Intracellular protein binding to asbestos induces aneuploidy in human lung fibroblasts.

Exposure to the natural mineral fiber asbestos causes severe lung-damaging fibrosis and cancer, yet it continues to be used as an industrial insulating material throughout the world. When cultured human lung cells are exposed to asbestos, individual fibers are engulfed into the cytoplasm where they induce significant mitotic aberrations leading to chromosomal instability and aneuploidy. The mechanisms of how asbestosis ultimately leads to lung cancer remain unclear. However, our experiments indicate that intracellular asbestos fibers induce aneuploidy and chromosome instability by binding to a subset of proteins that include regulators of the cell cycle, cytoskeleton, and mitotic process. Moreover, precoating of fibers with protein complexes efficiently blocked asbestos-induced aneuploidy in human lung cells without affecting their uptake by cells. These results provide new evidence that asbestos fibers can contribute to significant spindle damage and chromosomal instability by binding to proteins needed for the assembly and regulation of the cytoskeleton or the cell cycle.

Dynamic association of a tumor amplified kinase, Aurora-A, with the centrosome and mitotic spindle.

Aurora-A kinase, also known as STK15/BTAK kinase, is a member of a serine/threonine kinase superfamily that includes the prototypic yeast Ipl1 and Drosophila aurora kinases as well as other mammalian and non-mammalian aurora kinases involved in the regulation of centrosomes and chromosome segregation. The Aurora-A gene is amplified and overexpressed in a wide variety of human tumors. Aurora-A is centrosome-associated during interphase, and binds the poles and half-spindle during mitosis; its over-expression has been associated with centrosome amplification and multipolar spindles. GFP-Aurora-A was used to mark centrosomes and spindles, and monitor their movements in living cells. Centrosome pairs labeled with GFP-Aurora-A are motile throughout interphase undergoing oscillations and tumbling motions requiring intact microtubules and ATP. Fluorescence recovery after photobleaching (FRAP) was used to examine the relative molecular mobility of GFP-Aurora-A, and GFP-labeled alpha-tubulin, gamma-tubulin, and NuMA. GFP-Aurora-A rapidly exchanges in and out of the centrosome and mitotic spindle (t(1/2) approximately 3 sec); in contrast, both tubulins are relatively immobile indicative of a structural role. GFP-NuMA mobility was intermediate in both interphase nuclei and at the mitotic spindle (t(1/2) approximately 23-30 sec). Deletion mapping identifies a central domain of Aurora-A as essential for its centrosomal localization that is augmented by both the amino and the carboxyl terminal ends of the protein. Interestingly, amino or carboxy terminal deletion mutants that maintained centrosomal targeting exhibited significantly slower molecular exchange. Collectively, these studies contrast the relative cellular dynamics of Aurora-A with other cytoskeletal proteins that share its micro-domains, and identify essential regions required for targeting and dynamics.

Localization of human SMC1 protein at kinetochores.

Proper cohesion of sister chromatids is prerequisite for correct segregation of chromosomes during cell division. The cohesin multiprotein complex, conserved in eukaryotes, is required for sister chromatid cohesion. Human cohesion is composed of a stable heterodimer of the structural maintenance of chromosomes (SMC) family proteins, hSMC1 and hSMC3, and non-SMC components, hRAD21 and SA1 (or SA2). In yeast, cohesion associates with chromosomes from late G1 to metaphase and is required for the establishment and maintenance of both chromosome arm and centromeric cohesion. However, in human cells, the majority of cohesion dissociates from chromosomes before mitosis. Although it was recently shown that a small amount of hRAD21 localizes to the centromeres during metaphase, the presence of other cohesion components at the centromere has not been demonstrated in human cells. Here we report the mitosis-specific localization of hSMC1 to the kinetochores. hSMC1 is targeted to the kinetochore region during prophase concomitant with kinetochore assembly and remains through anaphase. Importantly, hSMC1 is targeted only to the active centromere on dicentric chromosomes. These results suggest that hSMC1 is an integral component of the functional kinetochore structure during mitosis.

Protein phosphatase 4 is involved in tumor necrosis factor-alpha-induced activation of c-Jun N-terminal kinase.

