The major limit to plant growth in acid soils is the presence of toxic aluminum (Al) cations, which inhibit root elongation.Aluminum tolerance in rye has been described as controlled by, at least, four independent and dominant loci (Alt1, Alt2, Alt3 and Alt4) located on 6RS, 3RS, 4RL and 7RS chromosome arms, respectively.Previous data have shown some discrepancies about the location of Alt3 locus that should be studied.On the one hand, the rye Alt3 locus has been assigned to 4RL because it was linked to an AFLP (AMAL4) reported to be located on that chromosome.On the other hand, this locus co-segregated with an RFLP marker (BCD1230) that was also linked to the wheat Alt BH gene, located on 4DL.In our opinion these data were discrepant and require explanation.Either the map position of BCD1230 should be the first report of homology between 4DL and 4RL or there was a mistake in the assignation of Alt3 locus to 4RL.Consequently, five molecular markers (B1, B4, B11, B26 and BCD1230) that were previously described as tightly linked to the Alt3 locus on the 4RL chromosome arm were studied in wheat-rye addition lines and three F 2 populations that segregate for the Alt4 locus.Our results show unequivocally that these markers are located on 7RS and closely linked to the Alt4 locus.In spite of previous data supporting the existence of an Alt locus in 4R chromosome, our study reveals that the Alt3 and Alt4 loci are in fact the same one that is located on 7RS.
Abstract At the beginning of mammalian life the genetic material from each parent meets when the fertilized egg divides. It was previously thought that a single microtubule spindle is responsible to spatially combine the two genomes and then segregate them to create the two-cell embryo. Utilizing light-sheet microscopy, we showed that two bipolar spindles form in the zygote, that independently congress the maternal and paternal genomes. These two spindles aligned their poles prior to anaphase but kept the parental genomes apart during the first cleavage. This spindle assembly mechanism provides a rationale for erroneous divisions into more than two blastomeric nuclei observed in mammalian zygotes and reveals the mechanism behind the observation that parental genomes occupy separate nuclear compartments in the two-cell embryo. One Sentence Summary: After fertilization, two spindles form around pro-nuclei in mammalian zygotes and keep the parental genomes apart during the first division.
Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract Antigen Ki-67 is a nuclear protein expressed in proliferating mammalian cells. It is widely used in cancer histopathology but its functions remain unclear. Here, we show that Ki-67 controls heterochromatin organisation. Altering Ki-67 expression levels did not significantly affect cell proliferation in vivo. Ki-67 mutant mice developed normally and cells lacking Ki-67 proliferated efficiently. Conversely, upregulation of Ki-67 expression in differentiated tissues did not prevent cell cycle arrest. Ki-67 interactors included proteins involved in nucleolar processes and chromatin regulators. Ki-67 depletion disrupted nucleologenesis but did not inhibit pre-rRNA processing. In contrast, it altered gene expression. Ki-67 silencing also had wide-ranging effects on chromatin organisation, disrupting heterochromatin compaction and long-range genomic interactions. Trimethylation of histone H3K9 and H4K20 was relocalised within the nucleus. Finally, overexpression of human or Xenopus Ki-67 induced ectopic heterochromatin formation. Altogether, our results suggest that Ki-67 expression in proliferating cells spatially organises heterochromatin, thereby controlling gene expression. https://doi.org/10.7554/eLife.13722.001 eLife digest Living cells divide in two to produce new cells. In mammals, cell division is strictly controlled so that only certain groups of cells in the body are actively dividing at any time. However, some