression was obtained by infecting Sf21 cells with GST-CDK7 baculovirus (MOI 0.3) at 27 C for 3 days. Cells were collected by centrifugation at 1000rpm and resuspended in 50 mM Tris-HCl (pH 8), 250 mM NaCl, 10% glycerol, and 10 mM DTT and Complete protease inhibitor cocktail (Roche, Basel, Switzerland). Cells were disrupted by sonication and the lysate, after clarification by centrifugation at 10,000 rpm, was loaded on GSH-Sepharose (Amersham Biosciences, San Francisco, CA). Resin was washed with 10 column volumes of the lysis buffer and then with 10 column volumes of 50 mM Tris-HCl (pH 8), 250 mM NaCl, 10% glycerol, 1 mM DTT, and 1 mM EDTA. Cleavage was performed overnight with 1 column volume of the above buffer containing 40 l of PreScission (Amersham Biosciences) protease/ml resin. Cleaved protein was recovered in 2 column volumes of the same buffer and loaded onto Superdex 200 26/60 (Amersham Biosciences). Size-exclusion chromatography was run using 50 mM Tris-HCl (pH 8), 200 mM NaCl, 10% glycerol, and 1 mM DTT. Cyclin H baculovirus was a gift from Prof. David Morgan (Dept. of Physiology, University of California, San Francisco, CA). pGEXMAT1, provided by Prof. Erich Nigg (Max Planck Institut, Martinsried, Germany), was used as a template for subsequent PCR. MAT228 (corresponding to the C-terminal region of MAT1 residues 228–309) was cloned into the vector pVL1392 (Pharmingen BD) modified with the insertion of MBP and PreScission protease cleavage site. CDK7/ cyclin H and CDK7/cyclin H/MAT228 complexes were produced by coinfection of respective viruses in Sf21 cells. Purifications were carried out as described above. pGEX-CTD was provided by Dr. Andre Furger (Dept. of Biochemistry, University of Oxford, UK) and used to transform DH5 cells. Cells were induced overnight with 0.25 mM IPTG at 20 C and sonicated in 50 mM Tris-HCl (pH 8), 200 mM NaCl, 10% glycerol, and 10 mM DTT and Complete protease inhibitor cocktail (Roche). Clarified lysate was loaded on GSH-Sepharose (Amerham Biosciences) and resin was then washed with 10 column volumes of lysis buffer and 10 column volumes of in 50 mM Tris-HCl (pH 8), 200 mM NaCl, 10% glycerol, and 1 mM DTT. GST-CTD was then eluted with 2 column volumes of 50 mM Tris-HCl (pH 8), 200 mM NaCl, 10% glycerol, and 10 mM glutathione. CDK2, pThr160-CDK2, and pThr160-CDK2/cyclin A were produced as previously described (Brown et al., 1999a) pET28a-KAP expression vector was kindly provided by Prof. David Barford (Institute of Cancer Research, London, UK) and subsequently used to purify KAP as described (Hanlon and Barford, 1998). Mutagenesis of pThr160-CDK2 and CDK7 to produce KAP interaction mutants (KM) were performed using the Site Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) and following manufacturer’s instructions. These mutants were expressed and purified adopting the same procedures used for wild-type proteins. Mass spectrometric analyses were performed on VG Fisons “Platform” single quadrupole mass spectrometer using samples desalted by drop dialysis.