Protein phosphatase 4 (PP4, previously named protein phosphatase X (PPX)), a PP2A-related serine/threonine phosphatase, has been shown to be involved in essential cellular processes, such as microtubule growth and nuclear factor kappa B activation. We provide evidence that PP4 is involved in tumor necrosis factor (TNF)-alpha signaling in human embryonic kidney 293T (HEK293T) cells. Treatment of HEK293T cells with TNF-alpha resulted in time-dependent activation of endogenous PP4, peaking at 10 min, as well as increased serine and threonine phosphorylation of PP4. We also found that PP4 is involved in relaying the TNF-alpha signal to c-Jun N-terminal kinase (JNK) as indicated by the ability of PP4-RL, a dominant-negative PP4 mutant, to block TNF-alpha-induced JNK activation. Moreover, the response of JNK to TNF-alpha was inhibited in HEK293 cells stably expressing PP4-RL in comparison to parental HEK293 cells. The involvement of PP4 in JNK signaling was further demonstrated by the specific activation of JNK, but not p38 and ERK2, by PP4 in transient transfection assays. However, no direct PP4-JNK interaction was detected, suggesting that PP4 exerts its positive regulatory effect on JNK in an indirect manner. Taken together, these data indicate that PP4 is a signaling component of the JNK cascade and involved in relaying the TNF-alpha signal to the JNK pathway.

Managing the centrosome numbers game: from chaos to stability in cancer cell division.

Aneuploid tumor cells can arise through multipolar mitosis caused by supernumerary centrosomes. Multipolar spindles, however, are antagonistic to cell viability. Thus, most cells derived from such an aberrant mitosis would be eliminated by apoptosis. A rare daughter cell, through chance acquisition of an appropriate chromosome complement and/or gene dosage, could survive and contribute to a clone of aneuploid tumor cells. Survival and perpetuation of the clone, however, requires an additional step - the resumption of mitotic stability through the assembly of a bipolar, not multipolar, spindle. Either selective inactivation of the extra centrosomes or their coalescence into two functional spindle poles corrects the problem of centrosome excess. Current data support coalescence as a mechanism for regulating the number of functional centrosomes in tumor cells.

Progesterone facilitates chromosome instability (aneuploidy) in p53 null normal mammary epithelial cells.

Mammary epithelial cells from p53 null mice have been shown recently to exhibit an increased risk for tumor development. Hormonal stimulation markedly increased tumor development in p53 null mammary cells. Here we demonstrate that mammary tumors arising in p53 null mammary cells are highly aneuploid, with greater than 70% of the tumor cells containing altered chromosome number and a mean chromosome number of 56. Normal mammary cells of p53 null genotype and aged less than 14 wk do not exhibit aneuploidy in primary cell culture. Significantly, the hormone progesterone, but not estrogen, increases the incidence of aneuploidy in morphologically normal p53 null mammary epithelial cells. Such cells exhibited 40% aneuploidy and a mean chromosome number of 54. The increase in aneuploidy measured in p53 null tumor cells or hormonally stimulated normal p53 null cells was not accompanied by centrosome amplification. These results suggest that normal levels of progesterone can facilitate chromosomal instability in the absence of the tumor suppressor gene, p53. The results support the emerging hypothesis based both on human epidemiological and animal model studies that progesterone markedly enhances mammary tumorigenesis.

The centrosome-associated Aurora/Ipl-like kinase family.

Because of the well-known role of the centrosome and mitotic apparatus in genome partitioning in normal cells, defects in pathways essential for mitotic regulation are likely implicated in the cascade of events leading to aneuploidy and neoplasia. Exogenous overexpression of AIM-1, for example, produces multinuclearity in human cells and increased ploidy as well as aneuploidy (Tatsuka et al., 1998). Overexpression in colorectal tumor cell lines is thought to have a causal relationship with multinuclearity and increased ploidy. Cytokinesis error caused by AIM-1 overexpression is a major factor in the predisposition to cancer. As previously mentioned, the involvement of BTAK/aur2/AIK in centrosome amplification and its oncogenic activity are compelling. Aur2 has also been implicated in oncogenesis, and defects in kinetochore function leading to chromosome instability in human tumors should not be minimized (Farruggio et al., 1999). Further studies are needed to provide a clearer definition of how these kinetic proteins are linked and regulated in normal mitosis and cancer. Thus, Boveri appears to have been correct in formulating his early hypothesis that a defective mitotic apparatus and centrosome number were central and causative in chromosome missegregation and cancer. One hundred years later, at the onset of a new millennium and with light-years of advanced technology in our favor, we are just now beginning to piece together the enzymes, substrates, and signaling pathways that support and explain his long-ignored but prophetic claim.

Aberrant cell cycle progression contributes to the early-stage accelerated carcinogenesis in transgenic epidermis expressing the dominant negative TGFbetaRII.

