- Email: [email protected]
Cyclin-dependent kinase 7: at the cross-roads of transcription, DNA repair and cell cycle control?
312
Cyclin-dependent kinase 7: at the cross-roads of transcription, DNA repair and cell cycle control? Erich A Nigg Cyclin-dependent
kinase (CDK)
7 was originally implicated
in cell cycle control by virtue of its ability to phosphorylate and activate other CDKs. Subsequently,
both CDK7
its partner, cyclin H, were found to be associated general transcription for CDK7
factor TFIIH, suggesting
in transcription
and RNA polymerase
additional
roles
and DNA repair. During the past
year, a third subunit associated been characterized,
and
with the
with CDK7
and
cyclin H has
and the functional link between
CDK7
in although a low level of activity can be detected unphosphorylated CDKZ-cyclin A complexes [3’,7,8]. In the case of CDKl, phosphorylation at Thr161 appears to coincide with the onset of cyclin binding, and, conversely, dephosphorylation occurs concomitantly with cyclin destruction upon exit from mitosis. One phosphatase able to dephosphorylate the T-loop residue in monomeric CDKs has recently been identified [9’].
II has been strengthened.
Address Department of Molecular Biology, University of Geneva, Sciences II, 30, quai Ernest-Ansermet, Cl-i-1 211 Geneva 4, Switzerland; e-mail: niggasc2a.unige.ch Current Opinion in Call Biology 1996, 8:312-317 0 Current Biology Ltd ISSN 0955-0674 Abbreviations cyclin dependent kinase activating kinase CAK cyclin-dependent kinase CDK carboxy-terminal domain CTD RNAPII RNA polymerase II suppressor of RNA polymerase B SRB transcription factor IIH TFIIH mitogen-activated protein kinase MAPK ‘m&age-a-trois’ 1 MAT1
Introduction Cyclin-dependent kinases (CDKs) are best known for their fundamental roles in cell cycle regulation. However, some members of the CDK family are involved in other cellular processes, notably in transcription. In vertebrates, cell cycle functions are firmly established for CDKl (the homolog of Schizosaccharotnycespombe Cdc2 and Saccharomycescemisiae CDCZS), CDKZ, CDK4 and CDK6, and such functions are likely for CDK3, whilst the jury is still out for CDKS, CDK7, and CDK8 [1,2,3*,4*]. The particular case of CDK7 constitutes the focus of this review. This nuclear kinase was originally identified as the catalytic subunit of a complex that is able to phosphorylate and thereby activate several cell cycle regulatory CDKs. However, recent findings suggest that the physiological role of CDK7 may be broader than originally surmised, or perhaps different altogether.
T-loop phosphorylation:
a common theme
In order to be active, most CDKs require not only a cyclin partner but also phosphorylation at one particular site, which corresponds to Thr161 in human CDKl, and which is located within the so-called T-loop of kinase subdomain VIII [3*,51,6~~].CDKl, CDK2 and CDK4 all require T-loop phosphorylation for maximum activity,
T-loop phosphorylation is important not only for the activation of CDKs but also for the activation of many other protein kinases [lo’]. From the analysis of several crystal structures it appears that this modification favours a kinase conformation which allows the access of substrate to the active site [11,12]. As exemplified by CAMP-dependent protein kinase, T-loop phosphorylation can be constitutive and serve a primarily structural role. In other cases, it constitutes an important regulatory switch, as illustrated best by the members of the mitogen-activated protein kinase (MAPK) family. Hence the identification of kinases (and phosphatases) acting on T-loops constitutes a subject of considerable general interest.
CDK7: a kinase in a ‘mbnage+trois’ In 1993, one enzyme capable of phosphorylating Thr161 in CDKl (and the corresponding residue within the T-loops of other CDKs) was purified from both vertebrate and invertebrate species [13-151. This enzyme was termed the CDK-activating kinase (CAK). Sequence analyses led to the surprising conclusion that the catalytic subunit of CAK was itself structurally related to CDKs, and had already been cloned under the designation MO15 [16]. Subsequently, MO15 was shown to bind a bonajde cyclin partner, cyclin H, and hence was renamed CDK7 [17,18]. At that point, the available evidence suggested that the CDK7-cyclin H complex might function as an upstream element in a cascade of CDK-cyclins. Somewhat surprisingly, however, CDK7’s associated CAK activity was found to be essentially constant throughout the cell cycle, except for a somewhat reduced activity in quiescent cells [19-211. CDK7 interacts not only with cyclin H, but also, in near-stoichiometric amounts, with a second polypeptide [17,21]. During the past year, this 36-37 kDa protein has been cloned from human [22**], mouse [23**], Xetzopusand starfish [24@*]cDNA libraries. Its structure predicts that it contains an amino-terminally located putative zinc-binding domain conforming to a typical C3HC4 (single-letter code for amino acids) RING-finger motif. The precise functions of RING-finger motifs remain unknown, but they appear more likely to mediate protein-protein (or
Cyclin-dependent
protein-RNA) interactions than specific DNA binding [ZS]. In r&u reconstitution experiments indicate that the CDK7-associated 36 kDa RING-finger protein can function as an assembly factor, promoting a stable interaction between CDK7 and cyclin H [22**-24**]. Accordingly, the protein was named MAT1 (for ‘menage-a-trois’ 1 [Z?]). The RING-finger domain of MAT1 is not required for this ternary-complex formation [Z?], suggesting that it is available for mediating interactions with other, as yet unidentified, proteins. In &JO, MAT1 appears to form a trimeric complex with CDK7 and cyclin H under most circumstances [22”,26”], but CDK7-cyclin H heterodimers have also been observed [23~~,24*~,26**]. Presently, there is no evidence that MAT1 could act as an inhibitor of the catalytic activity of CDK7, but it would be premature to exclude the possibility that, in &JO, MAT1 may modulate CDK7’s activity, subcellular localization and/or substrate specificity. MAT1 appears to represent an assembly factor which is specific for CDK7 and cyclin H, as it has not yet been observed with other CDK-cyclin complexes. However, functionally related assembly factors may exist for other CDK-cyclin complexes [27,28*]. The T-loop of CDK7 itself contains a site (Thr170 in human CDK7) which is phosphorylated in z&o [20,29]. Whereas the interaction between CDK7 and cyclin H in a yeast two-hybrid system seems to depend on the phosphorylation of Thr170 [18], the MATl-mediated stabilization of the CDK7-cyclin H complex is independent of the phosphorylation state of Thr170 [23”,24**]. These results indicate that MAT1 is able to bypass the requirement for T-loop phosphorylation, and that CDK7-cyclin H complexes can form by two distinct mechanisms [23”]. The relative contribution of the two mechanisms in vtio is unknown, but it is attractive to speculate that proteins functionally resembling MAT1 may provide one solution to a fundamental problem: in spite of what Babuschka dolls would seem to suggest, kinase cascades operating through T-loop phosphorylation cannot extend indefinitely; some protein is needed to begin the cascade.
CDK7 is present in transcription
factor IIH
An entirely new perspective on CDK7 function was opened when CDK7 was identified as a subunit of transcription factor IIH (TFIIH) and shown to phosphorylate the carboxy-terminal domain (CTD) of RNA polymerase II (RNAPII) [30**]. TFIIH is a multiprotein complex required not only for class II transcription but also for nucleotide-excision repair [31*,32*]. Its associated CTD-kinase activity is considered important for the promoter-clearance step of transcription, but the precise structural consequences of the phosphorylation of the CTD remain the subject of some debate [31*,32’,33]. Cyclin H and MAT1 are also present in TFIIH [23**,26**,34**,35**], and it is not known what, if anything, distinguishes the TFIIH-associated form of CDK7 from the quantitatively predominant free form. Taking advantage of the intimate association of CDK7
kinase 7 Nigg
313
with TFIIH, anti-CDK7 antibodies were recently shown to constitute a powerful tool for the rapid immuno-isolation of a transcriptionally active RNAPII holoenzyme from a mammalian source [36”]. Whether CDK7 really displays dual-substrate specificity in &JO remains to be further explored, but there is no question that the CDK7-cyclin H-MAT1 complex is able to phosphorylate both the T-loop of CDKs and the YSPTSPS (single-letter code for amino acids) repeats of the RNAPII CTD in vitro. At the level of their primary structures these substrates bear no obvious similarities, raising the question of how they both can fit into the active site of the CDK7-cyclin H-MAT1 complex. Previous studies have identified several potential CTD kinases, including CDKl-cyclin B, mammalian MAPKs, and the S. cemvisiae CDK-cyclin complex CTKl-CTKZ [33,37’]. Furthermore, genetic and biochemical data implicate another budding yeast CDK-cyclin pair, SRBlO-SRBl 1 (where SRB denotes suppressor of RNA polymerase B), in CTD phosphorylation [38”], and the mammalian CDKg-cyclin C complex may perform a similar function [39’]. However, direct evidence for CTD phosphorylation by either SRBlO or CDKS has not yet been reported. Considering the multitude of potential CTD kinases, it remains an important but difficult task to determine the physiological role of each enzyme. The importance of the catalytic activity of TFIIHassociated CDK7 has recently been explored using in vitro transcription systems [40*,41”]. No requirement for CDK7 activity could be observed for the adenovirus major late promoter [40’], but this result is difficult to interpret as transcription from this viral promoter does not depend on the CTD. In contrast, transcription from the CTD-dependent mammalian dihydrofolate reductase promoter did require the presence of an active CDK7 in the added TFIIH preparation [41”]. Although these studies do not rigorously prove that CDK7 phosphorylates the CTD directly, they clearly demonstrate a requirement for CDK7 activity during in vitro transcription, at least for certain promoters. The possibility that the CDK7-cyclin H-MAT1 complex may also function in DNA-repair processes has not yet been examined in depth. However, preliminary data do support a role of CDK7 in nucleotide-excision repair [30”,35**], and recent studies indicate that TFIIHassociated CDK7 activity drops in response to UV irradiation of cells [26”]. This latter observation is particularly intriguing because UV irradiation did not affect the activity of free (i.e. non-TFIIH-associated) CDK7. These observations clearly encourage more detailed studies into the possibility that TFIIH-associated CDK7 may function in signalling pathways that link the recognition of DNA damage to transcriptional responses, DNA repair, and, perhaps, cell cycle control.
