C9ORF78 partially localizes to centromeres and plays a role in chromosome segregation

Abstract

C9ORF78 is a poorly characterized protein found in diverse eukaryotes. Previous work indicated overexpression of C9ORF78 in malignant tissues indicating a possible involvement in growth regulatory pathways. Additional studies in fission yeast and humans uncover a potential function in regulating the spliceosome. In studies of GFP-tagged C9ORF78 we observed a dramatic reduction in protein abundance in cells grown to confluence and/or deprived of serum growth factors. Serum stimulation induced synchronous re-expression of the protein in HeLa cells. This effect was also observed with the endogenous protein. Overexpressing either E2F1 or N-Myc resulted in elevated C9ORF78 expression potentially explaining the serum-dependent upregulation of the protein. Immunofluorescence analysis indicates that C9ORF78 localizes to nuclei in interphase but does not appear to concentrate in speckles as would be expected for a splicing protein. Surprisingly, a subpopulation of C9ORF78 co-localizes with ACA, Mad1 and Hec1 in mitotic cells suggesting that this protein associates with kinetochores or centromeres. Levels of C9ORF78 at the centromere/kinetochore also increased upon activation of the mitotic checkpoint. Furthermore, knocking-down C9ORF78 caused mitotic defects. These studies uncover novel mitotic function and subcellular localization of C9ORF78.

SUMMARY:

C9ORF78 regulates chromosome segregation.

Introduction

Multiple mechanisms ensure high fidelity of chromosome segregation to create viable daughter cells. Since attachment of chromosomes to the spindle involves random search and capture, checkpoint and surveillance mechanisms exist to correct errors in attachment (1–4). Errors that go uncorrected before cell division result in aneuploidy or cell death. The mitotic checkpoint coordinates chromosome segregation by inhibiting anaphase onset until all chromosomes attain bipolar attachment to the spindle (4). Unattached kinetochores recruit checkpoint proteins Mad1 and Mad2 which catalyze the formation of Mitotic Checkpoint Complex (MCC) an inhibitor of the Anaphase Promoting Complex/Cyclosome (APC/C). MCC is composed of Mad2, Cdc20, Bub3 and BubR1 and is generated as long as unattached kinetochores persist. When all kinetochores attach to the spindle, loss of MCC allows activation of APC/C, an E3 ubiquitin ligase that targets Cyclin B and Securin for destruction allowing exit from mitosis (5). Certain cancer types show clear defects in MCC components which include mutations and abnormal expression levels (6–9). These defects weaken the mitotic checkpoint resulting in abnormal chromosome segregation and aneuploidy. While the mitotic checkpoint detects unattached kinetochores, more subtle attachment defects (ex. both kinetochores attached to same spindle pole) are corrected by the chromosomal passenger complex (CPC) surveillance system (1). CPC, composed of Aurora B, Borealin, Survivin, and INCENP localizes to the inner centromere and kinetochore and destabilizes defective attachments thereby triggering the mitotic checkpoint until attachment defects are corrected (10–12). At least 100 proteins localized to kinetochores coordinate attachment to the spindle and ensure operation of the mitotic checkpoint and CPC surveillance system (13). Many details relevant to the function of the kinetochore are unknown.

Human C9ORF78 is a nuclear protein containing 289 amino acids with no obvious functional domains. Published reports focused on this protein are minimal. In one study, C9ORF78 was identified as a human cancer antigen and was given the alias HCA59. A putative yeast orthologue, TLS1 regulates splicing and while the human protein was found transiently associated with the spliceosome (14, 15), it is not known if C9ORF78 regulates splicing in animals. Here we identify a subpopulation of C9ORF78 localized to kinetochores. Suppressing C9ORF78 expression causes defects in chromosome segregation. Consistent with a cell cycle role, we also find that C9ORF78 is a serum-induced protein that is also induced by E2F1 and Myc. These studies uncover a novel mitotic function of a poorly characterized human cancer antigen.

Materials and Methods

Cell Lines and Culture Conditions.