Mutations in the transforming growth factor beta type II receptor (TGFbetaRII) have been found in various malignant tumors, suggesting that loss of TGFbeta signaling plays a causal role in late-stage cancer development. To test whether loss of TGFbetaRII is involved in early-stage carcinogenesis, we have generated transgenic mice expressing a dominant negative TGFbetaRII (deltabetaRII) in the epidermis. These mice exhibited an increased susceptibility to chemical carcinogenesis protocols at both early and late stages. In the current study, parameters for cell cycle progression and chromosome instability were analysed in deltabetaRII tumors. DeltabetaRII papillomas showed an increased S phase in flow cytometry. Bromodeoxyuridine (BrdU) labeling and mitotic indices in deltabetaRII papillomas also showed a threefold increase compared to papillomas developing in non-transgenic mice. When papillomas further progressed to squamous cell carcinomas (SCC), both control and deltabetaRII SCC showed similar BrdU labeling indices and percentages of S phase cells. However, deltabetaRII SCC cells showed a sixfold increase in the G2/M population. Mitotic indices in deltabetaRII SCC also showed a threefold increase compared to non-transgenic SCC. Consistent with a perturbed cell cycle, deltabetaRII papillomas and SCC showed reduced expression of the TGFbeta target genes p15 (INK4b), p21 (WAF-1) and p27 (Kip1), inhibitors of cyclin-dependent kinases (cdks). However, most deltabetaRII papilloma cells exhibited normal centrosome numbers, and deltabetaRII SCC exhibited a similar extent of centrosome abnormalities compared to control SCC (35-40% cells). Most of deltabetaRII SCC exhibited diploid chromosome profiles. These data indicate that inactivation of TGFbetaRII accelerates skin tumorigenesis at early stages by the acceleration of loss of cell cycle control, but not by increased chromosome instability.

Analysis of centrosome abnormalities and angiogenesis in epidermal-targeted p53172H mutant and p53-knockout mice after chemical carcinogenesis: evidence for a gain of function.

We previously developed a transgenic mouse model that expresses in the epidermis a murine p53172R-->H mutant (p53m) under the control of a human keratin-1-based vector (HK1.p53m). In contrast to mice with wild-type p53 and p53-knockout mice, HK1.p53m mice exhibit increased susceptibility to chemical carcinogenesis, with greatly accelerated benign papilloma formation, malignant conversion, and metastasis. In the study presented here, we examined the expression pattern of several differentiation markers and observed that p53m tumors exhibited a less differentiated phenotype than tumors elicited in non-transgenic mice. Metastasis in p53m tumors was also associated with a poorly differentiated phenotype. To determine whether genomic instability was associated with a putative gain-of-function role for this p53m, in situ examination of centrosomes was performed in HK1.p53m and equivalent p53-null papillomas. In contrast to HK1.p53m papillomas, which had centrosome abnormalities at high frequencies (75% of cells contained more than three centrosomes/cell), p53-null tumors exhibited few abnormal centrosomes (4% of cells contained more than three centrosomes/cell). To determine whether angiogenesis played a role in the rapid progression of p53m tumors, the expression of vascular endothelial growth factor, a promoter of angiogenesis, and thrombospondin-1, an inhibitor of angiogenesis, was examined in tumors derived from either p53m or p53-knockout mice. Regardless of their p53 status (wild type, p53m, p53-/-), all of the papillomas exhibited similar levels of vascular endothelial growth factor expression and decreased expression of thrombospondin-1 as did normal epidermis. In addition, tumors from different p53 genotypes showed a similar density of blood vessels. Because p53 status did not appear to play an overt role in angiogenesis, these data suggest that p53m accelerates tumorigenesis primarily by exerting a gain of function associated with genomic instability.

Tumour amplified kinase STK15/BTAK induces centrosome amplification, aneuploidy and transformation.