Kinase Assays CDK7, CDK7/cyclin H, and CDK7/cyclin H/MAT228 activities were measured by following the incorporation of radiolabeled phosphate into substrate. CDK2 (2.6 g 76 pmol) or GST-CTD (5 g 76pmol) as substrates were incubated with 100 ng of CDK7, CDK7/cyclin H, or CDK7/cyclin H/MAT228 in 10 l kinase buffer (0.1 mM ATP, 10 mM MgCl2, 50 mM Tris-HCl (pH 7.5), 1 Ci -32P-labeled ATP (Amersham Biosciences). The reaction mixtures were incubated for 5 min at 20 C and terminated by the addition of SDS sample buffer. Autophosphorylated CDK7/cyclin H and CDK7/cyclin H/MAT228 were obtained by incubating the each complex in 50 mM Tris-HCl (pH 7.5), 0.1 mM ATP, and 10 mM MgCl2 for 30 min at 20 C. Fully phosphorylated CDK7, CDK7/cyclin H, and CDK7/cyclin H/MAT228 were obtained incubating each with 50 ng of pThr160-Cdk2/cyclin A in 50 mM Tris-HCl (pH 7.5), 0.1 mM ATP, and 10 mM MgCl2 for 30 min at 20 C. Samples were analyzed on 12% SDS–PAGE gels and exposed to the PhosphorImager image plate (Molecular Dynamics, Amersham Biosciences). Pixels were quantified using the ImageJ software (NIH) and converted into MBq using the equation: pixels A ln(Mbq) B. A and B are experimental parameters calculated from the image plate calibration. Data were normalized with respect to the enzyme concentration. KAP Assay Dephosphorylation reactions were performed by incubating 0.5 g of pThr160-CDK2 or KM-pThr160-CDK2 with 0.2 g of KAP or 2 g of CDK7 or KM-CDK7 with 0.8 g of KAP in 50 mM Tris (pH 7.5), 150 mM NaCl, 2 mM DTT, and 0.5 mM EDTA for 1 hr at room temperature. Samples were then loaded on SDS-PAGE gel and analyzed for bandshift by Coomassie blue stain. Crystallization Sitting drop crystallization trials were performed at 4 C using Molecular Dimensions (Soham, UK) Structure Screen I and II. Drops were set up using Genesis ProTeam 150 crystallization robot (Tecan, Reading, UK) and CrystalQuick 96-well plates (Greiner bio-one, Longwood, FL) mixing 0.2 l of CDK7 (7 mg/ml) with 0.2 l of testing solutions. Drops were equilibrated by vapor diffusion with 100 l of its testing solution in the plate reservoir. Final crystallization conditions were identified as 0.1 M Na Citrate (pH 6.4), 0.2 M Na Acetate (pH 6.4), 20% PEG 4000, 10% glycerol, 200 mM nondetergent sulphobetaine 201 (NDSB201), and 2 mM ATP. The solubilizing agent NDSB201 helped prevent excessive nucleation and promoted crystal growth. The crystals were transferred to 0.1 M Na Citrate (pH 6.4), 0.2 M Na Acetate (pH 6.4), 20% PEG 4000, 20% glycerol before freezing in liquid nitrogen. X-Ray Data Collection, Data Processing, and Structure Solution The crystals were small ( 50 10 5 m). An initial dataset was ? collected at beamline ID29, ESRF to 3.3 A (Table 1.). The crystals were space group P21 with four molecules per asymmetric unit. A second data set was collected at the micro-focus beamline, ID13 (ESRF) using a focal spot size of 10 m . The high brightness of the ? beam allowed data collection to 3 A resolution (Table 1). However, severe radiation damage allowed only 5–10 degrees of data to be collected from one part of the crystal before it was necessary to translate the crystal and continue data collection using an unexposed region. Each of these data segments was indexed separately with the program MOSFLM (Leslie, 1992). Only 80% overall com? pleteness was achieved because of radiation damage. The 3 A data ? set was scaled together with the 3.3 A dataset previously recorded. The combined data set with addition of reflections in both the high and low resolution shells resulted in data that gave a successful structure solution. The structure was solved by molecular replacement with the program MOLREP (CCP4, 1994) using a modified model of inactive CDK2 (Protein Data Bank accession code 1HCK) as the search object. MOLREP found three copies of the model (Rfactor 0.513, CC 0.294). Examination of the resulting electron density maps revealed a further copy. A search in MOLREP using the previous solution as fixed input and a search model corresponding to the C-terminal domain of CDK2 (residues 104–152, 164–237, and 240–