cells may escape these controls so that they divide rapidly and form tumors. A protein called Ki-67 is only produced in actively dividing cells, where it is located in the nucleus – the structure that contains most of the cell’s DNA. Researchers often use Ki-67 as a marker to identify which cells are actively dividing in tissue samples from cancer patients, and previous studies indicated that Ki-67 is needed for cells to divide. However, the exact role of this protein was not clear. Before cells can divide they need to make large amounts of new proteins using molecular machines called ribosomes and it has been suggested that Ki-67 helps to produce ribosomes. Now, Sobecki et al. used genetic techniques to study the role of Ki-67 in mice. The experiments show that Ki-67 is not required for cells to divide in the laboratory or to make ribosomes. Instead, Ki-67 alters the way that DNA is packaged in the nucleus. Loss of Ki-67 from mice cells resulted in DNA becoming less compact, which in turn altered the activity of genes in those cells. Sobecki et al. also identified many other proteins that interact with Ki-67, so the next step following on from this research is to understand how Ki-67 alters DNA packaging at the molecular level. Another future challenge will be to find out if inhibiting the activity of Ki-67 can hinder the growth of cancer cells. https://doi.org/10.7554/eLife.13722.002 Introduction The cell proliferation antigen Ki-67 (Ki-67 or Ki67) is constitutively expressed in cycling mammalian cells (Gerdes et al., 1983). It is therefore widely used as a cell proliferation marker to grade tumours. Ki-67 is a nuclear DNA-binding protein (MacCallum and Hall, 2000) with two human isoforms that have predicted molecular weights of 320kDa and 359kDa (Gerdes et al., 1991). The domain structure of Ki-67 is represented in Figure 1. All homologues contain an N-terminal Forkhead-associated (FHA) domain, which can bind both to DNA and to phosphorylated epitopes. The most characteristic feature of Ki-67 is the presence of multiple tandem repeats (14 in mice, 16 in human) containing a conserved motif of unknown function, the 'Ki-67 domain'. Two other conserved motifs include a Protein Phosphatase 1 (PP1)-binding motif (Booth et al., 2014) and a 31 amino acid conserved domain (CD) of unknown function, 100% identical between human and mouse, that includes a 22 amino acid motif conserved in all homologues. Ki-67 homologues also have a weakly conserved leucine/arginine rich C-terminus which can bind to DNA and, when overexpressed, promotes chromatin compaction (Scholzen et al., 2002; Takagi et al., 1999). Figure 1 Download asset Open asset Comparison of human and mouse Ki-67 structural elements. (A) Top: cartoon of human (long form) and mouse Ki-67 protein highlighting conserved elements and functional motifs. Domains are indicated by boxes (FHA, forkhead-associated domain; PP1, PP1-binding domain; CD, conserved domain; D-box: APC/C targeting destruction box motifs; KEN: APC/C-Cdh1 targeting KEN box motifs). Highly conserved regions are indicated by dotted line with percent of identical amino acids. Bottom: alignment of mouse Ki-67 repeats. (B) APC/C targeting motifs identified in human (both isoforms) and mouse Ki-67. https://doi.org/10.7554/eLife.13722.003 Ki-67 protein levels and localisation vary through the cell cycle. Its maximum expression is found in G2 phase or during mitosis (Endl and Gerdes, 2000b). In interphase, Ki-67 forms fibre-like structures in fibrillarin-deficient regions surrounding nucleoli (Verheijen et al., 1989b; Kill, 1996; Cheutin et al., 2003). Ki-67 also colocalises with satellite DNA (Bridger et al., 1998) and is found in protein complexes that bind to satellite DNA (Saksouk et al., 2014). It remains associated with nucleolar organiser regions of acrocentric chromosomes throughout interphase (Bridger et al., 1998). Ki-67 is a direct substrate of the cyclin-dependent