The centrosomes are thought to maintain genomic stability through the establishment of bipolar spindles during cell division, ensuring equal segregation of replicated chromosomes to two daughter cells. Deregulated duplication and distribution of centrosomes have been implicated in chromosome segregation abnormalities, leading to aneuploidy seen in many cancer cell types. Here, we report that STK15 (also known as BTAK and aurora2), encoding a centrosome-associated kinase, is amplified and overexpressed in multiple human tumour cell types, and is involved in the induction of centrosome duplication-distribution abnormalities and aneuploidy in mammalian cells. STK15 amplification has been previously detected in breast tumour cell lines and in colon tumours; here, we report its amplification in approximately 12% of primary breast tumours, as well as in breast, ovarian, colon, prostate, neuroblastoma and cervical cancer cell lines. Additionally, high expression of STK15 mRNA was detected in tumour cell lines without evidence of gene amplification. Ectopic expression of STK15 in mouse NIH 3T3 cells led to the appearance of abnormal centrosome number (amplification) and transformation in vitro. Finally, overexpression of STK15 in near diploid human breast epithelial cells revealed similar centrosome abnormality, as well as induction of aneuploidy. These findings suggest that STK15 is a critical kinase-encoding gene, whose overexpression leads to centrosome amplification, chromosomal instability and transformation in mammalian cells.

Expression of a p53 mutant in the epidermis of transgenic mice accelerates chemical carcinogenesis.

To develop an in vivo model for studying the role of the p53 tumor suppressor in skin carcinogenesis, a murine p53(172H) mutant (equivalent to human p53(175H)) was expressed in the epidermis of transgenic mice, utilizing a targeting vector based on the human keratin 1 gene (HK1.p53m). HK1.p53m mice developed normally and did not exhibit an obvious epidermal phenotype or develop spontaneous tumors. However, these mice demonstrated an increased susceptibility to a two-stage chemical carcinogenesis protocol, with the rate of formation and number of papillomas being dramatically increased as compared to non-transgenic controls. The majority of papillomas in control mice regressed after termination of 12-O-tetradecanoylphorbol-13-acetate (TPA) treatment, whereas p53m papillomas progressed to carcinomas and metastases. In addition, more advanced malignancy, i.e., undifferentiated spindle cell carcinomas, were exclusively observed in p53m mice. Increased bromodeoxyuridine (BrdU) labeling, accompanied by decreased expression of p21, was observed in HK1.p53m papillomas. In situ examination of centrosomes in HK1.p53m papillomas also revealed marked abnormalities, with 75% of the cells containing > or = 3 centrosomes/cell, whereas centrosome numbers in papillomas from control animals remained normal. These data suggest that the accelerated tumorigenesis observed in chemically-treated p53m mice is most likely due to increased genomic instability resulting from an inhibition of G1 arrest and abnormal amplification of centrosomes.

CENP-G: a new centromeric protein that is associated with the alpha-1 satellite DNA subfamily.

A new constitutive centromere-specific protein (CENP) has been identified as a result of its recognition as an autoantigen by serum from a patient with gastric antral vascular ectasia disease. Conventional immunoblotting and two-dimensional double blotting with both this antiserum and a known anti-centromere antiserum showed that this antiserum predominantly recognized a Mr 95,000 protein that is different from all known CENPs. We have named this new protein CENP-G. This protein was detected at the centromeric region throughout the cell cycle. In mitosis, it was restricted to the kinetochore inner plate as shown by immunogold labeling and electron microscopy. The centromeres of some human chromosomes are known to contain two subfamilies of alpha-satellite DNA. Using immunofluorescence combined with fluorescent in situ hybridization with subfamily-specific DNA probes, we revealed that CENP-G was specifically associated with one of the subfamilies, which we have named alpha-1, but not the other. The localization and the alpha-1-specific association suggested that CENP-G may play a role in kinetochore organization and function. Like CENP-B and C, but unlike CENP-A, this protein remained with the nuclear matrix after intensive extraction. While CENP-B is absent from the human Y chromosome, the existence of CENP-G on the Y chromosome has been proven by immunofluorescence and whole chromosome painting. CENP-G was also detected in CHO, Indian muntjac and Chinese muntjac cells, suggesting that it is conserved in evolution.

The cenpB gene is not essential in mice.

Centromere protein B (CENP-B) is a centromeric DNA-binding protein that binds to alpha-satellite DNA at the 17 bp CENP-B box sequence. The binding of CENP-B, along with other proteins, to alpha-satellite DNA sequences at the centromere, is thought to package the DNA into heterochromatin subjacent to the kinetochore of mitotic chromosomes. To determine the importance of CENP-B to kinetochore assembly and function, we generated a mouse null for the cenpB gene. The deletion removed part of the promoter and the entire coding sequence except for the carboxyl-terminal 35 amino acids of the CENP-B polypeptide. Mice heterozygous or homozygous for the cenpB null mutation are viable and healthy, with no apparent defect in growth and morphology. We have established mouse embryo fibroblasts from heterozygous and homozygous cenpB null littermates. Microscopic analysis, using immunofluorescence and electron microscopy of the cultured cells, indicated that the centromere-kinetochore complex was intact and identical to control cells. Mitosis was identical in fibroblasts derived from cenpB wild-type, heterozygous and null animals. Our studies demonstrate that CENP-B is not required for the assembly of heterochromatin or the kinetochore, or for completion of mitosis.

Dynamic continuity of nuclear and mitotic matrix proteins in the cell cycle.

The eukaryotic cell nucleus is a membrane-enclosed compartment containing the genome and associated molecules supported by a highly insoluble filamentous network known as the nucleoskeleton or nuclear matrix. The nuclear matrix is believed to play roles in maintaining nuclear architecture and organizing nuclear metabolism. Recently, advances in microscopic techniques and the availability of new molecular probes have made it possible to localize functional domains within the nuclear matrix and demonstrate dynamic interactions between both soluble and insoluble components involved in the control of multiple nuclear transactions. Like the cytoplasm and its skeleton, the nucleoplasm is highly structured and very crowded with an equally complex skeletal framework. In fact, there is growing evidence that the two skeletal systems are functionally contiguous, providing a dynamic cellular matrix connecting the cell surface with the genome. If we impose cell cycle dynamics upon this skeletal organization, it is obvious that the genome and associated nuclear matrix must undergo a major structural transition during mitosis, being disassembled and/or reorganized in late G2 and reassembled again in daughter nuclei. However, recent evidence from our laboratory and elsewhere suggests that much of the nuclear matrix is used to form the mitotic apparatus (MA). Indeed, both facultative and constitutive matrix-associated proteins such as NuMA, CENP-B, CENP-F, and the retinoblastoma protein (Rb) associate within and around the MA. During mitosis, the nuclear matrix proteins may either become inert "passengers" or assume critical functions in partitioning the genome into newly formed G1 nuclei. Therefore, we support the view that the nuclear matrix exists as a dynamic architectural continuum, embracing the genome and maintaining cellular regulation throughout the cell cycle.

Nuclear-mitotic apparatus protein: a structural protein interface between the nucleoskeleton and RNA splicing.

Vertebrate splicing factors are localized to discrete domains within the nuclei of somatic cells. The mechanism whereby such nuclear domains, identified as speckles by immunofluorescence microscopy, are generated is unclear. Recent studies suggest that the spatial order within the nucleus is maintained by nuclear matrix factors. Here we show that a protein in the nuclear matrix and mitotic apparatus [nuclear-mitotic apparatus protein, NuMA; Lydersen, B. & Pettijohn, D. (1980) Cell 22, 489-499] colocalizes with splicing factors in interphase nuclei and is associated with small nuclear ribonucleoproteins in a complex immunoprecipitated from HeLa extract with small nuclear ribonucleoprotein antibodies. Moreover, NuMA associates with splicing complexes that are reconstituted in vitro using wild-type pre-mRNA, but not with nonspecific RNA. Cumulatively, these observations suggest a function of NuMA or NuMA-like proteins in interphase cells in providing a bridge between RNA processing and the nucleoskeleton.

Localization of NuMA protein isoforms in the nuclear matrix of mammalian cells.

Using a monoclonal antibody 2D3 generated against a kinetochore-enriched human chromosome preparation, we identified a high molecular mass protein with nuclear staining in interphase and polar staining of the pericentriolar region in the mitotic spindle. Initially termed centrophilin, this protein associates with the minus-ends of spindle microtubules (MT) and appears to be important in spindle organization [Tousson et al., 1991: J. Cell Biol. 112:427-440]. Comparison of a partial cDNA sequence obtained for centrophilin with the full length cDNA sequence of nuclear mitotic apparatus protein (NuMA) [Compton et al., 1992: J. Cell Biol. 116:1395-1408; Yang et al., 1992: J. Cell Biol. 116:1303-1317] has indicated that NuMA and centrophilin are the same protein. Using a polyclonal NuMA antibody, we have provided further evidence that NuMA exists as isoforms as shown by peptide mapping and immunoblots. Sequential fractionation experiments along with immunofluorescence, immunoblotting, and EM immunogold labeling have demonstrated that NuMA isoforms are novel components of nuclear core filaments. Thus, NuMA, a long coiled-coil protein, appears to have dual functions in interphase and mitosis during the cell cycle. In interphase, NuMA likely plays a structural role in the nucleoskeleton that may be important in nuclear organization and functions, whereas in mitosis, NuMA appears to be associated with spindle MT organization and chromosome positioning.