Nucleus and gene expression
314
Lessons from yeast?
Figure 1
Taken at face value, the biochemical evidence obtained for the CDK7-cyclin H-MAT1 complex would indicate that CDK7 may act on both the T-loops of multiple CDKs and the CTD of RNAPII (see Fig. la). However, most of the evidence supporting this apparent dual specificity has been obtained from in vitro studies, and it is not clear whether both CAK activity and CTD-kinase activity constitute true physiological functions of this complex. Thus, it may be interesting to briefly consider two alternative scenarios (see Fig. 1b,c). On the one hand, one could argue that CDK7 functions primarily as a CAK, even in its TFIIH-associated form (Fig. lb); specifically, CDK7 might activate other CDK-cyclin complexes (e.g. SRBlO-SRBll or CTKl-CTKZ), and these might in turn phosphorylate the RNAPII CTD. On the other hand, an opposite view would emphasize the RNAPII-related role of CDK7 rather than its cell cycle role (Fig. lc). According to this model, other kinases would accomplish the CAK function in vim.
(a) CAK
, @
D
@
TFllH
9 pRb
$??%I
@
9
0
Lamin
RNAPII CTD
(b)
CAK GB c&k
Table 1 @
@
@ Potential yeasts.’
RNAPII CTD
H-MAT1
complex In
Metazoan organisms
S. cefetisiae
S. pombe
Catalytic subunit
CDK7 (MO1 5)
KIN26
Crkl /Mop1 7
Cyclin subunit
Cyclin H
CCL1
Mcs2
Assembly
MAT1
7
7
Yes
No
Yes
Yes
Yes
0
9
0 pRb
homologs of the CDK7-cyclln
Lamin
(c)
CAK @
subunit
CAK activity CID-klnasa
‘See references [17,16,22**,23**,42*,43’,44-461. to be allelic to Mcs6 [42’,501.
@$$@
9 pRb
9 RNAPII CTD
figure illustrating three extreme ways to think
about the physiological function of the CDK7-cyclin
tCrk1 /Mop1 is likely
Q Lamin
0 1996 Current Op~mon in Cell Biology
A highly schematic
activity Yes
H-MAT1
(CDK7)
complex. (a) The complex may display dual specificity as both a CAK and a CTD kinase (as part of TFIIH), not only in vitro but also in ho. According to this model, CDK7 would act upstream of other CDKs in addition to phosphotylating a non-CDK substrate; specifically, CDK7 is believed to phosphorylate the RNAPII CTD, much as CDK4-cyclin D or CDKl -cyclin B phosphorylates the retinoblastoma gene product (pRb) or nuclear lamins (Lamin), respectively. In contrast, (b) proposes that CDK7 functions primarily as a CAK, both for CDK-cyclin complexes involved in cell cycle control and for those CDK-cyclin complexes that might act as CTD kinases (represented by a shaded circle). Finally, (c) emphasizes the RNAPII-related role of CDK7 and postulates the existence of other kinases with CAK activity (represented by a shaded circle). (For the sake of clarity, only two out of several cell cycle regulatory CDK-cyclin complexes have been depicted downstream of CAK). Open arrows represent activation of one complex by another, or some other form of catalysis. Closed arrows represent the changing role of one complex.
Considering the present state of uncertainty about the function of CDK7, one might hope for illumination from studies in genetically tractable organisms. Potential homologs of CDK7 and cyclin H have indeed been identified in both fission and budding yeasts (see Table 1). In S. pot&, a complex formed between Crkl/Mopl (the putative CDK7 homolog) and Mcs2 (the putative cyclin H homolog) displays both CAK and CTD-kinase activity [42*,43*]. Crkl/Mopl is essential for viability [42*,43’], but the lethal phenotype is heterogeneous and might result from either defects in cell cycle progression, or defects in transcriptional regulation, or defects in both processes. Partial loss of Crkl/Mopl-Mcs2 kinase activity suppresses the mitotic-catastrophe phenotype caused by hyperactivated Cd&-Cdcl3 (which is the mitotic CDK-cyclin complex of S. porn&) [42’,43’,44]. These results clearly demonstrate a genetic interaction between Crkl/Mopl-Mcs2 and Cd&-Cdcl3, but they do not prove that the observed suppression reflects reduced
Cyclin-dependent
phosphorylation of the T-loop of Cd&. One could also argue that reduced phosphorylation of RNAPII might result in a lowered transcription rate of the genes encoding Cd&? and Cdcl3. In S. cerevisiae,the closest structural relatives of CDK7 and cyclin H are KIN28 and CCLl, respectively [45,46,47’]. The KIN28-CCL1 complex associates with TFIIH and readily phosphorylates the RNAPII CTD, but does not seem to display CAK activity [47*,48**,49*]. This might explain why S. cemisiae KIN28 cannot complement mutations in the S. pombe Crkl/Mopl gene, although vertebrate CDK7 can [42*,43’]. Loss of KIN28 function does not induce a uniform cell cycle arrest, but instead leads to a rapid decline in nascent mRNA synthesis [49’], supporting the view that the KIN28-CCL1 complex is a physiologically relevant CTD kinase. On the other hand, budding yeast extracts do contain CAK activity, albeit not associated with KIN28-CCLl [49*]. The purification and characterization of this CAK activity is eagerly awaited. Knowledge of the structure of the budding yeast CAK might help answer many questions concerning mammalian CAK, and -somewhat ironicallymight provide an illustration of the awesome power of yeast biochemistry.
Conclusions With the identification of MATl, a novel mechanism for the assembly of CDK-cyclin complexes has been revealed. Also, the evidence accumulates for a function of the TFIIH-associated CDK7-cyclin H-MAT1 complex in class II transcription and, possibly, nucleotide-excision repair. In contrast, the mechanisms involved in regulating this complex remain mysterious, and the cell cycle functions of the complex are still poorly understood. Answers to many of the open questions may come from a biochemical analysis of S. pombe Crkl/Mopl and a molecular characterization of an S. cerevisiae CAK.
References
and recommended
reading
5.
l
*
Jeffrey PD, Russo AA, Polysk K, Gibbs E, Hurwitz J, Massagub J, Pavletich NP: Mechanism of CDK activation revealed by the structure of a cyclinA-CDK2 complex. Nature 1995, 376:313-320. The crystal structure of a human CDKP-cyclin A complex is determined at 2.3 A resolution. This work provides the basis for a detailed understanding of the regulation and substrate specificity of CDK-cyclin complexes. The reported structure concerns unphosphorylated CDKP; it will be interesting to see how phosphorylation of Thrl60 alters the conformation of the T-loop of CDKP. 7.
Connell-Cowley L, Solomon MJ, Wei N, Harper JW: Phosphorylation independent activation of human cyclindependent kinase 2 by cyclin A in vitro. MO/ Biol Cell 1993, 4:79-92.
8.
Matsuoka M, Kate JY, Fisher RP, Mor an DO, Sherr C: Activation of cyclin-dependent kinase 4 (cdk4 B by mouse M015associated klnase. MO/ Cell Biol 1994, 14:7265-7275.
9. .
Poon RYC, Hunter T: Dephosphoryfation of Cdk2 Thrlss by the cyclin-dependent kinasa-interacting phosphatase KAP in the absence of cyclln. Science 1995, 27Q:QO-93. The phosphatase CDK-associated phosphatass (KAP) was originally identified in yeast two-hybrid screens as a protein which interacts with CDKs. Structurally, it resembles dual-specificity phosphatsses such as Cdc25. As shown here, however, it specifically dephosphorylates Thrl60 (but not Tyrl5) in monomeric Cdkl, at least when assayed in vitro. Binding to cyclin A protects Cdk2 from dephosphorylation. Thus, if KAP acts on CdkP in viva, it is likely to do so only after cyclin degradation. 10. .
