Errors in cell division result in aneuploidy, which is commonly observed in genetic disorders and cancer. Most of these errors are prevented by cell cycle checkpoints, which are cellular control systems that allow certain events to proceed only after the completion of specific events prior to them in the sequence. The spindle checkpoint, for example, is a surveillance system that can delay mitosis by preventing the activation of the anaphase-promoting complex (also called the cyclosome) if spindle organization is defective or chromosomes are incorrectly attached to the spindle. Genes encoding spindle checkpoint components were first isolated from the budding yeast Saccharomyces cerevisiae: they include the mitotic arrest-deficient (MAD) genes MAD1, MAD2, and MAD3; the budding uninhibited by benzimidazole (BUB) genes BUB1, BUB2, and BUB3; and the monopolar spindle gene MPS1. Mutated human homologs of BUB1 have been found in subtypes of colorectal cancer that exhibit chromosome instability.

A human homolog of Sgt1 (HsSgt1A) can rescue the yeast sgt1-null mutant from the lethality of its phenotype. This rescue strongly suggests that the functions of Sgt1 are highly conserved in yeast and humans. To examine how depletion of HsSgt1 proteins in mammalian cells affects chromosome segregation in those cells, we used RNA interference (in collaboration with Dr. Andrea Musacchio). We also performed experiments in which deletion mutants of human Sgt1 were overexpressed by the cells. We found that Sgt1 depletion or overexpression of the mutant Sgt1 causes mitotic defects and that in Sgt1-depleted HeLa cells, some previously characterized outer or central kinetochore proteins are not present at kinetochores.

Our immunoprecipitation-mass spectrometry analyses have shown that, like yeast Sgt1, human Sgt1 interacts with Hsp90, Hsp70, and Skp1. The Hsp90 inhibitor 17-allylaminogeldanamycin (17-AAG, a geldanamycin derivative), which is also a novel anticancer drug being tested clinical trials, induces a mitotic delay in human cells in culture. We found that the mitotic delay caused by 17-AAG depends on the spindle checkpoint, and a significant number of 17-AAG–treated cells had misaligned chromosomes at metaphase. Several known kinetochore proteins were not present at the kinetochores in 17-AAG–treated cells. We also found that Hsp90 is required from late S phase to early G2 phase for the recruitment of kinetochore proteins. Thus, we discovered a novel anticancer mechanism of 17-AAG. Synthetic lethality occurred in response to 17-AAG treatment and application of human Sgt1 siRNA. Our results strongly suggest that the Sgt1-Hsp90 complex is required for the assembly of kinetochore protein complexes in human cells.

However, human Mad2 depletion did not induce this DNA fragmentation. This novel mitotic cell death appeared to be independent of caspase activation. We found that AIF (apoptosis-inducing factor) and EndoG (endonuclease G), which are effectors of caspase-independent cell death, are released from mitochondria during the activation of CIMD and that the DNA fragmentation depends on AIF and EndoG. We also found that CIMD depends on p73 (a homolog of p53) but not p53. Therefore, our results suggest that human Bub1 is a crucial factor in regulating cell death during mitosis, and we propose that CIMD protects cells from aneuploidy by inducing the death of cells that are prone to substantial chromosome missegregation.