|
Laboratory of Molecular Biophysics
|
Next: Regulation of Phosphorylase Kinase, Up: Protein Kinases, Return to: Contents.
In response to mitogenic signals, cells progress from the resting phase, G0, to G1 during which they become committed to progression through the cell cycle. From G1 they enter S phase when the chromosomes are duplicated once and only once. After a second gap or growth phase, G2, they enter mitosis when the cell divides into two daughter cells. A member of the family of the cyclin dependent protein kinases, CDKs, initiates each of the transitions. In response to the interaction of mitogenic factors with their receptors at the outside of the cell, signalling to the nucleus results in expression of the D-type cyclins which associate with and activate the kinases CDK4 and CDK6. These CDK complexes phosphorylate the tumour suppressor protein pRb and, through mechanism not completely understood, promote the expression of cyclin E. Cyclin E associates with CDK2 to drive cells from G1 to S phase through phosphorylation of a limited number of targets including pRb. On entry into S phase, cyclin E is abruptly destroyed by the proteasome to which it is targeted by ubiquitination. Cyclin A, expressed in response to the CDK2/cyclin E activities, then associates with CDK2 to drive cells through S phase. CDK2/cyclin A phosphorylates a large number of target proteins including pRb, transcription factors, regulators of transcription factors and pre-replication complexes. Cyclin A remains present throughout G2 and associates with CDK1 during the transition from G2 to M phase after which it is abruptly degraded. Cyclin B then associates with CDK1 to drive cells through M phase in concert with other kinases and down stream regulator enzymes.
The CDKs are tightly regulated (reviewed in Morgan 1997; Shapiro and Harper 1999). In addition to the dependence on the association with cyclin, CDKs require activatory phosphorylation (on Thr160 in CDK2) by a cyclin dependent activatory kinase (CAK or CDK7/cyclin H) in the region of the kinase termed the activation segment. CDK activity may be inhibited by two distinct mechanisms, response to cellular protein inhibitors of the Ink4 and Cip1/Kip1 families and phosphorylation on the glycine rich loop by Wee1 and Myt1 kinases. The inhibitory phosphorylations are relieved by the action of the phosphatase Cdc25C, which in turn is subject to regulation and provides an important checkpoint control, especially in response to DNA damage.
N. K. Brown, E. Dubinina, in collaboration with J. A. Endicott and M. E. M. Noble
In the last decade, the structural basis of CDK2 activation has become well
understood (Pavletich 1999). The combination of association with cyclin A and
phosphorylation on Thr160 in CDK2 leads to the correct orientation of those
residues that bind ATP and the correct orientation of the activation segment
for protein substrate binding, as shown by our work on a phospho-CDK2/cyclin A
peptide substrate complex (Brown et al. 1999). The recognition features of
the CDK2 catalytic site ensure that the phosphorylatable residue of the
substrate (serine or threonine) is directed to the 
phosphate of ATP for
phosphoryl transfer. In recent years it has been recognised that almost all
CDK2 physiologically important substrates, (such as the tumour suppressor
protein pRb, the related protein p107, or CDC6 a component of the
pre-replication complex) have an additional site of recognition that contains
the sequence motif RXL (using the single letter code). This motif is usually
C-terminal to the site of phosphorylation and may be as close as 20 residues
or as distant as over 100 residues (Table 1).
Mutation of the RXL site
prevents phosphorylation of the substrate. Evidence suggests that the site
makes a poor substrate a good substrate.
| Substrate | Phosphorylation site | Recruitment site |
| pRb | S807PLK, S811PYK + others | LK873KLRFD |
| p107 | S640PIS + others | KR658TLFGE |
| CDC6 | S54 PRK, S74PPK | GR94 RLVFD |
| p53 | S315PQP | HK381KLMFK |
| p27 | T187PKK | CR30 NLFGP |
| E2F | S332PPP, S337PPS + DP1 | KR90 RLDLE |
|
|
We have an ongoing programme to determine the role of the 'recruitment site' in CDK2 specificity. Our original work (Brown et al. 1999) showed that the RXL motif from a p107 peptide bound to the cyclin A at an exposed site, identified previously as a conserved exposed hydrophobic site on cyclins A, B, D and E (Brown et al. 1995). The site is some 40Å from the catalytic site as the crow flies but about 53Å if the chain has to wrap around the cyclin (Figure 1a). We are currently exploring whether there is a direct route of communication from the catalytic site to the recruitment site through the use of peptides derived from p107 and CDC6 that contain both the canonical SPXK sequence for the phosphorylatable serine and the RXL motif down stream. We are also exploring co-crystallisation studies of pCDK2/cyclin A with intact substrates and the possibility that the CDK2/cyclin A complex has a third site of recognition for substrates.
