TI - RESULTS . AB - A metaphase chromosome fraction can be prepared from colcemid-treated tissue culture cells by Dounce homogenisation and sequential density gradients ( 27 ) . The metaphase chromosome scaffold generated by stringent extraction of histones from this fraction includes the kinetochore and contains several prominent protein bands upon SDS-PAGE resolution , notably topoisomerase II and ScII ( 27,29,32 , 35,36 ) . A less stringent procedure was followed here , to retain more chromosome -associated material in the search for candidate Aurora B SUBstrates . To confirm that the resultant fraction , shown in Figure 1A and B , indeed contained the chromosome scaffold proteins , we performed immunoblot analysis with antibodies specific for various proteins expected to be in the scaffold . As shown in Figure 1C , this partially-extracted scaffold contained CENP-A , CENP-B , CENP-C , INCENP , Aurora B and ScII . The monoclonal antibody MPM-2 , which recognises a mitotic phosphoepitope , also detected a number of bands . To confirm the reproducibility of this extraction protocol , 11 separate chromosome preparations were run on the same one-dimensional gel . Visual inspection of the band patterns of these different preparations revealed them to have essentially the same composition ( data not shown ) . To examine how this extraction procedure compared with the more stringent protocols previously published , we performed gel electrophoresis on equivalent amounts of chromosomal protein extracted by differing methods . As shown in Figure 1D , there exist clear differences between the protein scaffold fractions prepared by the different extraction methods , but further analysis will be necessary to determine to what extent the content of minor , possibly unknown , components of the scaffolds is increased under the different conditions . To define the components of this chromosome subfraction , we subjected the preparations to one-dimensional SDS - PAGE ( Fig 1A and B ) and excised individual bands from Coomassie-stained gels . MALDI-TOF analysis of the peptide composition of these bands was performed and proteins in the fraction identified by database comparison . Table 1 lists the proteins identified in this initial analysis of the extracted chromosome fraction and Figure 2 summarises these results graphically . Of the 62 proteins identified in the fraction , 29% are nuclear or chromosome -associated . Some 34% are primarily associated with membranes or the cytoskeleton , 18% are mitochondrial , 3% centrosomal and the remainder are from the cytoplasm , act as chaperones or have unknown properties . Since the extent to which the cytoskeletal proteins interact with the chromosomes in forming a scaffold is not clear , some of the cytoskeletal proteins may actually represent chromosome -associated material , rather than artefactual or contaminant protein . We hypothesise that the mitochondria may be co-fractionating with the chromosomes in the Percoll density gradients , which would explain the relatively high proportion of mitochondrial proteins identified . To test whether the identification of proteins by this method can define new chromosomal proteins , we examined the localisation of the putative GTP -binding protein NGB/CRFG during mitosis . Green fluorescent protein ( GFP ) -tagging in pEGFP-C1 and transfection of the gene encoding this protein into HeLa JW cells revealed that it forms subnuclear assemblies in interphase , as has been noted previously ( 46 ) , but that it is associated with chromosomes throughout mitosis ( Fig 3 ) . Transfection with pEGFP-N1-CRFG did not result in any detectable GFP signal and use of the vector alone gave rise to the typical diffuse staining of GFP throughout the cell cycle ( data not shown ) . The perichromosomal localisation , seen clearly in metaphase ( Fig 3B ) , has been observed for a number of nucleolar proteins , but does not exclude a role for this protein in mitotic chromosomes , nuclear function or in cell cycle control ( 47,48 ) . These findings confirm that the analysis of scaffolds in this manner is a useful means to discover new proteins associated with mitotic chromosomes . Since Aurora B itself was shown to be in the chromosome fraction , we tested whether there was a kinase activity associated with the preparation . A number of different buffer conditions were assayed and a reproducible pattern of PHOSphorylated bands was found to result upon incubation of the chromosomes with [gamma-32P] ATP (Fig 4A) , irrespective of the buffer used . This activity was lost upon heat treatment ( Fig 4B ) . At present , the identities of the endogenous kinases that carry out this PHOSphorylation