kinase CDK1 (Blethrow et al., 2008) and is hyperphosphorylated in mitosis. This may regulate its expression and / or localisation (Endl and Gerdes, 2000a). In HeLa cells, Ki-67 binds tightly to chromatin in interphase, whereas this binding is weakened in mitosis (Saiwaki et al., 2005) when it associates with condensed chromosomes before relocating to the chromosome periphery (Verheijen et al., 1989a). Ki-67 is generally assumed to be required for cell proliferation. An early study found that injection of antisense oligonucleotides that block Ki-67 expression inhibited cell proliferation (Schluter et al., 1993). Subsequent studies have shown that unperturbed Ki-67 expression levels are required for normal proliferation rates in various cell lines (Kausch et al., 2003; Rahmanzadeh et al., 2007; Zheng et al., 2006; 2009). However, no complete Ki-67 loss of function studies have been reported. Therefore, it remains unclear what its functions are and whether it is essential for cell proliferation. The dynamic localisation of Ki-67 has led to suggestions that it could coordinate nucleolar disassembly and reassembly at either side of mitosis (Schmidt et al., 2003). Indeed, Ki-67 is required to localise nucleolar granular components to mitotic chromosomes, thereby potentially playing a role in nucleolar segregation between daughter cells (Booth et al., 2014). It has also been reported that Ki-67 has roles in ribosome biogenesis (Rahmanzadeh et al., 2007), consistent with its nucleolar localisation and apparent role in cell proliferation. In this work, we characterise the cellular roles of Ki-67 using knockdown and genetic approaches. We find that mutant mice with disrupted Ki-67 expression are viable and fertile. Preventing Ki-67 downregulation upon cell cycle exit in vivo does not impede differentiation. Thus, Ki-67 expression can be uncoupled from cell proliferation. Instead, we show that Ki-67 is an essential mediator of heterochromatin organisation and long-range chromatin interactions, controlling gene expression. As it is expressed at high levels only in proliferating cells, our results suggest that Ki-67 links heterochromatin organisation to cell proliferation. Results Mouse development is not affected by genetic up- and downregulation of Ki-67 expression Given the tight correlation between Ki-67 expression and cell proliferation, it is often assumed that Ki-67 is required for cell proliferation and that its downregulation might promote cell cycle exit. We tested these hypotheses genetically. Ki-67 protein expression is regulated during the cell cycle, and we speculated that it might be a target for the APC/C-Cdh1 ubiquitin ligase complex. This complex is active in late mitosis and G1, and triggers degradation of substrates containing D-boxes and KEN boxes. Human Ki-67 isoforms contain two or three KEN boxes, whereas mouse Ki-67 contains two D-boxes (Figure 1A,B). Mouse Ki-67 contains an additional sequence, AQRKQPSR at 2680–2687, which is highly similar to the A-box, a third APC/C-Cdh1 recognition motif (Littlepage and Ruderman, 2002). To see whether Cdh1 regulates Ki-67, we analysed mouse embryo fibroblasts (MEFs) lacking the Fzr1 gene that encodes Cdh1 (Garcia-Higuera et al., 2008). Asynchronous Fzr1 heterozygous MEFs, that are at different stages of the cell cycle, had variable Ki-67 levels, whereas in the Fzr1 knockout MEFs Ki-67 was upregulated and more homogeneously expressed (Figure 2A). To see whether sustained Ki-67 expression in quiescent cells would have a negative impact on cell cycle arrest in vivo, we analysed Fzr1-knockout mice. Here, Ki-67 was overexpressed and uncoupled from cells that incorporate BrdU in all tissues examined (Figure 2B, Figure 2—figure supplement 1). Thus, Ki-67 expression is regulated by APC/C-Cdh1 in mice and its downregulation is not a prerequisite for cell cycle exit. Figure 2 with 1 supplement see all Download asset Open asset Maintenance of Ki-67 expression in quiescent cells in vivo by Cdh1 mutation. (A) Immunofluorescence analysis of Ki-67 in MEF cells isolated from embryo (E13.5) of Fzr1(+/Δ);Sox2-Cre and Fzr1(-/Δ);Sox2-Cre mice. Scale bar, 25 µm. (B) IHC staining of Ki-67 and BrdU in sagittal sections of embryo cerebellum (E18.5) of Fzr1 (+/Δ);Sox2-Cre and Fzr1 (-/Δ);Sox2-Cre mice. Bars, 200 µm. Bars in zoom, 50 µm. https://doi.org/10.7554/eLife.13722.004 We next investigated the functional consequences of Ki-67 downregulation for normal tissue development and homeostasis. To disrupt the gene encoding Ki-67, Mki67, in the mouse germline, we used a TALEN pair targeting the unique ATG start codon. This is predicted to generate null alleles (Figure 3—figure supplement 1A). After cytoplasmic injection of these TALEN-encoding mRNAs into zygotes, 10 out of 54 mice had mutations disrupting the coding sequence. We crossed founder mutant mice, and four gave germline transmission of the mutation. Due to mosaicism, this resulted in six lines: two lines with mutations eliminating the initiation codon and four lines with deletions which cause frameshifts immediately downstream of the ATG (Figure 3A). From these, we selected a 2-nucleotide deletion (2nt∆) mutant that retains the ATG initiation codon but has a frameshift in the next codon (Figure 3—figure supplement 1B), and a 21-nucleotide deletion (21nt∆) that eliminates the ATG (Figure 3—figure supplement 1C). We crossed these mice and, unexpectedly, obtained homozygous mutants at the expected Mendelian frequency that were indistinguishable from heterozygous or wild-type (WT) littermates (Figure 3B, Figure 3—figure supplement 1D). Both deletion mutant lines showed normal growth and were fertile. Sagittal sections from Mki672nt∆/2nt∆ mice did not reveal any obvious defects in tissue morphology (Figure 3—figure supplement 2). Since the intestinal epithelium is the most highly proliferative adult mouse tissue, we compared its morphology between WT and mutant mice. In WT animals, the proliferative crypt compartment was strongly stained for Ki-67 by immunohistochemistry (IHC), while only minimal levels of Ki-67 were observed in the differentiated cells on the villus (Figure 3C, top), as expected. In contrast, in the mutants, proliferating crypt cells showed only residual levels of Ki-67 staining by IHC (Figure 3C, bottom) or immunofluorescence (Figure 3—figure supplement 3). Immunoblotting of intestinal epithelium preparations could detect a weak band of similar size to WT Ki-67 (Figure 3D; Figure 3—figure supplement 4). The signal was, however, reduced by at least 90% in both mutants compared to WT tissue. Three different Ki-67 antibodies gave similar results. These are all extremely sensitive as they recognise the highly repeated Ki-67 domain. They should also detect N-terminally truncated Ki-67 that would result from translation from the ATG at position 433. qRT-PCR analysis showed that Ki-67 mRNA level was, unexpectedly, increased rather than reduced in the intestinal tissue (Figure 3—figure supplement 5). In the intestinal epithelium, analysis of Wnt signalling and differentiation of goblet and tuft cells showed no differences between WT and Mki6721nt∆/21nt∆ mice (Figure 3—figure supplement 6). These results show that high Ki-67 levels and an intact Ki-67 gene are not required for development or differentiation in vivo. Figure 3 with 11 supplements see all Download asset Open asset Mouse development with a mutated Ki-67 gene. (A) Table describing Ki-67 mutant mouse lines resulting from germline transmission of mutations generated by cytoplasmic injection of TALEN-encoding mRNA into zygotes. (B) Macroscopic appearance of littermate female mice at 10 weeks of age. Genotypes are specified. (C) IHC staining of Ki-67 in sagittal section of intestine from Mki67WT/WT, Mki67WT/2nt∆ and Mki672nt∆/2nt∆ mice. (D) Western blots of Ki-67 and cyclin A expression from intestine isolated from Mki67WT/WT, Mki67WT/2nt∆ and Mki672nt∆/2nt∆ mice. LC, loading control. (E) Western blot of Ki-67 in MEFs from WT, Mki67WT/2nt∆ and Mki672nt∆/2nt∆ mice. LC, loading control. (F) Flow cytometry profiles in WT, Mki67WT/2nt∆ and Mki672nt∆/2nt∆ MEFs showing EdU incorporation upon a 1 hr pulse and DNA content. https://doi.org/10.7554/eLife.13722.006 To see if cells from Ki-67 mutant mice had normal proliferation capacity we isolated embryonic fibroblasts (MEFs) from day-13 embryos. Homozygous Mki672nt∆/2nt∆MEFs had at least 90% lower Ki-67 levels (Figure 3E). We could not confirm by immunoblotting whether or not the protein was full-length or truncated since SDS-PAGE cannot resolve 15kDa differences between proteins of nearly 400 kDa, and no antibodies are available against the N-terminus of Ki-67. As with intestinal tissue, the loss of Ki-67 expression was not due to mRNA degradation, as shown by qRT-PCR (Figure 3—figure supplement 7). Indeed, in the homozygous mutant, the Ki-67 mRNA level was increased to a level comparable with that of proliferating NIH-3T3 cells. Mutant MEF proliferated comparably to controls, and flow cytometric assessment of EdU incorporation after a 1 hr EdU pulse showed similar numbers of replicating cells in Ki-67 WT and mutant cells (Figure 3F). The low level residual Ki-67 expression in homozygous Ki-67 mutants suggests that TALEN or the conceptually-related CRISPR approaches may not lead to complete loss of expression, even when the translation initiation codon has been mutated. To further investigate whether Ki-67 translation can occur with a mutated initiation codon, we used the same TALEN pair to generate monoclonal Ki-67 mutant mouse NIH-3T3 cell lines, allowing analyses of translation that are technically impossible using animal tissues. We also performed the same procedure in the absence of TALENs to isolate wild-type clones. We obtained nine mutants with very low Ki-67 expression. Cloning and sequencing showed that five had biallelic mutations around the ATG codon (Figure 3—figure supplement 8). As in mice, even though Ki-67 was visible by immunofluorescence, Ki-67 was barely detectable by Western blot in all clones analysed (Figure 3—figure supplement 9A). All clones proliferated efficiently (Figure 3—figure supplement 9B). qRT-PCR showed that mutants did not have decreased Ki-67 mRNA levels compared to WT NIH-3T3 cells; indeed, like mutant MEFs, clone 14 had a higher level (Figure 3—figure supplement 9C). We selected two clones for further analysis of Ki-67 translation. Clone 14 had lost the ATG codon in one allele, but had acquired an insertion of 4 nucleotides after the ATG in the second allele, generating the same shifted reading frame as the Mki672nt∆/2nt∆ mice. Clone 21 had lost the ATG codon in both alleles, thus mimicking the situation with Mki6721nt∆/21nt∆ mice. qRT-PCR quantification of Ki-67 mRNA from ribosome purifications shows that in clones 14 and 21 there was no decrease in ribosome association with Ki-67 mRNA; indeed, in clone 14 it increased, probably due to the higher Ki-67 mRNA level (Figure 3—figure supplement 10). To definitively determine levels of Ki-67 translation in the mutants, we performed SILAC quantitative mass spectrometry from exponentially growing WT or mutant NIH-3T3 cells cultured in light (L) (WT) or heavy-labelled (H) medium (clones 14 and 21). Chromatin was purified and run on SDS-PAGE. Peptides were purified from two gel slices, one around the predicted size of full length Ki-67 (>250 kDa; band 1) and one at a smaller size (130k Da-250 kDa, band 2). Ki-67 could not be positively identified in clones 14 and 21 (Figure 3—figure supplement 11). In peptides from WT cells (L), 44 peptides derived from Ki-67 were identified by MS/MS. In contrast, in peptides from mutant cell lines (H), no MS/MS spectra for Ki-67 could be identified in either band. Selecting the ‘re-quantify’ option in MaxQuant (that forces quantitation of identified light peaks against any peaks that have the expected difference in m/z ratio), the ratios H/L observed for putative Ki-67 peaks were in the range of most typical contaminants that are only found unlabelled in a SILAC experiment. In band 1, the median normalised H/L intensity ratio of the 5 corresponding m/z peaks in clone 14 was 0.095 (mean, 0.100, SD, 0.02), and in clone 21 (7 peptides) was 0.191 (mean 0.203, SD 0.05). In band 2, which would result from truncated or degraded Ki-67, there were only 3 corresponding peaks, with median H/L ratios for clones 14 and 21 of 0.358 (mean, 0.402, SD, 0.30) and 0.500 (mean, 0.454, SD 0.33) respectively. Taken together, these results show that if Ki67 is translated in the mutant cell lines, it is at trace levels that are not positively identifiable by state-of-the-art mass spectrometry. At best, translation can occur from the mutated Ki-67 gene with an estimated 10% efficiency. The product is most likely an N-terminally truncated protein lacking the conserved FHA-domain, arising from a downstream in-frame ATG. Cells lacking Ki-67 proliferate efficiently Given the above results, it remained possible that very low levels of Ki-67 remain after Mki67 gene mutation and that they might suffice to sustain cell proliferation. To rule out this possibility we devised a 'double TALEN' strategy to completely eliminate Ki-67 expression. We designed and synthesised an additional TALEN pair targeting a sequence downstream of the translation stop codon. We co-transfected the ATG TALEN pair and the Stop TALEN pair with a GFP knock-in construct containing homology arms (Figure 4A). We thus isolated several monoclonal cell lines in which Ki-67 mRNA was essentially eliminated (Figure 4B), indicating efficient nonsense-mediated decay (NMD). These cell lines had no residual Ki-67 protein expression (Figure 4C), confirming that the basal immunostaining seen in clones 14 and 21 indeed reflected trace level Ki-67 expression. We used Southern blotting of genomic DNA to characterise these alleles. The 3’ end of the Mki67 gene remained intact in all clones, while the 5’ end of mutants showed a rearrangement consistent with a tandem insertion of multiple copies of the knockin construct upstream of the Mki67 ORF (Figure 4—figure supplement 1). Thus, the Mki67 gene was severely disrupted but not deleted. These Ki-67-negative cells proliferated normally. Growth curves and DNA content profiles were indistinguishable from controls (Figure 4D). Time-lapse videomicroscopy showed that although individual clones had slightly different cell cycle lengths, cell division time was not significantly different between WT and mutant clones (Figure 4E). We noticed that mitotic cells of one of the clones lacking Ki-67 had an altered chromosomal periphery (Figure 4F). Such a phenotype has previously been reported in Ki-67 knockdown HeLa cells (Booth et al., 2014). Nevertheless, these cells could divide efficiently. Figure 4 with 1 supplement see all Download asset Open asset Cell proliferation without Ki-67. (A) Schematic representation of strategy for TALEN-mediated generation of Mki67 null allele. (B) qRT-PCR analysis of Ki-67 mRNA levels in NIH-3T3 WT clone W4 and Ki-67-negative 60, 65, 99 clones. (C) Top: Western blot of Ki-67 and Cyclin A in NIH-3T3 WT clone W4 and Ki-67-negative mutant clones 60, 65, 99; LC, loading control; below, Ki-67 immunofluorescence; bar, 10 µm. (D) Left, growth curves of WT and Ki-67 null cell lines 60, 65 and 99; right, cell cycle distribution analysed by flow cytometry. (E) Cell cycle length of WT clone W4 and Ki-67 null clones 60 and 65 as determined by time-lapse videomicroscopy. (F) Cells of clone 65 show altered chromosomal periphery in mitosis. The Ki-67 staining is deliberately overexposed to demonstrate absence of detectable Ki-67 in clone 65, even in metaphase. Bar, 5 µm. https://doi.org/10.7554/eLife.13722.019 Next, we tested whether Mki67 mutant clones had altered