A molecule with a sense of timing uration of kinase signaling determines how a cell responds to external cues like growth factors. But how a cell measures signaling time has been a mystery. Now, Leon Murphy, John Blenis (Harvard Medical School, Boston, MA), and colleagues have found that immediate early gene (IEG) products have a sensor mechanism that makes these distinctions. Their findings may provide a target for cancer treatment drugs. Growth factors PDGF and EGF elicit different responses in fibroblasts, with only PDGF inducing cells to enter S-phase. The group found that EGF transiently activated ERK MAP kinases, whereas PDGF stimulated prolonged activation. The IEG product c-Fos is known to be transcribed with similar ki-netics in response to either growth factor , but c-Fos was only phosphorylated when ERK activity was sustained. This not only stabilized c-Fos, but also primed it for further phosphorylation by revealing a DEF domain, an ERK-docking motif. Without secondary phos-phorylation, c-Fos was less able to promote cellular transformation. "Since several other IEGs also have DEF domains," says Blenis, "this group of IEGs may act as molecular sensors in many processes," including neuronal differentiation and the generation of an immune response. Blenis and Murphy hope it will be possible to antagonize these sensors to block specifically ERK-regulated proliferation, such as that seen in Ras-induced cancers, thus possibly avoiding toxic side effects caused by disturbing ERK's other homeostatic cellular functions.

We demonstrate that members of the erk-encoded family of mitogen-activated protein (MAP) kinases (pp44/421w/*) and members of the rsk-encoded protein kinases (RSKs or pp9Ok) are present in the cytoplasm and nucleus of HeLa cells. Addition of growth factors to serum-deprived cells results in increased tyrosine and threonine phosphorylation and in the activation of cytosolic and nuclear MAP phosphorylation, suggesting that protein-SerlThr kinases exist upstream in the signaling pathway that regulate these enzymes and that may in turn be regulated by proteintyrosine (Tyr) kinases or protein kinase C (13, 14, 52). At present, no evidence exists for the putative pp70S6K-protein kinase. However, recent studies have identified mitogenregulated, pp90Ysk-activating protein kinases (16,51).These studies showed that a 42-kDa insulin-stimulated, microtubule-associated protein 2 (MAP2) kinase could partially reactivate protein phosphatase-inactivated S6 protein kinase II (51), a Xenopus homolog of the pp9orsk family (3,32

A single unattached kinetochore can delay anaphase onset in mitotic tissue culture cells (Rieder, C.L., A. Schultz, R. Cole, G. Sluder. 1994. J. Cell Biol. 127:1301–1310). Kinetochores in vertebrate cells contain multiple binding sites, and tension is generated at kinetochores after attachment to the plus ends of spindle microtubules. Checkpoint component Mad2 localizes selectively to unattached kinetochores (Chen, R.-H., J.C. Waters, E.D. Salmon, and A.W. Murray. 1996. Science. 274:242–246; Li, Y., and R. Benezra. Science. 274: 246–248) and disappears from kinetochores by late metaphase, when chromosomes are properly attached to the spindle. Here we show that Mad2 is lost from PtK1 cell kinetochores as they accumulate microtubules and re-binds previously attached kinetochores after microtubules are depolymerized with nocodazole. We also show that when kinetochore microtubules in metaphase cells are stabilized with taxol, tension at kinetochores is lost. The phosphoepitope 3f3/2, which has been shown to become dephosphorylated in response to tension at the kinetochore (Nicklas, R.B., S.C. Ward, and G.J. Gorbsky. 1995. J. Cell Biol. 130:929–939), is phosphorylated on all 22 kinetochores after tension is reduced with taxol. In contrast, Mad2 only localized to an average of 2.6 out of the 22 kinetochores in taxol-treated PtK1 cells. Therefore, loss of tension at kinetochores occupied by microtubules is insufficient to induce Mad2 to accumulate on kinetochores, whereas unattached kinetochores consistently bind Mad2. We also found that microinjecting antibodies against Mad2 caused cells arrested with taxol to exit mitosis after ∼12 min, while uninjected cells remained in mitosis for at least 6 h, demonstrating that Mad2 is necessary for maintenance of the taxol-induced mitotic arrest. We conclude that kinetochore microtubule attachment stops the Mad2 interactions at kinetochores which are important for inhibiting anaphase onset.

The spindle assembly checkpoint delays anaphase until all chromosomes are attached to a mitotic spindle. The mad (mitotic arrest-deficient) and bub (budding uninhibited by benzimidazole) mutants of budding yeast lack this checkpoint and fail to arrest the cell cycle when microtubules are depolymerized. A frog homolog of MAD2 (XMAD2) was isolated and found to play an essential role in the spindle assembly checkpoint in frog egg extracts. XMAD2 protein associated with unattached kinetochores in prometaphase and in nocodazole-treated cells and disappeared from kinetochores at metaphase in untreated cells, suggesting that XMAD2 plays a role in the activation of the checkpoint by unattached kinetochores. This study furthers understanding of the mechanism of cell cycle checkpoints in metazoa and provides a marker for studying the role of the spindle assembly checkpoint in the genetic instability of tumors.