Hanks SK, Hunter T: The eukaryotic protein kinase superfamily: kinase (catalytic) domain structure and classification. FASEB J 1995, Q:576-598. A comprehensive review of the primary structures of protein kinsses. Besides containing other useful information, this compilation illustrates that many protein kinasas have phosphorylatable residues in positions corresponding roughly to the T-loop-phosphorylation sites of CDKs. 11.
Taylor SS, Knighton DR, Zheng J, Lynn F, Eyck TF, Sowadski JM: Structural framework for the protein kinase family. Annu Rev Cell Biol 1992, 8:429-462.
12.
Morgan DO, De Bondt HL: Protein klnase regulation: insights from crystal structure analysis. Curr Opin Cell Viol 1994, 6:239-246.
13.
Fesquet D: The MO1 5 gene encodes the catalytic subunit of a protein kinase that activates cdc2 and other cyclin-dependent kinases (CDKs) through phosphorylation of Thr161 and its homologues. EMBO J 1993, 12:3111-3121.
14.
Poon RYC, Ysmashita K, Adamczewski JP, Hunt T, Shuttleworth J: The cdc2-related orotein ~40M015 is the catalvtfc subunit of a protein kinase th& ten activate p33cdk* and 634dC*. EM60 J 1993, 12:3123-3132.
15.
Solomon MJ, Harper WJ, Shuttleworth J: CAK, the p34&* activating klnase, contains a protein identical to or closely related to p40Mols. EM30 J 1993, 12:3133-3142.
16.
Shuttleworth J, Godfrey R, Colman A: p40M015, a cdcl-related protein kinase involved in negative regulation of meiotfc maturation of Xenopus laevis oocytes. EM30 J 1990, 9:3233-3240.
1z
Fisher RP, Morgan DO: A novel cyclin associates with MO1 J/cdk7 to form the CDK-activating kinase. Cell 1994, 781713-724.
18.
Mlkelii TP, Tsssan JP, Nigg EA, Frutiger S, Hughes GJ, Weinberg RA: A cyclin associated with the CDK-activating klnase MO1 5. Nature 1994, 371:254-257.
19.
Brown Al, Jones T, Shuttleworth J: Expression and activity of p40M015, the catalytic subunit of cdk-activating kinase, during Xenopus oogenesis and embryogenesis. MO/ Viol Cell 1994, 5:921-932.
20.
Poon RYC, Ysmashita K, Howell M, Ershler MA, Belyavsky A, Hunt T: Cell cycle regulation of the p34c*2@3mk* activating kinase p4@01s. J Cell Sci 1994,107:2789-2799.
21.
Tssssn JP, Schultz SJ, Bsrtek J, Nigg EA: Cell cycle analysis of the activity, subcellular localisatlon, and subunlt composition of human CAK (CDK-activating kinase). J Cell Biol 1994, 127~467-478.
of special interest of outstanding interest
1.
Van den Heuvel S, Harlow E: Distinct roles for cyclln-dependent kinases in cell cycle control. Science 1993, 262:2050-2054.
2.
Sherr CJ: Gl phase progression: 79:551-555.
cycling on cue. Cell 1994,
Morgan DO: Principles of CDK regulation. Nature 1995, 374:131-134. i thorough review of the principles of CDK regulation.
3.
4. .
Nigg EA: Cyclin-dependent protein kinases: key regulators the eukaryotic cell cycle. Bioessays 1995, 17:471-480.
of
Solomon MJ: The function(s) of CAK, the p34d-activating klnase. Trends Biochem Sci 1994,19:496-500.
6. ..
Papers of particular interest, published within the annual period of review, have been highlighted as: .
315
A recent review of both the regulation and the substrates of CDK-cyclin complexes.
Acknowledgements I would like to thank Jean-Pierre Tassan and Andrew Fry for helpful comments on the manuscript. Due to space limitations, literature citation is not comprehensive and I apologize to all colleagues who received less explicit credit than they deserved. Work in the author’s laboratory was supported by grants from the Swiss National Science Foundation (grant 31-33615.92) and the Swiss Cancer League (grant FOR 447).
kinase 7 Nigg
316
22. ..
Nucleus and gene expression
Tassan JP, JaquenoudM, Fry AM, Frutiger S, Hughes GJ, Nigg EA: In vitro assembly of a functional human CDK7-cyclln H complex requires MATI, a novel 36 kDa RING finger protein. EM60 J 1QQ5,14:5608-561 Z
Translation of CDK7, cyclin H and MAT1 in a reticulocyte lysate yields no stable heterodimeric complexes between any two proteins. However, active ternary complexes can readily be isolated after mixing all three proteins, and complex formation occurs independently of the RING-finger domain of MATI. Ternary complexes phosphorylate both the T-loop in CDK2 and a peptide mimicking the CTD of RNAPII (see also [23”,24”,26”]). Fisher RP, Jin P, Chamberlin HM, Morgan DO: Alternative mechanisms of CAK assembly require an assembly factor or an activatina kinase. Cell 1995. 83:47-57. Using lysates from- baculovirus-infecte; insect cells, the authors show that the formation and activity of recombinant CDK’I-cyclin H complexes are substantially increased by the presence of MATl. Interestingly, the requirement for MAT1 can be bypassed by prior phosphorylation of T-loop residues within CDKZ usina either CDKI-cvclin A or CDKP-cvclin A as T-loop kinasesl These’resulk indicate the existence of alternative-pathways for CDK7-cyclin H complex assembly, and they raise the possibility that CDK-cyclin complexes other than CDK’/-cyclin H-MAT1 may display Tloop-directed kinase activity in viva. See also [22”,24”,26”]. 23. ..
Devault A, Martinez AM, Fesquet D, Labbe JC, Tassan JP, Nigg EA, Cavadore JC, Dor(te M: MATI, a new RING-finger protein subunit StabiliZinQ cyclin H-cdk7 complexes In starfish and Xenopus CAK. EM/30 J lQQ5,14:5027-5036. In vitro assambly of CDK7-cyclin H-MAT1 complexes is studied in a reticulocyte lysate sytem. MAT1 is shown to confer stability to a ternary complex regardless of the phosphorylation state of the CDKP T-loop. Moreover, evidence is presented that, at least in oocytes, dimeric CDK7-cyclin H complexes co-exist with trimeric MAT1 -containing complexes. See also [22” *23” I26”I.
24. ..
25.
Freemont PS: The RING finger -a
novel protein sequence motif related to the zinc finger. Annu NY Acad Sci 1993, 684:174-l
26. ..
92.
Adamczewski JP, Rossignol M, Tassan JP, Nigg EA, Moncollin V, Egly JM: MATI, cdk7 and cyclin H form a kinase complex which is UV light sensitive upon association with TFIIH. EMBO J 1QBB
.___, in nress r.___.
MAT1 is shown to be present in those CDK7-cyclin H complexes that associate with TFlIH. Furthermore, evidence is presented that TFIIH-associated (but not bulk!) CDK7 activity drops in response to UV irradiation of cells. These observations raise the possibility that CDKI might participate in a checkpoint signalling pathway that relays the presence of DNA damage to the transcriptional apparatus. 27.
Kato JY, Matsuoka M, Strom DK, Sherr CJ: Regulation of cyclin D-dependent kinase 4 (cdk4) by cdk4-activating klnase. MO/ Cell Biol 1994, 14~2713-2721.
Gerber MR, Farrell A, Raymond JD, Herskowitz I, Morgan DO: Cdc37 is required for association of the protein klnase Cdc28 with G1 and mitotlc cyclins. Proc Nat/ Acad Sci USA 1995, 92:4651-4655. The Cdc37 gene of S. cerevisiae has previously been shown to be required for passage through the G1 phase of the cell cycle. A strain harboring a mutation in Cdc37o is now reoorted to contain reduced levels of CdcPB-cyclin compl’exes. These ra’sults suggest that Cdc37p is somehow required for the formation of CDK-cyclin complexes in budding yeast. However, the Cdc37 gene product is not stably associated with Cdc28-cyclin complexes, suggesting that its mode of action is different from that of vertebrate MATI. 28. .
29.
Labb(, JC, Martinez AM, Fesquet D, Capony JP, Darbon JM, Derancouri J, Devault A, Morin N, Cavadore JC, Do& M: P40Mels associates with p36 subunit and requires both nuclear translocation and Thr176 phosphorylation to generate cdk-activating activity in Xenopus oocytes. EM80 J 1 QQ4, 13:5155-5164.
A recent review on TFIIH subunits and their involvement in transcription, DNA repair, and cell cycle regulation. See also [32*1. 32. .
Maldonado E, Reinberg D: News on initiation and elongation of transcription by RNA polymerase II. Curr Opin Cell Biol 1 QQ5, 7:352-361. See annotation [31 *I. 33.
34. ..
