Use of BAC TransgeneOmics for identification and characterization of mitotic protein complexes in human cells.
(A) Schematic outline of the workflow established for BAC tagging, GFP localization and tandem affinity purification (TAP)-mass spectrometry (MS) of mitotic proteins. IFM, immunofluorescence microscopy.
(B) Representative images of fixed HeLa cells, obtained by immunofluorescence microscopy with antibodies to GFP. Examples of cells are shown that stably express LAP-tagged proteins that have different locations in interphase, metaphase or telophase, as indicated in the images. Images are connected by lines that represent a subset of the localization trajectories observed for different proteins. The red and blue lines represent the trajectories of MIS12 and MCC, respectively. Dotted lines represent all observed trajectories that include centrosome localization in metaphase. Size bar, 10 μm.
(C) Time-lapse images of a mitotic HeLa cell stably expressing H2B-mCherry (red) and the CPC subunit INCENP-LAP (green). Numbers indicate time (minutes) before and after anaphase onset. Size bar, 10 μm.

Heat map showing hierarchical clustering of the localization trajectories of 102 spindle and centrosome proteins.
Immunofluorescence microscopy images of fixed HeLa cells expressing LAP-tagged proteins (listed on the y-axis) in five mitotic phases were classified into 33 staining patterns (mouse ortholog proteins were imaged unless suffixed as “_human”). The frequency with which each staining pattern was observed for each protein in each phase is represented by a square with color ranging from black (0% occurrence) to bright green (100% occurrence). The number codes for the staining patterns are as follows: 1: centrosome (PCM-like); 2: centrosome (fine dots); 3: dots near centrosome; 4: centrosome and partial microtubule (MT); 5: microtubule; 6: chromosome; 7: kinetochore; 8: cortex (whole); 9: cortex (partial); 10: dots in cytoplasm; 11: ER-like meshwork; 12: nuclear envelope; 13: nucleus; 14: dots in nucleus; 15: nucleolus; 16: whole cell; 17: spindle MT (whole); 18: spindle MT (K-fiber); 19: spindle matrix-like; 20: chromosome (axis); 21: chromosome (periphery); 22: kinetochore (sisters); 23: kinetochore (inner); 24: microtubule (whole); 25: microtubule (K-fiber); 26: spindle midzone; 27: cleavage furrow; 28: microtubule next to midbody; 29: microtubule (unclear); 30: midbody; 31: midbody ring; 32: midbody next to ring; 33: Golgi-like. The ‘status’ column shows the characterization status of each gene, as defined by literature searching in PubMed, according to the following color scheme: blue, reported to be involved in mitosis; yellow, some function reported but not in the context of mitosis; red, functionally uncharacterized. The dendrogram on the left indicates the relative similarities of the trajectories of individual proteins to the entire trajectory pattern. The clustered heat map was divided into ten subclusters based on a visual inspection of clustered localization patterns (subclusters indicated by alternating green and yellow shading). The names of subunits of previously identified complexes are boxed in different colors (Centralspindlin, light blue; CPC, magenta; γ-TuRC, dark blue; Aurora A-TPX2, bright green; chTOG-TACC3, orange; Augmin/HAUS, red; SKA, olive-green).

Combined interaction and localization map for a selection of 10 out of 107 small clusters (cluster size 2 to 20 proteins). Protein interaction data obtained by tandem affinity purification-mass spectrometry were analyzed by SFCM clustering. The resulting clusters (enclosed by orange dashed lines) were manually annotated based on literature knowledge (29) (for composition of reference complexes as described in the literature see fig. S6). The resulting curated clusters are shown as gray ellipses: complexes containing reciprocal interactions are enclosed by solid gray lines, while those without reciprocal interactions are denoted by dashed gray lines. Inter- and intra-cluster interactions are represented by dashed and solid blue lines, respectively, with the number of interactions corresponding to the thickness of these lines. Localization as determined by GFP imaging is shown in the color codes as indicated. Note that cluster and protein complexes can have interactions with several other clusters/complexes. For example, MCC co-purified and therefore co-clustered both with APC/C and with the MIS12/NDC80 cluster. The former association reflects binding of MCC to APC/C in prometaphase cells (30), from which these proteins were purified, whereas MCC association with MIS12/NDC80 may reflect recruitment of MCC components to unattached kinetochores in prometaphase (31). For an expanded ‘master map’, integrating localization and interaction data for more proteins, see fig. S8.