Hepatic gene therapy: efficient gene delivery and expression in primary hepatocytes utilizing a conjugated adenovirus-DNA complex.

Receptor-mediated endocytosis is an effective method for gene delivery into target cells. We have previously shown that DNA molecules complexed with asialoglycoprotein can be efficiently endocytosed by primary hepatocytes and the internalized DNA can be released from endosomes by the use of a replication-defective adenovirus. Because the DNA and virus enter target cells independently, activity enhancement requires high concentrations of adenoviral particles. In this study, adenoviral particles were chemically conjugated to poly(L-lysine) and bound ionically to DNA molecules. Quantitative delivery to primary hepatocytes was achieved with significantly reduced viral titer when the asialoorosomucoid-poly(L-lysine) conjugate was included in the complex. The conjugated adenovirus was used to deliver a DNA vector containing canine factor IX to mouse hepatocytes, resulting in the expression of significant concentrations of canine factor IX in the culture medium. The results suggest that receptor-mediated endocytosis coupled with an efficient endosomal lysis vector should permit the application of targeted and efficient gene delivery into the liver for gene therapy of hepatic deficiencies.

Identification of a membrane protein from T84 cells using antibodies made against a DIDS-binding peptide.

The outwardly rectified chloride channel of secretory epithelial cells is inhibited by disulfonic stilbene (DS) compounds such as 4,4'-diisothiostilbene-2,2'-disulfonic acid (DIDS) [R. J. Bridges, R. T. Worrell, R. A. Frizzell, and D. J. Benos, Am. J. Physiol. 256 (Cell Physiol. 25): C902-C912, 1989]. A 13-amino acid peptide (P49) corresponding to the putative DS binding site region of the murine anion exchange protein was synthesized, and polyclonal antibodies were generated against it and then purified over a P49 affinity column. The resulting monospecific antibodies reacted on Western blots with a 95- to 100-kDa protein from human erythrocytes and a 55- to 60-kDa protein from the human colonic tumor cell line, T84. The reaction with T84 protein did not appear to represent recognition of an anion exchanger because anion efflux from T84 cells was independent of external Cl-. In addition, monoclonal antibodies raised against human band 3 recognized the band 3 protein in human red cell ghost preparations but recognized nothing in T84 cell membrane preparations. In T84 cells, DIDS protected the 60-kDa protein from antibody binding. The anti-P49 antibody blocked outwardly rectified Cl- channels incorporated into planar lipid bilayer membranes from rat colon. Immunocytochemical data reveal specific binding of the anti-P49 antibody to perinuclear cytoplasmic vesicles. Forskolin caused these antibody-labeled vesicles to migrate from the perinuclear region to the plasma membrane under conditions and with a time course identical to that seen for stimulation of Cl- transport in these cells. Our results suggest that the protein may be a part of a chloride channel complex of secretory epithelial cells.

The centromere-kinetochore complex: a repeat subunit model.

The three-dimensional structure of the kinetochore and the DNA/protein composition of the centromere-kinetochore region was investigated using two novel techniques, caffeine-induced detachment of unreplicated kinetochores and stretching of kinetochores by hypotonic and/or shear forces generated in a cytocentrifuge. Kinetochore detachment was confirmed by EM and immunostaining with CREST autoantibodies. Electron microscopic analyses of serial sections demonstrated that detached kinetochores represented fragments derived from whole kinetochores. This was especially evident for the seven large kinetochores in the male Indian muntjac that gave rise to 80-100 fragments upon detachment. The kinetochore fragments, all of which interacted with spindle microtubules and progressed through the entire repertoire of mitotic movements, provide evidence for a subunit organization within the kinetochore. Further support for a repeat subunit model was obtained by stretching or uncoiling the metaphase centromere-kinetochore complex by hypotonic treatments. When immunostained with CREST autoantibodies and subsequently processed for in situ hybridization using synthetic centromere probes, stretched kinetochores displayed a linear array of fluorescent subunits arranged in a repetitive pattern along a centromeric DNA fiber. In addition to CREST antigens, each repetitive subunit was found to bind tubulin and contain cytoplasmic dynein, a microtubule motor localized in the zone of the corona. Collectively, the data suggest that the kinetochore, a plate-like structure seen by EM on many eukaryotic chromosomes is formed by the folding of a linear DNA fiber consisting of tandemly repeated subunits interspersed by DNA linkers. This model, unlike any previously proposed, can account for the structural and evolutional diversity of the kinetochore and its relationship to the centromere of eukaryotic chromosomes of many species.