Serirawa H, M;ikelii TP, Conaway JW, Conaway RC, Weinberg RA, Young RA: Association of cdk-activating kinase subunits with transcription factor TFIIH. Nature 1 QQ5, 374:260-282. This paper shows that not only CDK7 but also cyclin H is a subunit of TRIH, and that this TFIIH-associated kinase phosphorylates both the T-loop of CDKs and the CTD of RNAF‘ll (see also [30”,35**]). 35. ..
Shiekhattar R, Mermelstein F, Fisher R, Drapkin R, Dynlacht B, Wessling HC, Morgan DO, Reinberg D: Cdk-activating kinase complex is a component of human transcription factor TFIIH. Nature 1 QQ5, 374~203-207. See annotations [30” 1 34”l . 36. ..
31. .
Seroz T, Hwang JR, Moncollin V, Egly JM: TFIIH: a link between transcription, DNA repair and cell cycle regulation. Gun Opin Gener Dev 1995, 5:217-221.
Ossipow V, Tassan JP, Nigg EA, Schibler U: A mammalian RNA polymerase II holoenzyme containing all components
required for promoter-specific transcription initiation. Cell 1QQ5, 63:137-l 46. This study shows that a monoclonal antiCDK7 antibody can be used for the rapid purification of a RNAPII holoenzyme from a highly concentrated rat liver nuclear extract. The antiCDK7 immunoprecipitate contains all components required for transcription initiation at both viral and cellular promoters. This is the first description of a mammalian RNAPII holoenzyme. 37.
.
Sterner DE. Lee JM. Hardin SE. Greenleaf AL: The veast carboxyl-t&mlnal r&eat dom& kinase CTDK-i Is a divergent cyclin-cyclin dependent klnate complex. MO/ Cell Biol 1 QQ5,
15:5716-5724. The budding yeast CTD kinase 1 is shown to contain a CDK-cyclin pair (CTKI-CTK2). in addition to a third subunit (CTK3). CTK2 is structurally related to mammalian cyclin C. CTK3 shares no obvious sequence similariiy with MATl, but it is noteworthy that two distinct CDKs with CTD-kinase activity (CDK7 in metazoans and CTKI in S. cerevisiae) both exist in heterotrimeric complexes. 38. ..
Liao SM, Zhang J, Jeffery DA, Koleske AJ, Thompson CM, Chao DM, Viljoen M, Van Vuuren JJ, Young RA: A kinase-cyclln pair in the RNA polymerase II holoenzyme. Nature lQQ5, 374:193-l 96. Two S. cerevisiae SRB gene products (initially isolated as suppressors of mutations in the RNAPII CTD). SRBI 0 and SRBI I. are shown to constitute a CDK-cyclin pair, SRBI 1 being structurally relatei to mammalian cyclin C. SRBI 0-SRBI 1 is associated with the yeast RNAPII holoenzyme, and is implicated in both positive and negative regulation of transcription. Genetic and biochemical data further indicate that SRBI 0-SRBI 1 acts on the RNAPII CTD, either directly or indirectly. See also 139.1. 39. .
Tassan JP, Jaquenoud M, LBopold P, Schultz SJ, Nigg EA: Identification of human cyclin-dependent kinase 8, a putative protein klnase partner for cyclin C. froc Nat/ Acad Sci USA 1 QQ5,92:8871-8875. This study identifies CDKB, which is a partner for mammalian cyclin C. Intriguingly, the closest structural relative of CDK8 is SRBI 0, suggesting that the CDKB-cyclin C complex may also play an important role in the regulation of transcription. See also [38**]. Mlkel;i TP, Parvin JD, Kim J, Huber LJ, Sharp PA, Weinberg RA: A kinase-deficient transcription factor IIH is functional in basal and activated transcriotion. Proc Nat/ Acad Sci USA 1 QQ5. 92:5174-5178. . TFIIH containing a catalytically inactive CDK7 is shown to be functional in transcribing the adenovirus major late promoter in an in vitro system. However, this result is difficult to interpret because transcription from this viral promoter does not depend on the CTD. See also I41 **I. 40. .
41.
..
Roy R, Adamczewski JP, Seroz T, Vermeulen W, Tassan JP, Schaeffer L, Nigg EA, Hoejimakers JHJ, Egly JM: The MO15 cell cycle kinase is associated with the TFIIH transcription-DNA repair factor. Cell 1 QQ4, 79:1093-i 101. CDKI (also known as MO151 is shown to be associated with TFllH and to phosphorylate the CTD of RNAPII. This provocative finding provides a potential link between transcription, DNA repair and cell cycle regulation (see also [34** 1 35” I48”I) . 30. ..
Dahmus ME: The role of multisite phosphorylatlon in the regulation of RNA polymerase II activity. Pro9 Nucleic Acid Res Mel Biol 1QQ4, 4&l 43-l 79.
Akoulitchev S. Mlkeil
TP. Weinbera RA. Reinbera D:
Requirement’for TFIIH kinase ac&ity in transcyiption by RNA polymerase II. Nature 1 QQ6, 377:557-560.
This study provides direct evidence that CDK7 activity is required in TFIIH in order to transcribe the dihydrofolate reductase gene in vitro. This result strongly reinforces the notion that the CDK7-associated CTD-kinase activity is important for transcription elongation. See also [40*1. Buck V, Russell R, Millar JBA: Identification of a cdk-activating kinase in fission yeast EMBO J 1QQ5, 14:6173-6183. ; novel CDK, termed Crkl, is identified in S. pombe. Crkl is structurally related to mammalian CDKI and budding yeast KIN28, and presumably allelic to Mcs6. It interacts with the cyclin Mcs2, which in turn is related to mammalian cyclin H and budding yeast CCL1 The gene encoding Crkl 42.
Cyclin-dependent
is an essential gene. The Crkl -Mcs2 complex displays both CAK and CTDkinase activity in vitro. Genetic data strongly indicate an interaction between Crkl -Mcsl and Cdcl-Cdcl3, but the biochemical basis for this interaction has not yet been definitively established. Such biochemical data would ba critical for determining whether the fission yeast Crkl -Mcs2 complex functions as a CAK or CTD kinase or in both roles. See also 143.1. 43. .
Damagnez V, MPkelP TP, Cottarel G: Schizoseccheromyces pombe Mopl-Mcs2 is related to mammalian CAK EM80 J 1995, 14:6164-6172. This paper reports the isolation of a fission yeast CAK and CTD kinase which is related to mammalian CDK7 and budding yeast KIN28 The kinase is identical to Crkl , which is described in 142’1, but in this paper it is referred to as Mop1 Evidence is presented that fission yeast Mop1 and Mcs2 can associate with the heterologous mammalian partners cyclin H and CDK7, respectively. 44.
Molz L, Beach D: Characterisation of the fission yeast mcs2 cyclin and its associated protein kinase activity. EM80 J 1993, 12:1723-l 732.
45.
Simon M, SBraphin B, Faye G: KIN28, a yeast split gene coding for a putative protein kinase homologous to CDC28. EM60 J 1986, 5:2697-2701.
46.
Valay JG, Simon M, Faye G: The Kin28 protein kinase is associated with a mlln in Saccharomyces cerevisiae. J MO/ Biol 1 99gv 234:3071310.
47. .
Valay JG, Simon M, Dubois MF, Bensaude 0, Facca C, Fays G: The KIN28 gene Is required both for RNA polymerase II
kinase
7 Nigg
317
mediated transcription and phosphorylation of the Rpblp CTD. J Mel Biol 1995, 249535-544. This paper shows that transcription by RNAPII is drastically reduced in budding yeasts harboring a temperature-sensitive mutation in the KIN28 gene, and that reduced transcriptional activity correlates with reduced phosphotylation of the RNAPll CTD. These results strongly support the notion that KIN28 acts on RNAPll in ho. Furthermore, genetic interactions are identified betwsen KIN28 and RADB, SIN4, ST11 and CDC37. See also [49-l. 48. l*
Feaver WJ, Svejstrup JQ, Henry NL, Kornberg RD: Relationship of CDK-activating kinase and RNA polymerase II CTD kinase TFIIH/TFIIK. Cell 1994, 79:1103-l 109. The first demonstration that KIN28 interacts with TFIIH in S. cerevisiae. See also [30”]. 49. .
Cismowski MJ, Laff GM, Solomon MJ, Reed SI: KIN28 encodes a C-terminal domain klnase that controls mRNA transcrlption In Seccheromyces ceretisiee but lacks cyclln-dependent kinase-activating kinase (CAK) activity. MO/ Cell Biol 1995, 15:2983-2992. KIN28 is shown to phosphotylate the RNAPII CTD but not a mammalian CDK in vitro. Furthermore, mutations in KIN28 affect transcription but not the phosphorylation status of CDC28. These data support a role of KIN28 in transcription rather than cell cycle regulation. Budding yeast do, however, possess CAK activity, although the relevant protein has not yet been identified. SW also [47*1. 50.
Mdz L, Booher R, Young P, Beach D: cdc2 and the regulation of mitosis: six interacting mcs genes. Genetics 1989, 122~773-702.
Cyclin-dependent kinase 7: at the cross-roads of transcription, DNA repair and cell cycle control? Erich A Nigg Cyclin-dependent
kinase (CDK)
7 was originally implicated
in cell cycle control by virtue of its ability to phosphorylate and activate other CDKs. Subsequently,
both CDK7
its partner, cyclin H, were found to be associated general transcription for CDK7
factor TFIIH, suggesting
in transcription
and RNA polymerase
additional
roles
and DNA repair. During the past
year, a third subunit associated been characterized,
and
with the
with CDK7
and
cyclin H has
and the functional link between
CDK7
in although a low level of activity can be detected unphosphorylated CDKZ-cyclin A complexes [3’,7,8]. In the case of CDKl, phosphorylation at Thr161 appears to coincide with the onset of cyclin binding, and, conversely, dephosphorylation occurs concomitantly with cyclin destruction upon exit from mitosis. One phosphatase able to dephosphorylate the T-loop residue in monomeric CDKs has recently been identified [9’].
II has been strengthened.
Address Department of Molecular Biology, University of Geneva, Sciences II, 30, quai Ernest-Ansermet, Cl-i-1 211 Geneva 4, Switzerland; e-mail: niggasc2a.unige.ch Current Opinion in Call Biology 1996, 8:312-317 0 Current Biology Ltd ISSN 0955-0674 Abbreviations cyclin dependent kinase activating kinase CAK cyclin-dependent kinase CDK carboxy-terminal domain CTD RNAPII RNA polymerase II suppressor of RNA polymerase B SRB transcription factor IIH TFIIH mitogen-activated protein kinase MAPK ‘m&age-a-trois’ 1 MAT1
Introduction Cyclin-dependent kinases (CDKs) are best known for their fundamental roles in cell cycle regulation. However, some members of the CDK family are involved in other cellular processes, notably in transcription. In vertebrates, cell cycle functions are firmly established for CDKl (the homolog of Schizosaccharotnycespombe Cdc2 and Saccharomycescemisiae CDCZS), CDKZ, CDK4 and CDK6, and such functions are likely for CDK3, whilst the jury is still out for CDKS, CDK7, and CDK8 [1,2,3*,4*]. The particular case of CDK7 constitutes the focus of this review. This nuclear kinase was originally identified as the catalytic subunit of a complex that is able to phosphorylate and thereby activate several cell cycle regulatory CDKs. However, recent findings suggest that the physiological role of CDK7 may be broader than originally surmised, or perhaps different altogether.
T-loop phosphorylation:
a common theme
In order to be active, most CDKs require not only a cyclin partner but also phosphorylation at one particular site, which corresponds to Thr161 in human CDKl, and which is located within the so-called T-loop of kinase subdomain VIII [3*,51,6~~].CDKl, CDK2 and CDK4 all require T-loop phosphorylation for maximum activity,
T-loop phosphorylation is important not only for the activation of CDKs but also for the activation of many other protein kinases [lo’]. From the analysis of several crystal structures it appears that this modification favours a kinase conformation which allows the access of substrate to the active site [11,12]. As exemplified by CAMP-dependent protein kinase, T-loop phosphorylation can be constitutive and serve a primarily structural role. In other cases, it constitutes an important regulatory switch, as illustrated best by the members of the mitogen-activated protein kinase (MAPK) family. Hence the identification of kinases (and phosphatases) acting on T-loops constitutes a subject of considerable general interest.
CDK7: a kinase in a ‘mbnage+trois’ In 1993, one enzyme capable of phosphorylating Thr161 in CDKl (and the corresponding residue within the T-loops of other CDKs) was purified from both vertebrate and invertebrate species [13-151. This enzyme was termed the CDK-activating kinase (CAK). Sequence analyses led to the surprising conclusion that the catalytic subunit of CAK was itself structurally related to CDKs, and had already been cloned under the designation MO15 [16]. Subsequently, MO15 was shown to bind a bonajde cyclin partner, cyclin H, and hence was renamed CDK7 [17,18]. At that point, the available evidence suggested that the CDK7-cyclin H complex might function as an upstream element in a cascade of CDK-cyclins. Somewhat surprisingly, however, CDK7’s associated CAK activity was found to be essentially constant throughout the cell cycle, except for a somewhat reduced activity in quiescent cells [19-211. CDK7 interacts not only with cyclin H, but also, in near-stoichiometric amounts, with a second polypeptide [17,21]. During the past year, this 36-37 kDa protein has been cloned from human [22**], mouse [23**], Xetzopusand starfish [24@*]cDNA libraries. Its structure predicts that it contains an amino-terminally located putative zinc-binding domain conforming to a typical C3HC4 (single-letter code for amino acids) RING-finger motif. The precise functions of RING-finger motifs remain unknown, but they appear more likely to mediate protein-protein (or
Cyclin-dependent
protein-RNA) interactions than specific DNA binding [ZS]. In r&u reconstitution experiments indicate that the CDK7-associated 36 kDa RING-finger protein can function as an assembly factor, promoting a stable interaction between CDK7 and cyclin H [22**-24**]. Accordingly, the protein was named MAT1 (for ‘menage-a-trois’ 1 [Z?]). The RING-finger domain of MAT1 is not required for this ternary-complex formation [Z?], suggesting that it is available for mediating interactions with other, as yet unidentified, proteins. In &JO, MAT1 appears to form a trimeric complex with CDK7 and cyclin H under most circumstances [22”,26”], but CDK7-cyclin H heterodimers have also been observed [23~~,24*~,26**]. Presently, there is no evidence that MAT1 could act as an inhibitor of the catalytic activity of CDK7, but it would be premature to exclude the possibility that, in &JO, MAT1 may modulate CDK7’s activity, subcellular localization and/or substrate specificity. MAT1 appears to represent an assembly factor which is specific for CDK7 and cyclin H, as it has not yet been observed with other CDK-cyclin complexes. However, functionally related assembly factors may exist for other CDK-cyclin complexes [27,28*]. The T-loop of CDK7 itself contains a site (Thr170 in human CDK7) which is phosphorylated in z&o [20,29]. Whereas the interaction between CDK7 and cyclin H in a yeast two-hybrid system seems to depend on the phosphorylation of Thr170 [18], the MATl-mediated stabilization of the CDK7-cyclin H complex is independent of the phosphorylation state of Thr170 [23”,24**]. These results indicate that MAT1 is able to bypass the requirement for T-loop phosphorylation, and that CDK7-cyclin H complexes can form by two distinct mechanisms [23”]. The relative contribution of the two mechanisms in vtio is unknown, but it is attractive to speculate that proteins functionally resembling MAT1 may provide one solution to a fundamental problem: in spite of what Babuschka dolls would seem to suggest, kinase cascades operating through T-loop phosphorylation cannot extend indefinitely; some protein is needed to begin the cascade.
CDK7 is present in transcription
factor IIH
An entirely new perspective on CDK7 function was opened when CDK7 was identified as a subunit of transcription factor IIH (TFIIH) and shown to phosphorylate the carboxy-terminal domain (CTD) of RNA polymerase II (RNAPII) [30**]. TFIIH is a multiprotein complex required not only for class II transcription but also for nucleotide-excision repair [31*,32*]. Its associated CTD-kinase activity is considered important for the promoter-clearance step of transcription, but the precise structural consequences of the phosphorylation of the CTD remain the subject of some debate [31*,32’,33]. Cyclin H and MAT1 are also present in TFIIH [23**,26**,34**,35**], and it is not known what, if anything, distinguishes the TFIIH-associated form of CDK7 from the quantitatively predominant free form. Taking advantage of the intimate association of CDK7
kinase 7 Nigg
313
with TFIIH, anti-CDK7 antibodies were recently shown to constitute a powerful tool for the rapid immuno-isolation of a transcriptionally active RNAPII holoenzyme from a mammalian source [36”]. Whether CDK7 really displays dual-substrate specificity in &JO remains to be further explored, but there is no question that the CDK7-cyclin H-MAT1 complex is able to phosphorylate both the T-loop of CDKs and the YSPTSPS (single-letter code for amino acids) repeats of the RNAPII CTD in vitro. At the level of their primary structures these substrates bear no obvious similarities, raising the question of how they both can fit into the active site of the CDK7-cyclin H-MAT1 complex. Previous studies have identified several potential CTD kinases, including CDKl-cyclin B, mammalian MAPKs, and the S. cemvisiae CDK-cyclin complex CTKl-CTKZ [33,37’]. Furthermore, genetic and biochemical data implicate another budding yeast CDK-cyclin pair, SRBlO-SRBl 1 (where SRB denotes suppressor of RNA polymerase B), in CTD phosphorylation [38”], and the mammalian CDKg-cyclin C complex may perform a similar function [39’]. However, direct evidence for CTD phosphorylation by either SRBlO or CDKS has not yet been reported. Considering the multitude of potential CTD kinases, it remains an important but difficult task to determine the physiological role of each enzyme. The importance of the catalytic activity of TFIIHassociated CDK7 has recently been explored using in vitro transcription systems [40*,41”]. No requirement for CDK7 activity could be observed for the adenovirus major late promoter [40’], but this result is difficult to interpret as transcription from this viral promoter does not depend on the CTD. In contrast, transcription from the CTD-dependent mammalian dihydrofolate reductase promoter did require the presence of an active CDK7 in the added TFIIH preparation [41”]. Although these studies do not rigorously prove that CDK7 phosphorylates the CTD directly, they clearly demonstrate a requirement for CDK7 activity during in vitro transcription, at least for certain promoters. The possibility that the CDK7-cyclin H-MAT1 complex may also function in DNA-repair processes has not yet been examined in depth. However, preliminary data do support a role of CDK7 in nucleotide-excision repair [30”,35**], and recent studies indicate that TFIIHassociated CDK7 activity drops in response to UV irradiation of cells [26”]. This latter observation is particularly intriguing because UV irradiation did not affect the activity of free (i.e. non-TFIIH-associated) CDK7. These observations clearly encourage more detailed studies into the possibility that TFIIH-associated CDK7 may function in signalling pathways that link the recognition of DNA damage to transcriptional responses, DNA repair, and, perhaps, cell cycle control.
Nucleus and gene expression
314
Lessons from yeast?
Figure 1
Taken at face value, the biochemical evidence obtained for the CDK7-cyclin H-MAT1 complex would indicate that CDK7 may act on both the T-loops of multiple CDKs and the CTD of RNAPII (see Fig. la). However, most of the evidence supporting this apparent dual specificity has been obtained from in vitro studies, and it is not clear whether both CAK activity and CTD-kinase activity constitute true physiological functions of this complex. Thus, it may be interesting to briefly consider two alternative scenarios (see Fig. 1b,c). On the one hand, one could argue that CDK7 functions primarily as a CAK, even in its TFIIH-associated form (Fig. lb); specifically, CDK7 might activate other CDK-cyclin complexes (e.g. SRBlO-SRBll or CTKl-CTKZ), and these might in turn phosphorylate the RNAPII CTD. On the other hand, an opposite view would emphasize the RNAPII-related role of CDK7 rather than its cell cycle role (Fig. lc). According to this model, other kinases would accomplish the CAK function in vim.
(a) CAK
, @
D
@
TFllH
9 pRb
$??%I
@
9
0
Lamin
RNAPII CTD
(b)
CAK GB c&k
Table 1 @
@
@ Potential yeasts.’
RNAPII CTD
H-MAT1
complex In
Metazoan organisms
S. cefetisiae
S. pombe
Catalytic subunit
CDK7 (MO1 5)
KIN26
Crkl /Mop1 7
Cyclin subunit
Cyclin H
CCL1
Mcs2
Assembly
MAT1
7
7
Yes
No
Yes
Yes
Yes
0
9
0 pRb
homologs of the CDK7-cyclln
Lamin
(c)
CAK @
subunit
CAK activity CID-klnasa
‘See references [17,16,22**,23**,42*,43’,44-461. to be allelic to Mcs6 [42’,501.
@$$@
9 pRb
9 RNAPII CTD
figure illustrating three extreme ways to think
about the physiological function of the CDK7-cyclin
tCrk1 /Mop1 is likely
Q Lamin
0 1996 Current Op~mon in Cell Biology
A highly schematic
activity Yes
H-MAT1
(CDK7)
complex. (a) The complex may display dual specificity as both a CAK and a CTD kinase (as part of TFIIH), not only in vitro but also in ho. According to this model, CDK7 would act upstream of other CDKs in addition to phosphotylating a non-CDK substrate; specifically, CDK7 is believed to phosphorylate the RNAPII CTD, much as CDK4-cyclin D or CDKl -cyclin B phosphorylates the retinoblastoma gene product (pRb) or nuclear lamins (Lamin), respectively. In contrast, (b) proposes that CDK7 functions primarily as a CAK, both for CDK-cyclin complexes involved in cell cycle control and for those CDK-cyclin complexes that might act as CTD kinases (represented by a shaded circle). Finally, (c) emphasizes the RNAPII-related role of CDK7 and postulates the existence of other kinases with CAK activity (represented by a shaded circle). (For the sake of clarity, only two out of several cell cycle regulatory CDK-cyclin complexes have been depicted downstream of CAK). Open arrows represent activation of one complex by another, or some other form of catalysis. Closed arrows represent the changing role of one complex.
Considering the present state of uncertainty about the function of CDK7, one might hope for illumination from studies in genetically tractable organisms. Potential homologs of CDK7 and cyclin H have indeed been identified in both fission and budding yeasts (see Table 1). In S. pot&, a complex formed between Crkl/Mopl (the putative CDK7 homolog) and Mcs2 (the putative cyclin H homolog) displays both CAK and CTD-kinase activity [42*,43*]. Crkl/Mopl is essential for viability [42*,43’], but the lethal phenotype is heterogeneous and might result from either defects in cell cycle progression, or defects in transcriptional regulation, or defects in both processes. Partial loss of Crkl/Mopl-Mcs2 kinase activity suppresses the mitotic-catastrophe phenotype caused by hyperactivated Cd&-Cdcl3 (which is the mitotic CDK-cyclin complex of S. porn&) [42’,43’,44]. These results clearly demonstrate a genetic interaction between Crkl/Mopl-Mcs2 and Cd&-Cdcl3, but they do not prove that the observed suppression reflects reduced
Cyclin-dependent
phosphorylation of the T-loop of Cd&. One could also argue that reduced phosphorylation of RNAPII might result in a lowered transcription rate of the genes encoding Cd&? and Cdcl3. In S. cerevisiae,the closest structural relatives of CDK7 and cyclin H are KIN28 and CCLl, respectively [45,46,47’]. The KIN28-CCL1 complex associates with TFIIH and readily phosphorylates the RNAPII CTD, but does not seem to display CAK activity [47*,48**,49*]. This might explain why S. cemisiae KIN28 cannot complement mutations in the S. pombe Crkl/Mopl gene, although vertebrate CDK7 can [42*,43’]. Loss of KIN28 function does not induce a uniform cell cycle arrest, but instead leads to a rapid decline in nascent mRNA synthesis [49’], supporting the view that the KIN28-CCL1 complex is a physiologically relevant CTD kinase. On the other hand, budding yeast extracts do contain CAK activity, albeit not associated with KIN28-CCLl [49*]. The purification and characterization of this CAK activity is eagerly awaited. Knowledge of the structure of the budding yeast CAK might help answer many questions concerning mammalian CAK, and -somewhat ironicallymight provide an illustration of the awesome power of yeast biochemistry.
Conclusions With the identification of MATl, a novel mechanism for the assembly of CDK-cyclin complexes has been revealed. Also, the evidence accumulates for a function of the TFIIH-associated CDK7-cyclin H-MAT1 complex in class II transcription and, possibly, nucleotide-excision repair. In contrast, the mechanisms involved in regulating this complex remain mysterious, and the cell cycle functions of the complex are still poorly understood. Answers to many of the open questions may come from a biochemical analysis of S. pombe Crkl/Mopl and a molecular characterization of an S. cerevisiae CAK.
References
and recommended
reading
5.
l
*
Jeffrey PD, Russo AA, Polysk K, Gibbs E, Hurwitz J, Massagub J, Pavletich NP: Mechanism of CDK activation revealed by the structure of a cyclinA-CDK2 complex. Nature 1995, 376:313-320. The crystal structure of a human CDKP-cyclin A complex is determined at 2.3 A resolution. This work provides the basis for a detailed understanding of the regulation and substrate specificity of CDK-cyclin complexes. The reported structure concerns unphosphorylated CDKP; it will be interesting to see how phosphorylation of Thrl60 alters the conformation of the T-loop of CDKP. 7.
Connell-Cowley L, Solomon MJ, Wei N, Harper JW: Phosphorylation independent activation of human cyclindependent kinase 2 by cyclin A in vitro. MO/ Biol Cell 1993, 4:79-92.
8.
Matsuoka M, Kate JY, Fisher RP, Mor an DO, Sherr C: Activation of cyclin-dependent kinase 4 (cdk4 B by mouse M015associated klnase. MO/ Cell Biol 1994, 14:7265-7275.
9. .
Poon RYC, Hunter T: Dephosphoryfation of Cdk2 Thrlss by the cyclin-dependent kinasa-interacting phosphatase KAP in the absence of cyclln. Science 1995, 27Q:QO-93. The phosphatase CDK-associated phosphatass (KAP) was originally identified in yeast two-hybrid screens as a protein which interacts with CDKs. Structurally, it resembles dual-specificity phosphatsses such as Cdc25. As shown here, however, it specifically dephosphorylates Thrl60 (but not Tyrl5) in monomeric Cdkl, at least when assayed in vitro. Binding to cyclin A protects Cdk2 from dephosphorylation. Thus, if KAP acts on CdkP in viva, it is likely to do so only after cyclin degradation. 10. .