One of the questions that intrigues us is how CDK2/cyclin E and CDK2/cyclin A substrate specificity is achieved. What special properties do the two cyclins confer on the CDK2 that are so important for orderly cell cycle progression? In general CDK2/cyclin A has a broader specificity for in vivo substrates than CDK2/cyclin E. There are a few substrates that are phosphorylated better by CDK2/cyclin E, (e.g., the p220 protein NPAT that is involved in histone gene expression) but many more for which CDK2/cyclin A is a better catalyst. For this end we have revitalised our programme to express cyclin E in baculoviral infected insect cells. We shall use the effective polycistronic E. coli expression system for preparation of GST-phospho-CDK2 fusion to purify the cyclin E for crystallisation studies.
Table 1 shows that recruitment motifs tend to fall into two classes. One in which the RXL motif is followed by a hydrophobic residue (as in p107, CDC6, p53, p27) and the other in which it is followed by a charged residue (pRb, E2F). The structure of the p107 peptide bound to pCDK2/cyclin A showed that the hydrophobic phenylalanine made important contributions to binding (Figure 1b). What happens when this residue is charged as in E2F? We are working in collaboration with Steve Gamblin, Ivo Tews and Sheraz at the National Institute for Medical Research, Mill Hill to provide comparative structural and isothermal calorimetry data on the relative affinities of peptides bound at the cyclin site.
Erika de Moliner and Nick Brown
Components of the cell cycle machinery are frequently altered in cancer. Deregulated CDK activity provides cells with selective growth advantages, which combined with aberrant checkpoint control at the G1/S and G2/M boundaries, leads to undesirable cell proliferation. The CDKs offer multiple mechanisms for intervention in the transformed state. In S phase CDK2/cyclin A is important for phosphorylation and inactivation of the E2F/DP1 transcription factor. Inhibition of CDK2/cyclin A results in elevated E2F concentrations leading to S phase arrest and apoptosis. An extensive programme on structure based drug design of CDK2 inhibitors is in progress in collaboration with colleagues at the University of Newcastle and AstraZeneca (see section 8.2). Here we describe recent structural results on natural product inhibitors.
Staurosporine is a microbial alkaloid that attracted considerable attention since it was first characterised in 1986 and shown to be a potent inhibitor of protein kinase C (PKC). Staurosporine proved too toxic for use as a therapeutic agent but UCN-01 (7-hydroxy staurosporine), isolated from Streptomyces and reported in 1987, is a more selective protein kinase inhibitor. UCN-01 has antiproliferative effects and cytostatic properties in several human tumour cell lines. It has given favourable pharmacokinetic and toxicology properties and can be tolerated at doses equivalent to plasma concentrations of 35-50 µM (Senderowicz and Sausville 2000; Sausville et al. 2001). UCN-01 is currently undergoing clinical trials for cancer treatment in the US and in Japan.