Crystal Structure of Human CDK7 2077

298) successfully identified the orientation of the fourth molecule (Rfactor 50.5, CC 0.306). The difficulty in locating the fourth molecule arose because of a clash between the second and fourth molecules in the asymmetric unit in the region following the hinge corresponding to residues 97–100 of CDK2. This initially suggested that there were two different conformations of CDK7 present within the asymmetric unit. However, upon rebuilding it became apparent that CDK7 differed from CDK2 in this region and the same backbone conformation was in fact present in all four copies. As a result, 4-fold averaging was applied to maps during rebuilding. Rebuilding was carried out using O (Jones et al., 1991), followed by rounds of refinement in REFMAC5 (Murshudov et al., 1997) resulting in a final structure with Rfactor 0.219, Rfree 0.296 (Table 1). Acknowledgments We wish to thank the scientists at ESRF for their help with data collection, namely David Flot and Christian Riekel at ID13 and the scientists at ID29. We are grateful to David Barford, Robert Fisher, Andre Furger, Ernst Laue, David Morgan, and Erich Nigg for their gifts of plasmids. We acknowledge the contributions of John Sinclair in the early stages of this work. GL is supported by the BBSRC CASE award with Pharmacia Nerviano. This work has been supported by the MRC (L.N.J.). The authors declare that they have no conflicting financial interests. Received: July 19, 2004 Revised: August 16, 2004 Accepted: August 22, 2004 Published: November 9, 2004 References Akoulitchev, S., and Reinberg, D. (1998). The molecular mechanism of mitotic inhibition of TFIIH is mediated by phosphorylation of CDK7. Genes Dev. 12, 3541–3550. Akoulitchev, S., Chuikov, S., and Reinberg, D. (2000). TFIIH is negatively regulated by cdk8-containing mediator complexes. Nature 407, 102–106. Bax, B., Carter, P.S., Lewis, C., Guy, A.R., Bridges, A., Tanner, R., Pettman, G., Mannix, C., Culbert, A.A., Brown, M.J., et al. (2001). The structure of phosphorylated GSK-3beta complexed with a peptide, FRATide, that inhibits beta-catenin phosphorylation. Structure 9, 1143–1152. Beattie, J.F., Breault, G.A., Ellston, R.P., Green, S., Jewsbury, P.J., Midgley, C.J., Naven, R.T., Minshull, C.A., Pauptit, R.A., Tucker, J.A., and Pease, J.E. (2003). Cyclin-dependent kinase 4 inhibitors as a treatment for cancer. Part 1: identification and optimisation of substituted 4,6-bis anilino pyrimidines. Bioorg. Med. Chem. Lett. 13, 2955–2960. Breault, G.A., Ellston, R.P., Green, S., James, S.R., Jewsbury, P.J., Midgley, C.J., Pauptit, R.A., Minshull, C.A., Tucker, J.A., and Pease, J.E. (2003). Cyclin-dependent kinase 4 inhibitors as a treatment for cancer. Part 2: identification and optimisation of substituted 2,4-bis anilino pyrimidines. Bioorg. Med. Chem. Lett. 13, 2961–2966. Brotherton, D.H., Dhanaraj, V., Wick, S., Brizuela, L., Domaille, P., Volyanik, E., Xu, X., Parisini, E., Smith, B.O., Archer, S.J., et al. (1998). Crystal structure of the complex of the cyclin D-dependent kinase Cdk6 bound to the cell-cycle inhibitor p19INK4d. Nature 395, 244–250. Brown, N.R., Noble, M.E.M., Endicott, J.A., and Johnson, L.N. (1999a). The structural basis for specificity of substrate and recruitment peptides for cyclin-dependent kinases. Nat. Cell Biol. 1, 438–443. Brown, N.R., Noble, M.E.M., Lawrie, A.M., Morris, M.C., Tunnah, P., Divita, G., Johnson, L.N., and Endicott, J.A. (1999b). Effects of phosphorylation of threonine 160 on cyclin-dependent kinase 2 structure and activity. J. Biol. Chem. 274, 8746–8756. Busso, D., Keriel, A., Sandrock, B., Poterszman, A., Gileadi, O., and Egly, J.-M. (2000). Distinct regions of MAT1 regulate cdk7 kinase and TFIIH transcrition activities. J. Biol. Chem. 275, 22815–22823.