Hanks SK, Hunter T: The eukaryotic protein kinase superfamily: kinase (catalytic) domain structure and classification. FASEB J 1995, Q:576-598. A comprehensive review of the primary structures of protein kinsses. Besides containing other useful information, this compilation illustrates that many protein kinasas have phosphorylatable residues in positions corresponding roughly to the T-loop-phosphorylation sites of CDKs. 11.
Taylor SS, Knighton DR, Zheng J, Lynn F, Eyck TF, Sowadski JM: Structural framework for the protein kinase family. Annu Rev Cell Biol 1992, 8:429-462.
12.
Morgan DO, De Bondt HL: Protein klnase regulation: insights from crystal structure analysis. Curr Opin Cell Viol 1994, 6:239-246.
13.
Fesquet D: The MO1 5 gene encodes the catalytic subunit of a protein kinase that activates cdc2 and other cyclin-dependent kinases (CDKs) through phosphorylation of Thr161 and its homologues. EMBO J 1993, 12:3111-3121.
14.
Poon RYC, Ysmashita K, Adamczewski JP, Hunt T, Shuttleworth J: The cdc2-related orotein ~40M015 is the catalvtfc subunit of a protein kinase th& ten activate p33cdk* and 634dC*. EM60 J 1993, 12:3123-3132.
15.
Solomon MJ, Harper WJ, Shuttleworth J: CAK, the p34&* activating klnase, contains a protein identical to or closely related to p40Mols. EM30 J 1993, 12:3133-3142.
16.
Shuttleworth J, Godfrey R, Colman A: p40M015, a cdcl-related protein kinase involved in negative regulation of meiotfc maturation of Xenopus laevis oocytes. EM30 J 1990, 9:3233-3240.
1z
Fisher RP, Morgan DO: A novel cyclin associates with MO1 J/cdk7 to form the CDK-activating kinase. Cell 1994, 781713-724.
18.
Mlkelii TP, Tsssan JP, Nigg EA, Frutiger S, Hughes GJ, Weinberg RA: A cyclin associated with the CDK-activating klnase MO1 5. Nature 1994, 371:254-257.
19.
Brown Al, Jones T, Shuttleworth J: Expression and activity of p40M015, the catalytic subunit of cdk-activating kinase, during Xenopus oogenesis and embryogenesis. MO/ Viol Cell 1994, 5:921-932.
20.
Poon RYC, Ysmashita K, Howell M, Ershler MA, Belyavsky A, Hunt T: Cell cycle regulation of the p34c*2@3mk* activating kinase p4@01s. J Cell Sci 1994,107:2789-2799.
21.
Tssssn JP, Schultz SJ, Bsrtek J, Nigg EA: Cell cycle analysis of the activity, subcellular localisatlon, and subunlt composition of human CAK (CDK-activating kinase). J Cell Biol 1994, 127~467-478.
of special interest of outstanding interest
1.
Van den Heuvel S, Harlow E: Distinct roles for cyclln-dependent kinases in cell cycle control. Science 1993, 262:2050-2054.
2.
Sherr CJ: Gl phase progression: 79:551-555.
cycling on cue. Cell 1994,
Morgan DO: Principles of CDK regulation. Nature 1995, 374:131-134. i thorough review of the principles of CDK regulation.
3.
4. .
Nigg EA: Cyclin-dependent protein kinases: key regulators the eukaryotic cell cycle. Bioessays 1995, 17:471-480.
of
Solomon MJ: The function(s) of CAK, the p34d-activating klnase. Trends Biochem Sci 1994,19:496-500.
6. ..
Papers of particular interest, published within the annual period of review, have been highlighted as: .
315
A recent review of both the regulation and the substrates of CDK-cyclin complexes.
Acknowledgements I would like to thank Jean-Pierre Tassan and Andrew Fry for helpful comments on the manuscript. Due to space limitations, literature citation is not comprehensive and I apologize to all colleagues who received less explicit credit than they deserved. Work in the author’s laboratory was supported by grants from the Swiss National Science Foundation (grant 31-33615.92) and the Swiss Cancer League (grant FOR 447).
kinase 7 Nigg
316
22. ..
Nucleus and gene expression
Tassan JP, JaquenoudM, Fry AM, Frutiger S, Hughes GJ, Nigg EA: In vitro assembly of a functional human CDK7-cyclln H complex requires MATI, a novel 36 kDa RING finger protein. EM60 J 1QQ5,14:5608-561 Z
Translation of CDK7, cyclin H and MAT1 in a reticulocyte lysate yields no stable heterodimeric complexes between any two proteins. However, active ternary complexes can readily be isolated after mixing all three proteins, and complex formation occurs independently of the RING-finger domain of MATI. Ternary complexes phosphorylate both the T-loop in CDK2 and a peptide mimicking the CTD of RNAPII (see also [23”,24”,26”]). Fisher RP, Jin P, Chamberlin HM, Morgan DO: Alternative mechanisms of CAK assembly require an assembly factor or an activatina kinase. Cell 1995. 83:47-57. Using lysates from- baculovirus-infecte; insect cells, the authors show that the formation and activity of recombinant CDK’I-cyclin H complexes are substantially increased by the presence of MATl. Interestingly, the requirement for MAT1 can be bypassed by prior phosphorylation of T-loop residues within CDKZ usina either CDKI-cvclin A or CDKP-cvclin A as T-loop kinasesl These’resulk indicate the existence of alternative-pathways for CDK7-cyclin H complex assembly, and they raise the possibility that CDK-cyclin complexes other than CDK’/-cyclin H-MAT1 may display Tloop-directed kinase activity in viva. See also [22”,24”,26”]. 23. ..
Devault A, Martinez AM, Fesquet D, Labbe JC, Tassan JP, Nigg EA, Cavadore JC, Dor(te M: MATI, a new RING-finger protein subunit StabiliZinQ cyclin H-cdk7 complexes In starfish and Xenopus CAK. EM/30 J lQQ5,14:5027-5036. In vitro assambly of CDK7-cyclin H-MAT1 complexes is studied in a reticulocyte lysate sytem. MAT1 is shown to confer stability to a ternary complex regardless of the phosphorylation state of the CDKP T-loop. Moreover, evidence is presented that, at least in oocytes, dimeric CDK7-cyclin H complexes co-exist with trimeric MAT1 -containing complexes. See also [22” *23” I26”I.
24. ..
25.
Freemont PS: The RING finger -a
novel protein sequence motif related to the zinc finger. Annu NY Acad Sci 1993, 684:174-l
26. ..
92.
Adamczewski JP, Rossignol M, Tassan JP, Nigg EA, Moncollin V, Egly JM: MATI, cdk7 and cyclin H form a kinase complex which is UV light sensitive upon association with TFIIH. EMBO J 1QBB
.___, in nress r.___.
MAT1 is shown to be present in those CDK7-cyclin H complexes that associate with TFlIH. Furthermore, evidence is presented that TFIIH-associated (but not bulk!) CDK7 activity drops in response to UV irradiation of cells. These observations raise the possibility that CDKI might participate in a checkpoint signalling pathway that relays the presence of DNA damage to the transcriptional apparatus. 27.
Kato JY, Matsuoka M, Strom DK, Sherr CJ: Regulation of cyclin D-dependent kinase 4 (cdk4) by cdk4-activating klnase. MO/ Cell Biol 1994, 14~2713-2721.
Gerber MR, Farrell A, Raymond JD, Herskowitz I, Morgan DO: Cdc37 is required for association of the protein klnase Cdc28 with G1 and mitotlc cyclins. Proc Nat/ Acad Sci USA 1995, 92:4651-4655. The Cdc37 gene of S. cerevisiae has previously been shown to be required for passage through the G1 phase of the cell cycle. A strain harboring a mutation in Cdc37o is now reoorted to contain reduced levels of CdcPB-cyclin compl’exes. These ra’sults suggest that Cdc37p is somehow required for the formation of CDK-cyclin complexes in budding yeast. However, the Cdc37 gene product is not stably associated with Cdc28-cyclin complexes, suggesting that its mode of action is different from that of vertebrate MATI. 28. .
29.
Labb(, JC, Martinez AM, Fesquet D, Capony JP, Darbon JM, Derancouri J, Devault A, Morin N, Cavadore JC, Do& M: P40Mels associates with p36 subunit and requires both nuclear translocation and Thr176 phosphorylation to generate cdk-activating activity in Xenopus oocytes. EM80 J 1 QQ4, 13:5155-5164.
A recent review on TFIIH subunits and their involvement in transcription, DNA repair, and cell cycle regulation. See also [32*1. 32. .
Maldonado E, Reinberg D: News on initiation and elongation of transcription by RNA polymerase II. Curr Opin Cell Biol 1 QQ5, 7:352-361. See annotation [31 *I. 33.
34. ..