Cell lines were cultured at 37°C in a humidified atmosphere containing 10% CO₂ in Dulbecco’s Modified Eagle’s Medium (Mediatech, Inc.) supplemented with 10% fetal bovine serum (Sigma) or calf serum (Atlanta Biologicals) and Penicillin/Streptomycin 100U/ml. WI38 (human embryonic lung fibroblast) were obtained from ATCC. HeLaM (a subclone of HeLa (10)) was a gift from Ganes Sen, Lerner Research Institute, Cleveland Clinic. HCT116 colon cancer cells was a gift from Bert Vogelstein, Johns Hopkins (16). hTERT RPE were a gift from Prasad Jallepalli, Memorial Sloan Kettering Cancer Center (17). MDA-MB-231 and MDA-MB-468 breast cancer cell lines were a gift from Song-Tao Liu, University of Toledo. NCI-H522 were from the NIH National Cancer Institute. HEK239 were a gift from Nikki Harter, Lerner Research Institute, Cleveland Clinic. REF-52 and variants expressing N-Myc or mutant p53 were a gift from Olga Chernova, Lerner Research Institute, Cleveland Clinic (18). 10T1/2 were originally obtained from ATCC. 10T1/2 variants expressing RAS, c-MYC or mutant p53 were generated by the corresponding author (19).

Plasmids and RNAi transfections.

Entry clones were prepared using C9ORF78 cDNA (DNAsu) and pENTR directional TOPO cloning kit (Invitrogen). The entry clone was confirmed by sequencing. To GFP tag C9ORF78 at N-terminus, the entry clone was subcloned into dECE-GFP. To knock-in GFP at the endogenous locus of C9ORF78, we used MMEJ assisted gene knock in using CRISPR-cas9 and PITCh system(20). Briefly to prepare repair template 20 nucleotide long microhomology region of C9ORF78 was cloned flanking EGFP-2A-Puro region in PITCh vector (upstream microhomology region- 5’ CCGCGTTACATAGCATCGTACGCGTACGTGTTTGGACTGATGACTATCATTATGAGAAGTTCAAGAAAATGAATCCCCCCGGATCCATGGTGAGCAAGGG 3’, downstream homology region- 5’ ACGCGTACGTGTTTGGAAGGCGATATTTACATCCCACTCTGCACAACTCAGTACCGTCAGGCACCGGGCTTGCG). gRNAs were designed against C9ORF78 last exon (Top- 5’ CACCGAGAAGTTCAAGAAAATGAAT 3’, Bottom- 5’ AAACATTCATTTTCTTGAACTTCTC 3’). These were cloned into pSpCas9BB-2A-Puro (Addgene) (21). To knock-down C9ORF78, 75nM ON-TARGET plus siRNA SMARTpool against C9ORF78 (020231-02) was transfected in 60% confluent 24 well plate using Lipofectamine 3000 (Invitrogen). Scrambled siRNA was used as a control. Cells were fixed/harvested 3 days post transfection.

Cloning and generation of stable cell lines:

To generate HeLaM cell line stably over-expressing GFP-C9ORF78, cells were transfected with 1 μg pDECE-GFP-C9ORF78 and a plasmid providing Hygromycin resistance. An isolated colony expressing GFP and resistant to Hygromycin was used for further experiments. To generate clones with GFP knocked-in at endogenous C terminus of C9ORF78, HeLa cells were transfected with 3 plasmids: a plasmid encoding C9ORF78 gRNA, PITCh gRNA and PITCh vector (that contains microhomology regions flanking GFP and Puromycin resistance gene). Colonies grown from single cell were screened for Puromycin resistance and nuclear GFP expression. Successful GFP tagging of C9ORF78 was confirmed by western blotting.

Western Blotting.

Cells were harvested by scraping and lysed in RIPA buffer [10 mM Tris (pH 7.4), 150 mM NaCl, 1% NP-40, 1% DOC, 0.1% SDS, 1 mg/ml aprotinin, 2 mg/ml leupeptin, 1 mg/ml pepstatin A, 1 mM DTT, 0.1M phenylmethylsulfonyl fluoride, 1 mM sodium fluoride and 1 mM sodium vanadate] for 20 minutes on ice. Insoluble debris was removed by centrifugation at 16,000 g for 20 minutes at 4°C. Equal amounts of protein for each sample (determined using BCA protein assay kit - Pierce) were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Gels were transferred to polyvinylidene difluoride membranes (Millipore), blocked in a solution containing 5% (w/v) non-fat dry milk dissolved in PBST [PBS containing 0.05% (v/v) Tween 20], and probed with antibodies to C9ORF78 (Atlas antibodies and Bethyl), α-tubulin (Sigma), Actin (Abcam), PRPF8 (Abbexa), GFP(SCBT), TERF2IP (Proteintech), POT1 (Proteintech), E2F1(SCBT) Signals were detected using horse-radish peroxidase conjugated secondary antibodies (Biorad) and enhanced chemiluminescence (Biorad).