CCP4 (Collaborative Computational Project Number 4) (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D Biol. Crystallogr. 50, 760–763. Chang, C.I., Xu, B.E., Akella, R., Cobb, M.H., and Goldsmith, E.J. (2002). Crystal structures of MAP kinase p38 complexed to the docking sites on its nuclear substrate MEF2A and activator MKK3b. Mol. Cell 9, 1241–1249. Chen, J., Larochelle, S., Li, X., and Suter, B. (2003). Xpd/Ercc2 regulates CAK activity and mitotic progression. Nature 424, 228–232. Cheng, A., Kaldis, P., and Solomon, M.J. (2000). Dephosphorylation of human cyclin-dependent kinases by protein phosphatase type 2C and 2 isoforms. J. Biol. Chem. 275, 34744–34749. Cheng, K.-Y., Lowe, E.D., Sinclair, J., Nigg, E.A., and Johnson, L.N. (2003). The crystal structure of the human polo-like kinase polobox domain and its complex with a phosphopeptide. EMBO J. 22, 5757–5768. Dahmus, M.E. (1996). Reversible phosphorylation of the C-terminal domain of RNA polymerase II. J. Biol. Chem. 271, 19009–19012. Dajani, R., Fraser, E., Roe, S.M., Yeo, M., Good, V.M., Thompson, V., Dale, T.C., and Pearl, L.H. (2003). Structural basis for recruitment of glycogen synthase kinase 3 beta to the axin-APC scaffold complex. EMBO J. 22, 494–501. Davies, T.G., Bentley, J., Arris, C.E., Boyle, F.T., Curtin, N.J., Endicott, J.A., Gibson, A.E., Golding, B.T., Griffin, R.J., Hardcastle, I.R., et al. (2002). Structure-based design of a potent purine-based cyclin dependent kinase inhibitor. Nat. Struct. Biol. 9, 745–749. Davis, S.T., Benson, B.G., Bramson, H.N., Chapman, D.E., Dickerson, S.H., Dold, K.M., Eberwein, D.J., Edelstein, M., Frye, S.V., Gampe, R.T., Jr., et al. (2001). Prevention of chemotherapy-induced alopecia in rats by CDK inhibitors. Science 291, 134–137. De Bondt, H.L., Rosenblatt, J., Jancarik, J., Jones, H.D., Morgan, D.O., and Kim, S.-H. (1993). Crystal structure of cyclin dependent kinase 2. Nature 363, 592–602. Devault, A., Martinez, A.-M., Fesquet, D., Labbe, J.-C., Morin, N., ? Tassan, J.-P., Nigg, E.A., Cavadore, J.-C., and Doree, M. (1995). ? MAT1 (menage a trois) a new RING finger protein subunit stabilising cyclin H-CDK7 complexes in starfish and xenopus CAK. EMBO J. 14, 5027–5036. Elia, A.E., Rellos, P., Haire, L.F., Chao, J.W., Ivins, F.J., Hoepker, K., Mohammad, D., Cantley, L.C., Smerdon, S.J., and Yaffe, M.B. (2003). The molecular basis for phosphodependent substrate targeting and regulation of Plks by the Polo-box domain. Cell 115, 83–95. Fabbro, D., Ruetz, S., Buchdunger, E., Cowan-Jacob, S.W., Fendrich, G., Liebetanz, J., Mestan, J., O’Reilly, T., Traxler, P., Chaudhuri, B., et al. (2002). Protein kinases as targets for anticancer agents: from inhibitors to useful drugs. Pharmacol. Ther. 93, 79–98. Feaver, W.J., Svejstrup, J.Q., Henry, N.L., and Kornberg, R.D. (1994). Relationship of CDK-activating kinase and RNA polymerase II CTD kinase TFIIH/TFIIK. Cell 79, 1103–1109. Fisher, R.P., and Morgan, D.O. (1994). A novel cyclin associates with MO15/CDK7 to form the CDK-activating kinase. Cell 78, 713–724. Fisher, R.P., Jin, P., Chamberlin, H.M., and Morgan, D.O. (1995). Alternative mechanisms of