Serirawa H, M;ikelii TP, Conaway JW, Conaway RC, Weinberg RA, Young RA: Association of cdk-activating kinase subunits with transcription factor TFIIH. Nature 1 QQ5, 374:260-282. This paper shows that not only CDK7 but also cyclin H is a subunit of TRIH, and that this TFIIH-associated kinase phosphorylates both the T-loop of CDKs and the CTD of RNAF‘ll (see also [30”,35**]). 35. ..
Shiekhattar R, Mermelstein F, Fisher R, Drapkin R, Dynlacht B, Wessling HC, Morgan DO, Reinberg D: Cdk-activating kinase complex is a component of human transcription factor TFIIH. Nature 1 QQ5, 374~203-207. See annotations [30” 1 34”l . 36. ..
31. .
Seroz T, Hwang JR, Moncollin V, Egly JM: TFIIH: a link between transcription, DNA repair and cell cycle regulation. Gun Opin Gener Dev 1995, 5:217-221.
Ossipow V, Tassan JP, Nigg EA, Schibler U: A mammalian RNA polymerase II holoenzyme containing all components
required for promoter-specific transcription initiation. Cell 1QQ5, 63:137-l 46. This study shows that a monoclonal antiCDK7 antibody can be used for the rapid purification of a RNAPII holoenzyme from a highly concentrated rat liver nuclear extract. The antiCDK7 immunoprecipitate contains all components required for transcription initiation at both viral and cellular promoters. This is the first description of a mammalian RNAPII holoenzyme. 37.
.
Sterner DE. Lee JM. Hardin SE. Greenleaf AL: The veast carboxyl-t&mlnal r&eat dom& kinase CTDK-i Is a divergent cyclin-cyclin dependent klnate complex. MO/ Cell Biol 1 QQ5,
15:5716-5724. The budding yeast CTD kinase 1 is shown to contain a CDK-cyclin pair (CTKI-CTK2). in addition to a third subunit (CTK3). CTK2 is structurally related to mammalian cyclin C. CTK3 shares no obvious sequence similariiy with MATl, but it is noteworthy that two distinct CDKs with CTD-kinase activity (CDK7 in metazoans and CTKI in S. cerevisiae) both exist in heterotrimeric complexes. 38. ..
Liao SM, Zhang J, Jeffery DA, Koleske AJ, Thompson CM, Chao DM, Viljoen M, Van Vuuren JJ, Young RA: A kinase-cyclln pair in the RNA polymerase II holoenzyme. Nature lQQ5, 374:193-l 96. Two S. cerevisiae SRB gene products (initially isolated as suppressors of mutations in the RNAPII CTD). SRBI 0 and SRBI I. are shown to constitute a CDK-cyclin pair, SRBI 1 being structurally relatei to mammalian cyclin C. SRBI 0-SRBI 1 is associated with the yeast RNAPII holoenzyme, and is implicated in both positive and negative regulation of transcription. Genetic and biochemical data further indicate that SRBI 0-SRBI 1 acts on the RNAPII CTD, either directly or indirectly. See also 139.1. 39. .
Tassan JP, Jaquenoud M, LBopold P, Schultz SJ, Nigg EA: Identification of human cyclin-dependent kinase 8, a putative protein klnase partner for cyclin C. froc Nat/ Acad Sci USA 1 QQ5,92:8871-8875. This study identifies CDKB, which is a partner for mammalian cyclin C. Intriguingly, the closest structural relative of CDK8 is SRBI 0, suggesting that the CDKB-cyclin C complex may also play an important role in the regulation of transcription. See also [38**]. Mlkel;i TP, Parvin JD, Kim J, Huber LJ, Sharp PA, Weinberg RA: A kinase-deficient transcription factor IIH is functional in basal and activated transcriotion. Proc Nat/ Acad Sci USA 1 QQ5. 92:5174-5178. . TFIIH containing a catalytically inactive CDK7 is shown to be functional in transcribing the adenovirus major late promoter in an in vitro system. However, this result is difficult to interpret because transcription from this viral promoter does not depend on the CTD. See also I41 **I. 40. .
41.
..
Roy R, Adamczewski JP, Seroz T, Vermeulen W, Tassan JP, Schaeffer L, Nigg EA, Hoejimakers JHJ, Egly JM: The MO15 cell cycle kinase is associated with the TFIIH transcription-DNA repair factor. Cell 1 QQ4, 79:1093-i 101. CDKI (also known as MO151 is shown to be associated with TFllH and to phosphorylate the CTD of RNAPII. This provocative finding provides a potential link between transcription, DNA repair and cell cycle regulation (see also [34** 1 35” I48”I) . 30. ..
Dahmus ME: The role of multisite phosphorylatlon in the regulation of RNA polymerase II activity. Pro9 Nucleic Acid Res Mel Biol 1QQ4, 4&l 43-l 79.
Akoulitchev S. Mlkeil
TP. Weinbera RA. Reinbera D:
Requirement’for TFIIH kinase ac&ity in transcyiption by RNA polymerase II. Nature 1 QQ6, 377:557-560.
This study provides direct evidence that CDK7 activity is required in TFIIH in order to transcribe the dihydrofolate reductase gene in vitro. This result strongly reinforces the notion that the CDK7-associated CTD-kinase activity is important for transcription elongation. See also [40*1. Buck V, Russell R, Millar JBA: Identification of a cdk-activating kinase in fission yeast EMBO J 1QQ5, 14:6173-6183. ; novel CDK, termed Crkl, is identified in S. pombe. Crkl is structurally related to mammalian CDKI and budding yeast KIN28, and presumably allelic to Mcs6. It interacts with the cyclin Mcs2, which in turn is related to mammalian cyclin H and budding yeast CCL1 The gene encoding Crkl 42.
Cyclin-dependent
is an essential gene. The Crkl -Mcs2 complex displays both CAK and CTDkinase activity in vitro. Genetic data strongly indicate an interaction between Crkl -Mcsl and Cdcl-Cdcl3, but the biochemical basis for this interaction has not yet been definitively established. Such biochemical data would ba critical for determining whether the fission yeast Crkl -Mcs2 complex functions as a CAK or CTD kinase or in both roles. See also 143.1. 43. .
Damagnez V, MPkelP TP, Cottarel G: Schizoseccheromyces pombe Mopl-Mcs2 is related to mammalian CAK EM80 J 1995, 14:6164-6172. This paper reports the isolation of a fission yeast CAK and CTD kinase which is related to mammalian CDK7 and budding yeast KIN28 The kinase is identical to Crkl , which is described in 142’1, but in this paper it is referred to as Mop1 Evidence is presented that fission yeast Mop1 and Mcs2 can associate with the heterologous mammalian partners cyclin H and CDK7, respectively. 44.
Molz L, Beach D: Characterisation of the fission yeast mcs2 cyclin and its associated protein kinase activity. EM80 J 1993, 12:1723-l 732.
45.
Simon M, SBraphin B, Faye G: KIN28, a yeast split gene coding for a putative protein kinase homologous to CDC28. EM60 J 1986, 5:2697-2701.
46.
Valay JG, Simon M, Faye G: The Kin28 protein kinase is associated with a mlln in Saccharomyces cerevisiae. J MO/ Biol 1 99gv 234:3071310.
47. .
Valay JG, Simon M, Dubois MF, Bensaude 0, Facca C, Fays G: The KIN28 gene Is required both for RNA polymerase II
kinase
7 Nigg
317
mediated transcription and phosphorylation of the Rpblp CTD. J Mel Biol 1995, 249535-544. This paper shows that transcription by RNAPII is drastically reduced in budding yeasts harboring a temperature-sensitive mutation in the KIN28 gene, and that reduced transcriptional activity correlates with reduced phosphotylation of the RNAPll CTD. These results strongly support the notion that KIN28 acts on RNAPll in ho. Furthermore, genetic interactions are identified betwsen KIN28 and RADB, SIN4, ST11 and CDC37. See also [49-l. 48. l*
Feaver WJ, Svejstrup JQ, Henry NL, Kornberg RD: Relationship of CDK-activating kinase and RNA polymerase II CTD kinase TFIIH/TFIIK. Cell 1994, 79:1103-l 109. The first demonstration that KIN28 interacts with TFIIH in S. cerevisiae. See also [30”]. 49. .
Cismowski MJ, Laff GM, Solomon MJ, Reed SI: KIN28 encodes a C-terminal domain klnase that controls mRNA transcrlption In Seccheromyces ceretisiee but lacks cyclln-dependent kinase-activating kinase (CAK) activity. MO/ Cell Biol 1995, 15:2983-2992. KIN28 is shown to phosphotylate the RNAPII CTD but not a mammalian CDK in vitro. Furthermore, mutations in KIN28 affect transcription but not the phosphorylation status of CDC28. These data support a role of KIN28 in transcription rather than cell cycle regulation. Budding yeast do, however, possess CAK activity, although the relevant protein has not yet been identified. SW also [47*1. 50.
Mdz L, Booher R, Young P, Beach D: cdc2 and the regulation of mitosis: six interacting mcs genes. Genetics 1989, 122~773-702.