The activities of cyclin-dependent kinases (CDKs), regulated primarily by the periodic expression of their cyclin binding partners, temporally order sequential cell cycle transitions through G1, S phase, G2, and mitosis. In mammalian cells, regulators of G1 transit include three D-type cyclins, as well as CDK4 and CDK6, and the CDK inhibitory proteins, p21 and p27 (1). In response to mitogens, individual D-type cyclins assemble with CDK4 or CDK6 and, paradoxically, with the p21 or p27 “inhibitors” to yield active higher-order holoenzymes that drive G1 progression and prime cells to enter S phase and begin DNA replication (2). On page 1330 of this issue, Guiley et al. (3) report the crystal structures of trimeric complexes containing cyclin D1, CDK4, and either p27 or p21 and reveal that active trimers containing tyrosine-phosphorylated p27 are surprisingly refractory to the U. S. Food and Drug Administration (FDA)–approved CDK4/6 inhibitors that are used to treat hormone-dependent breast cancer.
Newly synthesized CDK4, folded by the HSP90-chaperone system, can be sequestered by palbociclib or by p16, which increases unsequestered p27 and p21 and inhibits CDK2 complexes to block the G1/S transition. Extraction of CDK4 by p27 or p21 and cyclin D1 yields an inactive trimer capable of undergoing CAK-mediated Thr172 phosphorylation on its reconfigured activation segment. Activation of the trimer requires phosphorylation of tyrosine residues on p27.
GRAPHIC: A. KITTERMAN/SCIENCE
Cyclin D holoenzymes preferentially phosphorylate the tumor suppressor retinoblastoma (RB) protein, and a few other substrates, to cancel the antiproliferative activity of RB. By contrast, cyclin E–CDK2 (which acts at the G1/S transition) and cyclin A– and B–driven CDK2 and CDK1 complexes sequentially phosphorylate many hundreds of substrates (including RB) during the remainder of the cycle. Through their simultaneous binding to both cyclin and CDK subunits, p21 and p27 can inhibit CDKs 1, 2, 4, and 6 to enforce cell cycle arrest in response to mitogen withdrawal, antiproliferative cytokines, cell contact inhibition, or cellular stress. Unexpectedly, all cyclin D1–CDK4-dependent kinase activity toward RB requires higher-order complexes containing p27 or p21, which are required for holoenzyme assembly (2). Phosphorylation of bound p27, but not p21, by nonreceptor tyrosine kinases converts the inactive cyclin D1–CDK4-p27 complex to an active RB kinase (4) (see the figure). But it remains unknown how CDK4 holoenzyme assembly and activity are mediated by p27 and whether p21 is subject to similar regulatory controls.
Guiley et al. find that both p21 and p27 bind cyclin D1 and CDK4 and deform the adenosine triphosphate (ATP) binding pocket of CDK4, which is required for substrate phosphorylation. Yet p27 induces additional structural changes that independently shape the pocket to “prime” it for catalysis. Concordant release of an activation segment from the CDK4 ATP binding pocket exposes Thr172 that is phosphorylated by CDK-activating kinase (CAK, which comprises cyclin H and CDK7) to stimulate subsequent enzyme activity (5). Although none of these structural changes are enough for CDK4 activation, phosphorylation of p27 on Tyr74 weakens its association with CDK4 to elicit recombinant trimer activity; p21 contains a phenylalanine at the analogous position and remains tightly bound and inhibitory. However, there is conflicting evidence that p21, like p27, might function as a CDK4 activator under some circumstances (2), perhaps through other mechanisms. Phosphorylation of p27 Tyr88 and Tyr89, not visualized in the crystal structures of Guiley et al., likely makes an additional contribution to holoenzyme activity (3, 4).
A distinct group of inhibitors—called the INK4 proteins—specifically target CDK4 and CDK6 to arrest proliferation in G1 in an RB-dependent manner (1, 2). In particular, a stress-induced INK4 protein, p16, like RB, is a tumor suppressor, whereas CDK4 and CDK6 have proto-oncogenic activity. The finding that deregulation of the p16-CDK4/6-RB pathway aberrantly drives cancer cell proliferation spurred pharmaceutical development of specific CDK4/6 inhibitory drugs (palbociclib, ribociclib, abemaciclib), which, when used together with inhibitors of estrogen receptor (ER) signaling, were FDA-approved for the treatment of ER-positive breast cancer (6). Combinatorial therapy with CDK4/6 and ER inhibitors significantly increases progression-free survival in women with advanced, metastatic breast cancer (7–9), and clinical trials of CDK4/6 inhibitors are under way for other malignancies (6, 10).
Counterintuitively, Guiley et al. find that palbociclib does not inhibit the active phosphorylated p27-CDK4–cyclin D1 trimer and instead targets monomeric CDK4. Newly synthesized CDK4 requires the heat shock protein 90 (HSP90)–containing chaperone system for proper folding, and palbociclib disrupts this interaction to increase the relative abundance of the CDK4 monomer at the expense of trimer assembly. Therefore, palbociclib mimics p16, which also sequesters monomeric CDK4 to prevent its assembly with cyclin D1 and p27 or p21. The resulting increase in untethered p27 and p21 sets an elevated threshold to be overcome for activation of cyclin E–CDK2 and cyclin A–CDK2 at the G1/S transition (2). Thus, inhibition of both CDK4 and, indirectly, CDK2 by palbociclib or p16 orchestrates G1 arrest of cancer cells.
These findings advance the mechanistic understanding of p27-CDK4–cyclin D1 activation after more than two decades since the component molecules were discovered. The development of palbociclib and related drugs relied on inhibition of recombinant cyclin D and CDK4 produced using baculovirus vectors in insect cells, which express the HSP90-chaperone and CAK. This reconstituted dimeric enzyme retains activity, although vanishingly small quantities of such dimers are detected in mammalian cells. In continuously cycling cells, the rate of progression through G1 into S phase is determined by events in the previous cell cycle. Notably, cyclin D1 is preferentially degraded in S phase and expressed again in G2, raising the possibility that palbociclib “traps” monomeric CDK4 late in the cell cycle, preventing its assembly with cyclin D1 and leading to RB-dependent G1 arrest. It is unknown what factors determine the equilibrium between palbociclib-sensitive monomeric CDK4 and the drug-resistant p27-CDK4–cyclin D1 trimer, or whether p27 and p21 compete in forming active and inactive trimers. Moreover, proliferating cells bifurcate into two subpopulations after the anaphase stage of mitosis based on the amounts of replicative DNA damage and induction of p21 by the tumor suppressor p53 in the previous cycle (11, 12). Cells inheriting high p21 (and low CDK2) expression exhibit a greater dependency on renewed RB phosphorylation to progress through G1, implying that nonuniform responses to CDK4 inhibitors depend on variable inheritance of key regulators (e.g., p21, p27, cyclin D1).
Unlike CDK4, CDK6 is a weak client of the HSP90-chaperone and is generally dependent on cyclins D2 or D3 for activity (13). Robust CDK6 activity can confer palbociclib resistance in CDK4-driven cancers (14, 15), possibly because CDK6 assembles more efficiently than CDK4 into palbociclib-resistant trimers. Structures of CDK4 and CDK6 in complexes with the D-type cyclins are needed to understand differences in regulatory mechanisms and to improve drug development.
(원문: 여기를 클릭하세요~)