CAK assembly require an assembly factor or an activating kinase. Cell 83, 47–57. Goodrich, J.A., and Tjian, R. (1994). Transcription factors IIE and IIH and ATP hydrolysis direct promoter clearance by RNA polymerase II. Cell 77, 145–156. Gray, N.S., Wodicka, L., Thunnissen, A.M., Norman, T.C., Kwon, S., Espinoza, F.H., Morgan, D.O., Barnes, G., LeClerc, S., Meijer, L., et al. (1998). Exploiting chemical libraries, structure, and genomics in the search for kinase inhibitors. Science 281, 533–538. Harper, J.W., and Elledge, S.J. (1998). The role of Cdk7 in CAK function, a retro-retrospective. Genes Dev. 12, 285–289. Hanlon, N. and Barford, D. (1998) Purification and crystallisation of the CDK-associated protein phosphatase, KAP expressed in Escherichia coli. Protein Sci. 7, 508–511. Honma, T., Hayashi, K., Aoyama, T., Hashimoto, N., Machida, T., Fukasawa, K., Iwama, T., Ikeura, C., Ikuta, M., Suzuki-Takahashi, I.,

Structure 2078

et al. (2001a). Structure-based generation of a new class of potent Cdk4 inhibitors: new de novo design strategy and library design. J. Med. Chem. 44, 4615–4627. Honma, T., Yoshizumi, T., Hashimoto, N., Hayashi, K., Kawanishi, N., Fukasawa, K., Takaki, T., Ikeura, C., Ikuta, M., Suzuki-Takahashi, I., et al. (2001b). A novel approach for the development of selective Cdk4 inhibitors: library design based on locations of Cdk4 specific amino acid residues. J. Med. Chem. 44, 4628–4640. Ikuta, M., Kamata, K., Fukasawa, K., Honma, T., Machida, T., Hirai, H., Suzuki-Takahashi, I., Hayama, T., and Nishimura, S. (2001). Crystallographic approach to identification of cyclin-dependent kinase 4 (CDK4)-specific inhibitors by using CDK4 mimic CDK2 protein. J. Biol. Chem. 276, 27548–27554. Jeffrey, P.D., Russo, A.A., Polyak, K., Gibbs, E., Hurwitz, J., Massague, J., and Pavletich, N.P. (1995). Mechanism of CDK activation revealed by the structure of a cyclin A-CDK2 complex. Nature 376, 313–320. Jeffrey, P.D., Tong, L., and Pavletich, N. (2001). Structural basis of inhibition of CDK-cyclin complexes by INK4 inhibitors. Genes Dev. 14, 3115–3125. Johnson, L.N., Noble, M.E.M., and Owen, D.J. (1996). Active and inactive protein kinases. Cell 85, 149–158. Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Improved method for building models in electron density maps and the location of errors in these models. Acta Crystallogr. A47, 110–119. Kaldis, P., Russo, A.A., Chou, H.S., Pavletich, N.P., and Solomon, M.J. (1998). Human and yeast cdk-activating kinases (CAKs) display distinct substrate specificities. Mol. Cell. Biol. 9, 2545–2560. Kwek, K.Y., Murphy, S., Furger, A., Thomas, B., O’Gorman, W., Kimura, H., Proudfoot, N.J., and Akoulitchev, A. (2002). U1 snRNA associates with TFIIH and regulates transcriptional initiation. Nat. Struct. Biol. 9, 800–805. Larochelle, S., Pandur, J., Fisher, R.P., Salz, H.K., and Suter, B. (1998). Cdk7 is essential for mitosis and for in vivo Cdk-activating kinase activity. Genes Dev. 12, 370–381. Larochelle, S., Chen, J., Knights, R., Pandur, J., Morcillo, P., Erdjument-Bromage, H., Tempst, P., Suter, B., and Fisher, R.P. (2001). T-loop phosphorylation