Immunofluorescence

Cells were plated on sterilized coverslips. After appropriate treatments/transfections they were pre-extracted with permeabilization buffer [150 mM NaCl, 10 mM Tris (pH 7.7), 0.1% Triton X-100, and 0.1% BSA] for 0.5 min. Then fixed with 2% formaldehyde in phosphate buffered saline (PBS) for 10 min, followed by permeabilization buffer for 9 min. Fixed cells were blocked with PBS containing 0.1% BSA (PBSP) for 1 hr at room temperature. Cells were then stained with primary antibodies. Primary antibody dilutions were determined empirically for each antibody. Antibodies were visualized by incubating samples with Alexa-fluor-conjugated secondary antibodies (Invitrogen) 1:1000 diluted in PBSP. DNA was visualized by staining with Hoechst 33342 (Molecular Probes). Phenotypes were counted in blinded manner. Images were captured on SP8 Leica Confocal microscope or Wide field Olympus microscope. Laser intensities, Z-step size and other image acquisition settings were kept constant for a given experiment.

Image analysis.

For analysis of C9ORF78 signal post serum stimulation, only one Z plane was captured. Nuclear signal (Hoechst) was used as mask in ImageJ. At least 100 cells from each treatment were measured. Intensity of C9ORF78 and Ki67 for each nucleus was measured. To measure the C9ORF78/ACA ratio after RNAi knockdown, mitotic cells were analyzed by immunofluorescence. Images were captured on SP8 Leica confocal microscope, one Z plane was used and ROI were drawn using the ACA channel to define centromeres. Using the same ROI, raw signal intensities were measured for ACA and C9ORF78 channels. C9ORF78/ACA raw signal ratios were calculated, and frequency distributions shown. To estimate distance between C9ORF78 and Hec1 or Mad1, single confocal z-plane images were analyzed in ImageJ. First, average pixel intensity of each channel was measured along a line positioned on the axis of paired kinetochores (Supplemental Figure 3). Peaks of intensity were considered the center of the stained object (kinetochore or C9ORF78 region). Distance in μm was measured between the peaks.

Statistical methods.

Statistical analyses and drawing graphs were performed using R programming or MS excel. All results were confirmed by multiple independent experiments, quantified by students t-test. p value<0.05 were considered statistically significant. Quantitation of chromosome mis-alignment and micronuclei was conducted in a blinded manner.

GFP-Trap.

One × 10⁶ cells/10cm dish expressing unfused pGFP or GFP tagged to C9ORF78 were plated. For pulling down GFP tagged proteins, cells from 5 plates were pooled by scraping and whole cell lysates were collected as mentioned in western blotting. Manufacturer protocol (Chromotek) was used further. Briefly 25μl GFP-Trap MA beads equilibrated in dilution buffer [10mM Tris-Cl/pH7.5, 150mM NaCl, 0.5mM EDTA, protease inhibitors] were added to whole cell lysates diluted in dilution buffer. Beads were collected using a magnet, washed 3 times in dilution buffer, and analyzed by SDS-PAGE.

Adenoviral Vectors.

One 70% confluent HEK293 cells plate was infected with second-generation recombinant adenoviruses. Generated viruses were collected 2–3 days later from both the supernatant and adherent cells by three freeze-thaw cycles. HEK293 cells were seeded in 96 well plate and next day, the virus suspension was diluted at 1:100. Later, 10μl virus supernatant was added to cells in 90μl culture media. To calculate virions, further 10 fold serial dilutions were done. One week later, cytopathic effect (CPE) was tested based on cell morphology. Alternatively, viral titres were determined using immunofluorescence using antibodies against hexon protein. Infected cells were visualized using an EVOS microscope after 3X PBSP wash. To test the effect of E2F1 on C9ORF78, 0.25×10⁵ WI38 cells were plated in 6cm dishes in 10%FBS DMEM medium. Next day, cells were serum starved and incubated for 3 days. Viruses at indicated MOI were added. Cells were washed with PBS after overnight incubation with viruses, followed by addition of starvation medium. Cells were harvested for western blotting analysis 48 h after adding viruses.

Data Availability

All datasets, cell lines and reagents are freely available to the research community upon request.

Results

Regulation of C9ORF78 expression.

C9ORF78 (also known as HCA59 and TLS1) is found in diverse eukaryotes including yeast, plants, animals, and various protist orders indicating a relatively ancient origin (Fig. 1). However, the function of this protein in human cells has not bee clearly defined. In a previous study, C9ORF78 was identified as a cancer antigen and although it was reported to be overexpressed in cancer cell lines, these data were not shown. In a limited survey, we observed that C9ORF78 was significantly overexpressed in breast cancer cell lines MDA MB 231, MDA MB 468, lung cancer cell line NCI H522, colon cancer HCT116 and cervical cancer HeLaM compared to normal retinal epithelial cells (RPE) and normal WI38 fibroblasts (1.2–20X fold range, pvalue <0.05, Fig. 2). A genome-wide CRISPR screen suggested that C9ORF78 may be required for efficient growth of cancer cells in 3D culture (22). In a preliminary screen, we observed smaller colonies in HCT116 cultures transiently transfected with hC9ORF8 gRNAs. Also, we were unable to obtain clones of HeLaM cells lacking expression of C9ORF78 using CRISPR knock-out, potentially consistent with a role in proliferation.

Figure 1. Supplemental Figure: Molecular Phylogenetic analysis of C9ORF78 by Maximum Likelihood method.

The evolutionary history was inferred by using the Maximum Likelihood method based on the JTT matrix-based model (38). The tree with the highest log likelihood (−2543.4299) is shown. The percentage of trees in which the associated taxa clustered together is shown next to the branches. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The analysis involved 17 amino acid sequences. All positions containing gaps and missing data were eliminated. There were a total of 74 positions in the final dataset. Evolutionary analyses were conducted in MEGA³. Sequences were obtained via NCBI-BLAST search. No orthologue was detected in alveolates (Euglena, giardia, trypanosoma).

Figure 2. C9ORF78 overexpression in malignant transformed cell lines as compared to normal cells.

Cell lines indicated were analyzed by western blotting. All the cell types were plated at 100,000 cells/5cm plate in DMEM+10%FBS and harvested on day 3 after plating. Average results of triplicate experiments are shown in the bar graph. p values from a student’s t-test are indicated.

To further analyze C9ORF78 expression, we created an N-terminal GFP fusion and established stable clones in HeLa cells. We observed that GFP-C9ORF78 fluorescence was diminished in crowded cultures. Addition of fresh medium containing serum caused a gradual but consistent increase in GFP-C9ORF78 fluorescence (Fig. 3A). Image analysis of individual cells indicated maximal fluorescence at ~14 hours post medium change (Fig. 3B). Fluorescence imaging of interphase cells indicated that most GFP-C9ORF78 was localized to the nucleus (Fig. 3A). Similar results were obtained using immunofluorescence to detect the endogenous protein (Fig. 3C). Inducing DNA damage with etoposide had no apparent effect on C9ORF78 localization in HeLa cells (Fig. 3C). Next, we used immunofluorescence to measure the behavior of endogenous C9ORF78 in response to serum stimulation. Image analysis indicated low levels of staining in serum starved cells and an increase after serum induction (Fig. 3D). Serum induction was more dramatic in normal RPE cells compared to HCT116 colon cancer cells. Serum induction of C9ORF78 broadly paralleled the elevation of Ki67 staining to mark proliferating cells. The two antigens were correlated in RPE (R²=0.69) but not HCT116 cells (R²=0.26) after addition of 20% serum.

Figure 3. Stimulation of C9ORF78 expression by serum.

HeLa cells stably expressing GFP-C9ORF78 were grown to confluence and analyzed by time-lapse fluorescence imaging. Imaging was carried out at 37°C using a sealed flask containing 10% CO₂. (A) Examples of cells showing increased fluorescence after media change. (B) Fluorescence intensity in 10 cells was quantified to determine the kinetics of recovery. (C) Homogenous nuclear localization of C9ORF78. HeLa cells were analyzed by immunofluorescence. Examples of interphase cells that were untreated or exposed to etoposide (20μM) overnight. Centromeres were detected with human anti-centromere antibody. DNA was visualized with Hoechst 33342. (D) Endogenous C9ORF78 expression in response to serum stimulation. HCT116 or RPE cells were analyzed by quantitative immunofluorescence. Cells were grown for 3 days in low (0.5%) serum. New media containing either 0.5% (black dots) or 20% serum (red dots) was added and cells were fixed 48 hours later. Immunofluorescence staining was carried out using antibodies to C9ORF78 and the proliferation marker Ki67. Images were captured, and nuclear areas defined using ImageJ based on Hoechst staining. C9ORF78 and Ki67 intensity was then determined in the nuclear area. In the top panels, each dot represents one cell. Whisker plots are shown in the bottom panels.