Characterization of APC16, a previously unknown subunit of the APC/C.
(A) Silver-stained SDS-PAGE gel showing proteins immunoprecipitated using APC16 or CDC27 antibodies from extracts of HeLa cells, cultured under logarithmic growth conditions (L), or arrested in S phase by treatment for 18 hr with hydroxyurea (HU) or arrested in prometaphase by treatment for 18 hr with nocodazole (N). Numbers on the left indicate the molecular masses of reference proteins.
(B) Phosphorimage showing the ubiquitylation of Cyclin B1 (CCNB1), catalyzed by CDC27 and APC16 immunoprecipitates. [¹²⁵I]-labelled human CCNB1 fragment (amino acids 1 to 84) was incubated with CDC27 or APC16 immunoprecipitates from logarithmically growing HeLa cells, plus E1 and E2 enzymes, ubiquitin and ATP, with or without the recombinant co-activator protein CDH1, for the times indicated, then analyzed by SDS-PAGE and phosphorimaging. CON indicates empty protein-A beads (left) and a condensin antibody immunoprecipitate (right). The asterisk marks a contaminating band present in the CCNB1 sample.
(C) Immunoblots showing the co-sedimentation of APC16 with core APC/C subunits CDC16 and APC10, following density gradient centrifugation. An extract of logarithmically-growing HeLa cells was subjected to centrifugation through a 10-30% sucrose density gradient. 28 fractions were collected and analyzed by SDS-PAGE and immunoblotting with the antibodies indicated.
(D) Three-dimensional model of the human APC/C obtained by electron microscopy (30), showing the location of APC16, as determined by antibody labeling.

Characterization of the γ-TuRC interacting proteins MOZART1 and MOZART2B.
(A) Silver-stained SDS-PAGE gel of LAP-purified TUBG1, TUBGCP3, TUBGCP6, MOZART2B and MOZART1. Numbers on the left indicate the molecular masses of reference proteins. Proteins annotated based on expected electrophoretic mobility are listed on the right. CCT, chaperonin containing T-complex.
(B) Mass spectrometry results obtained from the samples in (A), showing identification of γ-TuRC subunits plus MOZART1, MOZART2A and MOZART2B. Entries highlighted in orange are bait proteins, yellow indicates proteins whose interaction with the bait has previously been reported, a white background indicates that the interaction was previously unknown. Asterisks indicate proteins identified by a single unique peptide.
(C) Immunofluorescence microscopy images showing centrosomal localization of TUBG1, MOZART2B and MOZART1 through the cell cycle (G1-phase, G2-phase, prophase (Pro) or prometaphase (PM)). HeLa cells expressing the indicated proteins were fixed and stained with antibodies to GFP (green) and with DAPI (red).
(D) Immunofluorescence microscopy images showing localization of TUBG1, MOZART2B and MOZART1 to centrosomes and the mitotic spindle in metaphase in fixed LAP-cells stained with antibodies to GFP and PLK1. All images in this figure are maximum-intensity projections of DeltaVision stacks. Size bar, 10 μm.
(E) Immunofluorescence microscopy images of HeLa cells treated for 72 hr with 100 nM siRNA specific to either TUBG1 or MOZART1. Cells were fixed and stained with DAPI (cyan) and TUBA1A antibodies (red). Size bar, 10 μm.
(F) Immunofluorescence microscopy images of TUBG1-LAP HeLa cells treated for 72 hr with 100 nM siRNA against TUBG1 to deplete the endogenous TUBG1 and, simultaneously, transfected either with luciferase siRNA (control siRNA) or MOZART1 siRNA. Cells were fixed and stained with DAPI, GFP antibodies (to visualize TUBG1-LAP) and CEP135 (to stain centrioles). Two distinct phenotypes were observed: CEP135 foci in close proximity to each other that colocalize with faint TUBG1-GFP foci (middle row) and CEP135 foci dispersed randomly dispersed throughout the cell with no detectable TUBG1-GFP foci (bottom row). Insets in the third column are magnified five-fold to show one centrosome. Size bar, 10 μm.
(G) Histogram showing quantification of the experiment described in (E). In addition, cells containing the mouse TUBG1-LAP or the mouse MOZART1-LAP (BAC +) were treated with 100 nM siRNA specific for TUBG1 or MOZART1, respectively, and fixed and stained as in E. C, control, TG, TUBG1 siRNA, M1, MOZART1 siRNA.
(H) Histogram showing quantification of the experiment described in (F). The number of cells with GFP positive (TUBG1 +) or negative (TUBG1 −) centrosomes was counted. The GFP-negative cells are a combination of both MOZART1-depletion phenotypes shown in (F).