stabilizes the CDK7-cyclin H-MAT1 complex in vivo and regulates its CTD kinase activity. EMBO J. 20, 3749–3759. Leslie, A.G.W. (1992). MOSFLM. In Joint CCP4 and ESF-EACMB Newsletter on Protein Crystallography (Warrington, U.K.: Daresbury Laboratory). Lowe, E.D., Tews, I., Cheng, K.Y., Brown, N.R., Gul, S., Noble, M.E., Gamblin, S.J., and Johnson, L.N. (2002). Specificity determinants of recruitment peptides bound to phospho-CDK2/cyclin A. Biochemistry 41, 15625–15634. Lu, H., Zawel, L., Fisher, L., Egly, J.M., and Reinberg, D. (1992). Human general transcription factor IIH phosphorylates the C-terminal domain of RNA polymerase II. Nature 358, 641–645. ? Martinez, A.-M., Afshar, M., Martin, F., Cavadore, J.-C., Labbe, J.-C., ? and Doree, M. (1997). Dual phosphorylation of the T-loop in CDK7: its role in controlling cyclin H binding and CAK activity. EMBO J. 16, 343–354. Murshudov, G.N., Vagen, A.A., and Dodson, E.J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D 53, 240–255. Nishiwaki, E., Turner, S.L., Harju, S., Miyazaki, S., Kashiwagi, M., Koh, J., and Serizawa, H. (2000). Regulation of CDK7-carboxyl-terminal domain kinase activity by the tumor suppressor p16(INK4A) contributes to cell cycle regulation. Mol. Cell. Biol. 20, 7726–7734. Noble, M.E., Endicott, J.A., and Johnson, L.N. (2004). Protein kinase inhibitors: insights into drug design from structure. Science 303, 1800–1805. Oelgeschlager, T. (2002). Regulation of RNA polymerase II activity by CTD phosphorylation and cell cycle control. J. Cell. Physiol. 190, 160–169. Ohkawa, K., Ishida, H., Nakanishi, F., Hosui, A., Ueda, K., Takehara, T., Hori, M., and Hayashi, N. (2004). Hepatitis C virus core functions

as a suppressor of cyclin-dependent kinase-activating kinase and impairs cell cycle progression. J. Biol. Chem. 279, 11719–11726. Palancade, B., and Bensaude, O. (2003). Investigating RNA polymerase II carboxyl-terminal domain (CTD) phosphorylation. Eur. J. Biochem. 270, 3859–3870. Poon, R.Y.C., and Hunter, T. (1995). Dephosphorylation of Cdk2 Thr160 by the cyclin dependent kinase-interacting phosphatase KAP in the absence of cyclin. Science 270, 90–93. Poon, R.Y.C., Yamashita, K., Howell, M., Erschler, M.A., Belyavsky, A., and Hunter, T. (1994). Cell cycle regulation of the p34cdc2/p33cdk2activating kinase p40Mo15. J. Cell Sci. 107, 2789–2799. Rosales, J., Han, B., and Lee, K.Y. (2003). Cdk7 functions as a cdk5 activating kinase in brain. Cell. Physiol. Biochem. 13, 285–296. Roy, R., Adamczewski, J.P., Seroz, T., Vermuelen, W., Tassan, J.P., Schaeffer, L., Nigg, E.A., Hoeijmakers, J.H., and Egly, J.M. (1994). The MO15 cell cycle kinase is associated with the TFIIH transcription-DNA repair factor. Cell 79, 1093–1101. Russo, A., Jeffrey, P.D., and Pavletich, N.P. (1996a). Structural basis of cyclin dependent kinase activation by phosphorylation. Nat. Struct. Biol. 3, 696–700. Russo, A.A., Jeffrey, P.D., Patten, A.K., Massague, J., and Pavletich, N.P. (1996b). Crystal structure of the p27Kip1 cyclin-dependentkinase inhibitor bound to the cyclin A-CDK2 complex. Nature 382, 325–331. Russo, A.A., Tong, L., Lee, J.O., Jeffrey, P.D., and