In addition to PKC inhibition, UCN-01 inhibits other protein kinases in the cell, including those involved in regulating antiapoptotic signals in non-proliferating cells. In vitro, UCN-01 is effective in cell cycle regulation. It inhibits CDK2, CDK4 and CDK6 assayed against pRb as substrate with IC50 values of 42, 32 and 58 nM, respectively (Kawakami et al. 1996). Paradoxically, in intact cells UCN-01 has been shown to activate CDK1 and CDK2. UCN-01 also potently abrogates control of the G2 checkpoint in cancer cells that are characterised by disrupted p53 function (Wang et al. 1996). Further, UCN-01 potentiates the cytotoxicity of a variety of DNA damage anti-cancer drugs including cisplatin, camptothecin, mitomycin C and ionising radiation. An explanation for the activation of CDK1 in cells was provided by the observation that UCN-01 is a potent inhibitor of Chk1 kinase against its substrate the phosphatase, Cdc25C (Graves et al. 2000). In response to DNA damage, normal cells arrest in G2 through the checkpoint that controls activation of CDK1 by dephosphorylation of inhibitory phosphorylation sites Thr14 and Tyr15 by Cdc25C. Through the inhibition of Chk1, UCN-01 prevents phosphorylation of the protein phosphatase, Cdc25C, on Ser216, a phosphorylation that is necessary to inhibit Cdc25C. The active phosphatase dephosphorylates CDK1 at pThr14 and pTyr15 thereby activating CDK1. UCN-01 sensitises cancer cells to DNA damaging reagents through abrogation of this cell cycle checkpoint, allowing premature activation of CDK1, which promotes mitosis in DNA damaged cells and their subsequent apoptosis. Thus, although at high concentrations UCN-01 can directly inhibit CDKs, at lower concentrations it can modulate cellular upstream regulators of the cell cycle. Because of its effects on DNA damaged cells, UCN-01 is a potential candidate for combination strategies in cancer treatment.
|
In order to elucidate the structural basis of kinase inhibition, we have determined the crystal structure of pCDK2/cyclin A co-crystallised with UCN-01 at 2.3Å resolution (De Moliner et al in preparation). We are grateful to Dr R. J. Schultz at the Drug Synthesis and Chemistry Branch, Developmental Therapeutics Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute for the gift of UCN-01. The inhibitor UCN-01 bound to the fully active pCDK2/cyclinA binary complex at the ATP binding site pocket occupying the same region of ATP adenine and ribose rings (Figure 2). The specific hydrogen bond pattern of ATP in the hinge region of CDK2 is conserved in the UCN-01 complex: UCN-01 O5 accepts a hydrogen bond from Leu83 N and UCN-01 N1 donates a hydrogen bond to Glu81 O. The binding of UCN-01 to CDK2, as with staurosporine (Lawrie, A.M. et al. 1997), is characterised by complementarity of shape between UCN-01 and CDK2 and contacts with a number of hydrophobic side chains. The 7-hydroxy group points into a pocket that is lined by the hydrophobic side chains of Val64, Phe80, Leu134 and Ala144. A polar component to this pocket achieved by a water molecule incorporated at a distance of 3.5Å from the 7-hydroxyl. The water in turn is hydrogen bonded to Asp145 N and a second water and forms a network linking the 7-hydroxyl group to other polar groups.
The addition of the 7-hydroxyl to staurosporine results in a compound that is of therapeutic value. What difference does this extra hydroxyl group make? The exploitation of UCN-01 as cancer drug relies on its reduced toxic effects compared with staurosporine. UCN-01 can be tolerated at µM levels while staurosporine is toxic at 10 nM levels. Part of this reduced toxicity may be due to the ability of UCN-01 to be a more selective kinase inhibitor. We have examined the sequences and structures of other kinases in order to search for an explanation of the greater selectivity of UCN-01. Sequence alignment of several of the key protein kinases indicate that the residues that contact the O7 hydroxyl are mostly conserved in many kinases. For Chk1 a crystal structure is available in the active conformation (Chen et al. 2000). Superposition of the Chk1 structure onto the pCDK2/cyclin A/UCN-01 structure showed conservation of the hydrophobic character of the contact residues and the polar residues at the ATP binding site but there are differences in the contacts to the 7-hydroxyl. In Chk1 Phe80 of CDK2 is a leucine, which could allow more flexibility in its contact to one of the indole cabazole rings of UCN-01. More importantly, in Chk1 Ala144 of CDK2 is a serine. The structure of Chk1 shows that the serine is in the right position to hydrogen bond to O7, thus providing a possible rationalisation for the high affinity of Chk1 for UCN-01 (IC50 estimated between 8-25 nM).
Erika de Moliner and Nick Brown.
The compound 4,5,6,7 tetrabromo 2-azabenzimidazole (TBB) is an ATP-site directed kinase inhibitor that is selective for casein kinase 2 (CK2), CDK2 and phosphorylase kinase (PhK). TBB, a gift from Professor L. A. Pinna, University of Padua, was co- crystallised with phospho-CDK2/cyclin A and the structure determined at 2.3Å resolution. Comparison of the mode of binding to CDK2 with that already observed by the Padua group (Battistutta et al. Protein Sci. (2001) in press) with CK2 revealed a surprise. Despite conservation of the ATP binding sites, TBB binds to the two kinases in different ways. In CK2 the bulky bromines are turned out of the catalytic site but in CDK2 the bromines are directed in towards the hinge region of the kinase. Contacts involve a halogen/carbonyl oxygen interaction, which is closer than a van der Waals interaction. Such contacts have been observed in small molecule crystal structures of organic bromine compounds. The origins of the different binding modes and the selectivity of TBB are being further explored.
Atlanta Cook and Ed Lowe.
The catalytic machinery at the active sites of protein kinases is highly
conserved. Our previous work with phosphorylase kinase (Lowe et al. 1997;
Skamnaki et al. 1999) had shown the importance of a key catalytic aspartate
for promotion of catalysis. These studies suggested that once both substrates
are bound in the correct orientation the ATP is poised to promote
phospho-transfer to the substrate serine hydroxyl by a direct in-line attack
of the lone pair electrons of the serine on the phosphorus atom of the ATP.
The aspartate acts as a proton acceptor from the serine in the reaction cycle
and a lysine residue together with a Mg2+ are involved in transition state
stabilisation. In order to test if further groups might participate in the
reaction cycle, especially in stabilisation of the transition state, we have
co-crystallised pCDK2/cyclin A with ADP, a peptide substrate and soaked with
NO3. The trigonal NO3 ion mimics the phosphorane intermediate in the
phosphoryl transfer reaction. The results of structure determination of this
complex at 2.7Å resolution showed the nitrate ion bound between the
ß-phosphate of ADP and the serine OH of the peptide substrate in just the
expected geometry for phosphoryl transfer. There was little conformational
change in the enzyme and no further side chains moved into the catalytic site
to stabilise the complex. We are currently working in collaboration with
Professor N. G. Oikonomakos to establish the kinetic properties of the
NO3 ion as an inhibitor.
Nick Brown.
XDRP1 (Xenopus DSK-related protein 1, 62 kDa) was identified in a two-hybrid screen using the N-terminus of cyclin A1 as bait. Its overexpression inhibits degradation of the cyclin in frog egg extracts. Mammalian proteins showing high sequence similarity with XDRP1 have also been cloned and they all contain an N-terminal ubiquitin-like domain (UBL) and a C-terminal ubiquitin-associated domain (UBA). The intervening sequences contain several repeats bearing the motif MXNPD. Thus they are all members of the ubiquitin domain protein (UDP) family of proteins.
Recent evidence has indicated that for at least some of these family members, the UBL domain interacts with subunit(s) of the proteasome while the UBA domain may bind (poly)ubiquitin and this goes some way towards explaining why their overexpression affects the degradation of substrates via the ubiquitin mediated proteasomal pathway. We have clones for XDRP1 (gift of H. Kobayashi) and a human homologue (gift of R. Poon) in the laboratory and have successfully expressed the human protein as a GST fusion bearing the 3C protease cleavage site. However, the cleaved and purified protein has proved difficult to concentrate and initial crystallisation screens gave negative results. Partial proteolysis combined with N-terminal protein sequencing was carried out in an attempt to discover stable fragments more amenable to structural analysis. Trypsin digestion of the human protein produced transient 45 and 30 kDa fragments which were rapidly broken down to more stable products with masses of 18, 16 and 14 kDa as judged by SDSPAGE. Sequence analysis of these smaller fragments (each of which gave a major and a minor signal) are summarised in the figure below. Based on these results, PCR will be used to engineer the UBL and UBA domains, as well as larger constructs encompassing the fragments obtained by trypsin digestion enabling functional and structural analyses.
Next: Regulation of Phosphorylase Kinase, Up: Protein Kinases, Return to: Contents.