kinetics of cell cycle exit or entry. To do this, we quantified cells that could replicate by measuring 5-ethynyl-2-deoxyuridine (EdU) incorporation into DNA. We found that 42% of WT Mki67 cells could still incorporate some level of EdU even after 72 hr with 0.1% serum, but only 13% or 28% of mutant clones 60 or 65, respectively, could do so (Figure 5A). This suggested that Mki67 disruption rendered cells slightly more sensitive to serum starvation. Upon addition of serum to quiescent cells, WT and Mki67 mutants entered the cell cycle with similar kinetics (Figure 5A). Ki-67 remained completely undetectable in mutants (Figure 5B). Figure 5 with 2 supplements see all Download asset Open asset Cells lacking Ki-67 enter the cell cycle efficiently. (A) Top, re-entry of cell cycle in NIH-3T3 WT clone W4 and Ki-67-negative mutant clones 60 and 65 after serum starvation-induced cell cycle arrest. Progression of cell cycle entry analysed by FACS using EdU staining. Bottom, quantification of cell cycle phases in this experiment. (B) Western blot analysis of Ki-67 (upper panel) and cyclin A2 (lower panel) upon cell cycle entry. LC, loading control. https://doi.org/10.7554/eLife.13722.021 As Ki-67 is frequently used to assess proliferation in human cancer cells, we tested whether human cells lacking Ki-67 can proliferate. We generated stable human cell lines with inducible or constitutive expression of shRNA that silenced Ki-67 or a non-silencing control (Figure 5—figure supplement 1A, 2A). We used the non-transformed human fibroblast cell line BJ-hTERT, and two commonly used cancer cell lines, HeLa and U2OS, which are of epithelial and mesenchymal origin, respectively. In BJ-hTERT, inducing shRNA expression largely prevented Ki-67 expression but had no detectable effect on the kinetics of entry into the cell cycle, as judged by expression of cell cycle regulators and EdU incorporation (Figure 5—figure supplement 1B). Further reducing residual Ki-67 levels with siRNA also did not affect cell proliferation (Figure 5—figure supplement 1C). Similarly, constitutive knockdown of Ki-67 in stable shRNA-expressing HeLa or U2OS cells had no effect on cell cycle distribution nor on the expression of cell cycle regulators (Figure 5—figure supplement 2B). Analysing single cells by immunofluorescence showed that knockdown U2OS cells with undetectable Ki-67 expression incorporated EdU in a similar manner to control cells, demonstrating efficient DNA synthesis (Figure 5—figure supplement 2C). Taken together, these results show that although Ki-67 elimination might have minor effects on cell cycle exit and mitosis, mammalian cells can nevertheless proliferate efficiently in the absence of detectable Ki-67. Ki-67 interacts with proteins involved in nucleolar processes and chromatin regulation To investigate possible molecular functions of Ki-67, we identified interacting proteins. To do this, we expressed FLAG-tagged versions of full-length human Ki-67 or an unrelated protein (TRIM39) in U2OS cells, and pulled down proteins from nuclear extracts with anti-FLAG antibody (Figure 6—figure supplement 1). These were analysed by label-free mass spectrometry. This approach identified 406 proteins specific to the Ki-67 pulldown (Figure 6—figure supplement 2). These included known Ki-67 partners: CDK1, an established Ki-67 kinase (Blethrow et al., 2008), nucleolar protein NIFK (Takagi et al., 2001), protein phosphatase 1 (Booth et al., 2014), and five subunits (HCFC1, HSPA8, MATRIN3, RBBP5 and WDR5) of a histone methylase complex that interacts with the nuclear receptor coregulator NRC (also known as NCOA6; Garapaty et al., 2009). Gene ontology and STRING analysis classified the Ki-67 interactors as being enriched in two general processes: chromatin regulation and ribosomal biogenesis (Figure 6, Figure 6—figure supplement 2). Specifically, interactors could be subdivided into groups involved in chromatin modification and transcription, ribosomal subunit biogenesis, pre-rRNA processing, protein translation, and splicing. These interactions suggested that Ki-67 might be involved not only in ribosomal biogenesis, as previously suggested (Rahmanzadeh et al., 2007), but also in regulating chromatin. Among chromatin regulators interacting with Ki-67, we found the NRC-interacting methylase complex; KMT2D, ASH2L and SUZ12 proteins which are components of MLL and PRC2/EED-EZH1 complexes which regulate histone H3 methylation (Patel et al., 2009; Pasini et al., 2004); SETD1A, HCFC1, HDAC2, YY1 and RCOR1, which are components of H3K4 demethylase complexes and co-repressors; and UHRF1, which binds to H3K9me3-modified chromatin and is involved in both maintaining DNA methylation and heterochromatin formation (Bostick et al., 2007; Guetg et al., 2010; Rottach et al., 2010). Ki-67 also interacts with TIP5, the major component of the nucleolar remodelling complex (NoRC) which is required to establish and maintain perinucleolar heterochromatic rDNA, as well as NoRC-interacting proteins TTF1 and DNM3. Figure 6 with 2 supplements see all Download asset Open asset Ki-67 interacts with proteins involved in nucleolar processes and chromatin. Simplified STRING analysis reveals network interactions between proteins associating with Ki-67. The full network is shown in Figure 6—figure supplement 2; data is provided in Figure 6—source data 1. https://doi.org/10.7554/eLife.13722.024 Figure 6—source data 1 Ki-67 interacting proteome. Table showing proteins identified by mass spectrometry eluted from immunoprecipitates of cells transfected with FLAG-tagged Ki-67, an unrelated protein (FLAG-tagged TRIM39) or empty vector. Proteins specifically interacting only with Ki-67 are presented on the first worksheet. Data are available via ProteomeXchange with identifier PXD003551. https://doi.org/10.7554/eLife.13722.025 Download elife-13722-fig6-data1-v2.xls Ki-67 is required for perichromosomal region formation during nucleologenesis We first focused on a possible role of Ki-67 in ribosome biogenesis, a process linked to nucleolar assembly and structure, and which is required for cell proliferation (Hernandez-Verdun et al., 2010). One of the candidate Ki-67 interactors involved in rRNA biogenesis was the Pescadillo homologue PES1, which participates in pre-rRNA processing and is localised in the nucleolar granular components (GC) (Rohrmoser et al., 2007; Tafforeau et al., 2013). During interphase, we found that Ki-67 localised at the cortical periphery of the GC, visualised using PES1. Ki-67 formed a boundary between the perinucleolar heterochromatin (clearly visible as a 'ring' in the DAPI staining) and the GC (Figure 7A, Figure 7—figure supplement 1). Whereas nucleolar disruption using Actinomycin D or the CDK inhibitors DRB or Roscovitine caused nuclear relocalisation of Ki-67 and GC proteins (Figure 7—figure supplement 2), depletion of Ki-67 did not affect the gross structure of the nucleolus, as determined by PES1 staining (Figure 7A, Figure 7—figure supplement 3). Figure 7 with 3 supplements see all Download asset Open asset Ki-67 localises PES1 to mitotic chromosomes. (A) Analysis of the interphase localisation of PES1 and Ki-67 proteins by immunofluorescence in HeLa cells 72 hr after transfection with control siRNA (scramble; Scr) or Ki-67 RNAi. Right, line scans showing the distribution of fluorescence signals within indicated nucleoli (dashed line). Images were captured in confocal mode with a spinning-disk microscope. Bar, 5 μm. (B) Analysis of the mitotic localisation of PES1 and Ki-67 proteins by immunofluorescence in HeLa cells 72 hr after transfection with control siRNA (scramble; Scr) or Ki-67 RNAi. Bar, 5 μm. https://doi.org/10.7554/eLife.13722.028 During mitosis, the nucleolus undergoes a dramatic cycle of disassembly and reassembly (Hernandez-Verdun et al., 2010). Briefly, soon after the onset of mitosis, when transcription is shut d