Pavletich, N.P. (1998). Structural basis for inhibition of the cyclin-dependent kinase Cdk6 by the tumour suppressor p16INK4a. Nature 395, 237–243. Schiekhattar, R., Mermelstein, F., Fisher, R.P., Drapkin, R., Dynlacht, B., Wessling, H.C., Morgan, D.O., and Reinberg, D. (1995). CDKactivating kinase complex is a component of human transcription factor TFIIH. Nature 374, 283–287. Schulze-Gahmen, U., and Kim, S.H. (2002). Structural basis for CDK6 activation by a viral-encoded cyclin. Nat. Struct. Biol. 9, 177–181. Serizawa, H. (1998). Cyclin-dependent kinase inhibitor p16INK4A inhibits phosphorylation of RNA polymerase II by general transcription factor TFIIH. J. Biol. Chem. 273, 5427–5430. Serizawa, H., Makela, T.P., Conaway, J.W., Conaway, R.C., Weinberg, R.A., and Young, R.A. (1995). Association of Cdk-activating kinase subunits with transcription factor TFIIH. Nature 374, 280–282. Sharrocks, A.D., Yang, S.H., and Galanis, A. (2000). Docking domains and substrate-specificity determination for MAP kinases. Trends Biochem. Sci. 25, 448–453. Shilatifard, A., Conaway, R.C., and Conaway, J.W. (2003). The RNA polymerase II elongation complex. Annu. Rev. Biochem. 72, 693–715. Solomon, M.J., Lee, T., and Kirschner, M.W. (1992). The role of phosphorylation in p34cdc2 activation: identification of an activating kinase. Mol. Biol. Cell 3, 13–17. Song, H., Hanlan, N., Brown, N.R., Noble, M.E.M., Johnson, L.N., and Barford, D. (2001). Phospho-protein/protein interactions: the crystal structure of Kinase Associated Phosphatase (KAP) in complex with CDK2. Mol. Cell 7, 615–626. Soni, R., O’Reilly, T., Furet, P., Muller, L., Stephan, C., ZumsteinMecker, S., Fretz, H., Fabbro, D., and Chaudhuri, B. (2001). Selective in vivo and in vitro effects of a small molecule inhibitor of cyclindependent kinase 4. J. Natl. Cancer Inst. 93, 436–446. Tarricone, C., Dhavan, R., Peng, J., Areces, L.B., Tsai, L.H., and Musacchio, A. (2001). Structure and regulation of the CDK5-p25 (nck5a) complex. Mol. Cell 8, 657–669. Tassan, J., Schultz, S.J., Bartek, J., and Nigg, E.A. (1994). Cell cycle analysis of the activity, subcellular localisation, and subunit composition of human CAK. J. Cell Biol. 127, 467–478. Tassan, J.-P., Jaquenoud, M., Fry, A.M., Frutger, S., Hughes, G.J., and Nigg, E.A. (1995). In vitro assembly of a functional human CDK7cyclin H complex requires MAT1, a novel 36 kDa RING finger protein. EMBO J. 14, 5608–5617.

Crystal Structure of Human CDK7 2079

Wallenfang, M.R., and Seydoux, G. (2002). cdk-7 Is required for mRNA transcription and cell cycle progression in Caenorhabditis elegans embryos. Proc. Natl. Acad. Sci. USA 99, 5527–5532. Watanabe, Y., Fujimoto, H., Watanabe, T., Maekawa, T., Masutani, C., Hanaoka, F., and Ohkuma, Y. (2000). Modulation of TFIIH-associated kinase activity by complex formation and its relationship with CTD phosphorylation of RNA polymerase II. Genes Cells 5, 407–423. Yankulov, K.Y., and Bentley, D. (1997). Regulation of CDK7 substrate specificity by MAT1. EMBO J. 16, 1638–1646. Accession Numbers The coordinates have been deposited in the Protein Data Bank (code 1UA2).


相关文章:
更多相关标签: