Courtesy of Philipp Kaldis, © 2003 Elsevier
After several groups reported discovering cyclin E and cyclin-dependent kinase 2 (CDK2) in 1991, a consensus emerged. It held that these protein partners are crucial in promoting the cell cycle's G1- to S-phase transition and driving cancer-cell proliferation. Therapeutically minded investigators accordingly began to seek inhibitors that would target CDK2's ATP-binding pocket and catalytic site.
But five recently published papers demolish this consensus view. Two researchers knocked down CDK2 protein expression in cancer cell lines, and the cells continued to divide.1 Four teams of scientists knocked out the CDK2 or cyclin E genes (CCNE1 and CCNE2), and the mice were viable, or their cells proliferated in culture.2-5 Many of the rodents, however, were sterile.
When Mariano Barbacid, director of the Spanish National Cancer Center in Madrid, heard that his newly generated CDK2-null mice were running around in the animal facility, "My first thought was, we've done something wrong," he says. The day after he presented his study at a June 2002 symposium near Helsinki, he recalls conferees telling him that they had "spent the whole night talking about the damned CDK2-knockout mice because it was really fairly shocking."
The new papers should likewise roil the cell-cycle field, prompting arguments over whether scientists overinterpreted earlier CDK2 and cyclin E findings. Biologists will have to revisit what these proteins really do in vivo and which proteins might compensate for their loss. Theorists must account for how these proteins contribute to meiosis. And cancer researchers might reevaluate efforts to develop drugs aimed at CDK2 or indeed at any single cell-cycle protein.
CYCLING CELLS, SUBTLE DEFECTS During the cell's G1 phase, two partnerships--CDK4/cyclin D and later CDK2/cyclin E --are believed to promote mitosis by phosphorylating the retinoblastoma protein. Phosphorylation prevents the protein from inhibiting various transcription factors necessary for cell-cycle progression.
In 1997, when Barbacid was at Bristol-Myers Squibb, his team began to validate targets for anticancer drugs by knocking out genes for various cyclin-dependent kinases (CDKs). After the elimination of CDK4 led to diabetes, CDK2 was next on the chopping block.3 Independent of Barbacid's effort, Philipp Kaldis, a principal investigator at the National Cancer Institute, also generated a CDK2 knockout.5
The two groups' observations, while not identical, don't conflict. The knockout mice survive to adulthood, and their embryonic fibroblasts proliferate in culture. Compared to wild-type cells, however, those derived from knockouts divide more slowly, show delayed entry into S phase, and resist oncogenic transformation. The animals are also sterile. Spermatocyte defects become apparent at the pachytene (chromosomal crossover) stage of meiotic prophase I; oocyte abnormalities occur at the dictyate (arrested) stage, which lasts weeks in mice and decades in humans.
As the knockout studies were proceeding, Frank McCormick, director of the Cancer Research Institute at the University of California, San Francisco, was approaching CDK2 from a different angle. He and postdoc Osamu Tetsu (now an assistant professor at UCSF) found that overexpression of the p27KIP1 protein, an endogenous CDK2 inhibitor, does not stop colon cancer cells from proliferating.
Further CDK2-inhibition methods, including antisense, RNA interference, and dominant-negative mutants, also failed to quench mitosis in cells cultured from colon cancers, osteosarcomas, and cervical cancers.1 But McCormick says he had trouble publishing the paper "because the referees simply didn't believe it." He speculates that Barbacid's public disclosure of his viable CDK2 knockouts finally mollified the paper's critics.
Longstanding attempts to create a cyclin E knockout encountered a different sort of obstacle: the discovery of cyclin E2, reported by three groups in 1998 and 1999. The two cyclins have overlapping biochemical properties and expression patterns, so both genes had to be eliminated to achieve a full effect. Working separately but cooperating, two teams made single and double knockouts. One was headed by Peter Sicinski of Harvard Medical School;2 the other, by Bruno Amati, now at the European Institute of Oncology in Milan.4
Amati remembers being surprised to find that mice lacking both E-type cyclins die "relatively late," just before embryonic day 12. Placental abnormalities appear to be the fatal flaw. Yet fibroblasts derived from double-knockout embryos proliferate, albeit a bit more slowly than wild-type cells. Cyclin E-null cells also cannot reenter the cell cycle from the quiescent G0 state, and they resist oncogenic transformation. Another dramatic result: About half of male mice without cyclin E2 are sterile and exhibit testicular abnormalities, while mutant females are reproductively normal.
IMPLICATIONS FOR CANCER In reaction to the new papers, McCormick asserts that "there's no real evidence that CDK2 does anything" in somatic cells. Other researchers, less skeptical about previous hypotheses, believe that CDK2 plays a cell-cycle role that's usurped by other kinases in knockout mice. But establishing that one enzyme compensates for another in a mammalian system is "very difficult," notes Kaldis. Test-tube or cell-culture experiments might reveal that two enzymes phosphorylate the same substrates, but such studies cannot prove what happens in vivo, he explains.
Courtesy of Peter Sicinski, © 2003 Elsevier
Multiple knockouts provide better evidence of enzymatic compensation, so Barbacid's group and undoubtedly others are trying to generate a mouse that lacks CDK2, CDK4, and CDK6 (which is very similar to CDK4). Barbacid is also interested in introducing germ-line point mutations to create dead kinases, a method that he says should surpass gene knockouts at predicting how drug inhibitors might work.
Knockout mice, Sicinski points out, merely indicate which cellular functions must be performed by the missing gene's protein product, and which functions can rely on other proteins. In the case of E-type cyclins, he adds, "the first group, very surprisingly, is very small, and the second group is very large. We expected the opposite." To account for the knockouts' resistance to oncogenic transformation, he speculates that "there is a critical threshold, a critical level of cyclin/CDK activity, that cells need to undergo transformation," and that the knockout cells do not reach this level.
Though CDK2 no longer appears crucial for cell proliferation, researchers do not yet consider it irrelevant to cancer. McCormick plans to investigate whether CDK2 inhibition retards the proliferation of breast- cancer cells. Philip W. Hinds, an associate pathology professor at Harvard Medical School, wants to learn whether knocking down CDK2 will induce cellular senescence, an innate tumor-suppressive process.
The recent papers might discourage drug companies from trying to develop inhibitors of a single CDK. Bristol-Myers Squibb's CDK2 inhibitor BMS-387032 has been tested in small Phase I clinical trials, but observers question the degree to which it targets only one CDK. Adrian M. Senderowicz, director of the molecular therapeutics unit at the National Institute of Dental and Craniofacial Research, suggests confirming pharmaceutical specificity this way: Observe whether a drug stops the proliferation of cells from knockout animals that lack the CDK supposedly inhibited by that drug.
"Tumor cells are smarter than we think, and we may need to target more than one [CDK]," Senderowicz contends. He says that flavopiridol, a compound he has tested on patients and in the lab, hits at least seven CDKs by occupying their ATP-binding pockets. CYC202 (roscovitine), a CDK2 inhibitor undergoing Phase II trials, affects CDK1 and CDK5. Other potential drugs, such as UCN-01 and perifosine, induce endogenous CDK inhibitors.
MEIOSIS AND INFERTILITY The recent findings raise the question of which cyclin teams up with CDK2 during meiosis. "In testicular extracts, cyclin A1 associates with CDK2," notes Debra J. Wolgemuth, a genetics and development professor at Columbia University College of Physicians and Surgeons. Her lab found that male mice lacking cyclin A1 are sterile because spermatogenesis halts in late meiotic prophase I. But the meiotic defect in cyclin A1 (CCNA1) knockouts seems to happen slightly later than in CDK2 knockouts, so Wolgemuth is unclear about when the genes' protein products might form a putative partnership.
A second issue is which meiotic substrates might be phosphorylated by CDK2. "One of the important events that is occurring as cells are exiting prophase and getting into metaphase is that the synaptonemal complex, which presumably glues together homologous chromosomes, is dissolved," says Mary Ann Handel, a senior research scientist at Jackson Laboratory in Bar Harbor, Maine. "That complex would be a prime candidate for a substrate." Still another issue is whether cyclin A1, cyclin E2, and CDK2 mutations are linked to cases of unexplained human infertility.
Perhaps these and other questions should have arisen a few years ago. The CDK2 and cyclin E knockouts were likely delayed because biologists expected a phenotype that "you really don't like to analyze, which is early embryonic lethality," says Hinds. "At some point, that stops being an excuse, and you have to do the experiment." Some deaths did, in fact, ensue--not only of embryos but also of old preconceptions.
Douglas Steinberg (email@example.com) is a freelance writer in New York City.
1. O. Tetsu, F. McCormick, "Proliferation of cancer cells despite CDK2 inhibition," Cancer Cell, 3:233-45, 2003.
2. Y. Geng et al., "Cyclin E ablation in the mouse," Cell, 114:431-43, Aug. 22, 2003.
3. S. Ortega et al., "Cyclin-dependent kinase 2 is essential for meiosis but not for mitotic cell division in mice," Nat Genet, 35:25-31, September 2003.
4. T. Parisi et al., "Cyclins E1 and E2 are required for endoreplication in placental trophoblast giant cells," EMBO J, 22:4794-803, Sept. 15, 2003.
5. C. Berthet et al., "CDK2 knockout mice are viable," Curr Biol, 13:1775-85, Oct. 14, 2003.