are unknown . To identify new potential SUBstrates of Aurora B , we next incubated the heat-inactivated , extracted chromosomes with recombinant human Aurora B kinase . As shown in Figure 4B , incubation of the chromosomes with the recombinant enzyme resulted in the PHOSphorylation of a number of proteins , notably giving rise to a strong band at the approximate size of the histones . Distinct bands were observed that indicated SUBstrates with apparent molecular masses of approximately 16 , 20 , 22 , 32 , 39 , 52 , 56 , 62 , 76 , 105 and 170 kDa (Fig 4B) . That these were not the same as those found following the incubation of the chromosomes with radiolabel but without recombinant enzyme indicates that Aurora B is not the only kinase in the fraction and that the recombinant enzyme we prepared has a restricted specificity for its activity . Next , to identify candidate subSTRates for Aurora B , we performed one-dimensional SDS-PAGE and excised bands that corresponded in size to those PHOSphorylated in parallel experiments ( Fig 4B ) . MALDI-TOF analysis of these bands identified the proteins listed in Table 2 . MALDI-TOF analysis of a negative control region , which contained no distinct bands , gave a complex mixture of peptides and no protein was identified . This list includes a number of potentially relevant Aurora B SUBstrates , based on intracellular distribution . It should be noted that the most common investigator-derived , artefactual keratin contaminants are keratins 1 , 2 and 10 , so that the keratins 8 and 17 found here are likely real components of the HeLa preparation . Three nuclear proteins--the GTP -binding protein NGB , topo IIalpha and CAP-C--were identified as candidate Aurora B SUBstrates in this screen , with the DEAD/H box polypeptide 30 representing a possible fourth . It should be noted that the experiment performed to identify potential Aurora B SUBstrates ( using ProFound ) was carried out at the beginning of our study , separately to the work used to generate the bulk of the data in Table 1 ( using MS-Fit , generating MOWSE scores ) , so that the peptides identified in the screen in Table 2 are clearly not the only components of the bands we examined . Since it is known that topo IIalpha is a phosPHOprotein ( 49-54 ) , we then tested whether it could serve as an in vitro SUBstrate for Aurora B . Recombinant topo IIalpha and recombinant Aurora B were co-incubated in the presence of [gamma-32P] ATP and we found that Aurora B did indeed PHOSphorylate the topoisomerase , as well as itself ( Fig 5 ) . As a control for the specificity of this PHOSphorylation , we incubated topo IIalpha with a recombinant kinase -inactive Aurora B in which a key active site residue was mutated from lysine to arginine (7) . This protein failed to PHOSphorylate topo IIalpha ( Fig 5 ) , confirming that this is an activity derived only from functional Aurora B . It is noteworthy here that the Aurora B activity was very dependent on buffer choice , even though the kinase was active on chromosomes to essentially the same extent in either Tris - or HEPES-containing buffers ( data not shown ) . This may reflect the likely presence of co factors necessary for optimal activity of the enzyme , e.g . INCENP , in the chromosome preparation , which are not available to the reaction with recombinant protein . Attempts to localise phosphorylaTION sites using Fe (III) -IMAC and electrospray tandem mass spectrometry ( 55 ) revealed one endogenously PHOSphorylated residue in the recombinant topoisomerase II . No Aurora B phosphorylaTION sites in topo IIalpha were identified , due perhaps to the low efficiency of the PHOSphorylation activity and to the large size and complexity of the topoisomerase molecule ( ie , there are approximately 200 serines and threonines in the peptide sequence ) . Since there is , as yet , no clear consensus site for the Aurora B kinase , it has been difficult to further explore this observation . In order for the in vitro data to have any significance in vivo , topoisomerase II and Aurora B must be able to interact during mitosis . Both are known to be nuclear proteins during mitosis and to examine the relationship between them we performed immunofluorescence localisation experiments . As shown in Figure 6 , topo IIalpha is located along the axis of the chromosomes , while Aurora B is centromeric , as expected . These findings show that atleast a subfraction of the topo IIalpha population is available for phosphorylaTION by Aurora B and , given recent evidence for the mobility of the topoisomerase ( 56,57 ) , suggests that Aurora B may be able to phosphorylaTE a significant amount of this SUBstrate during mitosis .