Endogenous C9ORF78 was robustly induced by serum in the normal human diploid fibroblast cell strain WI38. Induction of C9ORF78 in WI38 was evident by western blotting and immunofluorescence (Fig. 4A, Supplemental Figure 1). Genome-wide ChIP sequencing suggests that C9ORF78 may be regulated by E2F, Myc and Elf1 (23, 24). These transcription factors are also upregulated in response to serum stimulation. WI38 cells were starved so as to cause exit from cell cycle and hence suppress endogenous expression of E2F1. Overexpressing E2F1 using a recombinant adenovirus caused dose-dependent upregulation of C9ORF78 in serum starved WI38 cells (Fig. 4A, B, C). Next we analyze REF52, an immortalized, contact-inhibited rat embryonic fibroblast cell line (25). Derivatives of REF52 over-expressing N-Myc showed higher levels of C9ORF78 as compared to parental REF52 cells (Fig. 4D, E)(18). We also analyzed REF52#141 cells which express a dominant-negative mutant of p53. Mutant p53 expression had no significant effect on C9ORF78 expression (Fig. 4D and E). Similarly, overexpression of wild type p53 or p21 using either inducible systems or with recombinant adenoviruses had no significant effect on C9ORF78 expression (Supplemental Figure 2).

Figure 4. Regulation of C9ORF78 by serum, E2F1, and Myc.

Western blotting was used to measure endogenous C9ORF78 in either WI38 cells or nontransformed rat cell line REF52. (A) WI38 cells were either induced by serum stimulation or infected with recombinant adenoviruses at the indicated multiplicities of infection (MOI). Adenovirus with no transgene (CMV) was used as a control for virus expressing E2F1. E2F1 overexpression is shown in (B). (C) C9ORF78 levels detected by western blotting were quantified in three independent experiments. *p< 0.05 versus CMV (D and E) Parental REF52 cells or clones overexpressing either N-Myc or the C141Y dominant-negative p53 mutant were analyzed by western blotting. Example of western analysis is shown in (D) and quantitation of triplicate experiments is shown in (E). Average and standard deviations are shown. *p< 0.05 versus REF52. (F) C9ORF78 expression in mouse fibroblasts transformed with combinations of RAS (R), c-MYC (M) or mutant p53 (P). Cells were incubated in 0.5% FBS for 4 days and then analyzed by immunofluorescence using antibodies against C9ORF78 and cell cycle protein RPA. Average nuclear pixel intensity of both antigens was measured using ImageJ with the nucleus defined as the region of Hoechst 33342 staining.

Next, we analyzed mouse fibroblasts induced to undergo neoplastic transformation by expression of combinations of H-RAS^G12V (R), c-MYC (M) and mutant p53^R193P (P)(19). Compared to non-transformed parental 10T1/2 cells, RP6 cells showed lower levels of C9ORF78 while RM5 and RMP6 cell lines showed higher levels of this protein (Fig. 4F). Thus, in this mouse system, high expression of C9ORF78 correlates with overexpression of c-MYC. C9ORF78 expression was also correlated with cell cycle protein RPA; this effect was most evident in RMP6 cells (Fig. 4F). Together, these observations indicate that E2F1 and Myc regulation at least partly explains upregulation of C9ORF78 by serum. However, transcriptional induction is unlikely to explain elevation of GFP-C9ORF78 which is transcriptionally driven by a CMV promoter in the transfected plasmid (Fig. 3).

Localization of C9ORF78 in mitosis.

We analyzed asynchronously growing HeLa cells using immunofluorescence with antibodies to detect the endogenous protein. Confocal imaging revealed dispersed cytoplasmic staining during mitosis but also foci that coincided with anti-centromere (ACA) antibody staining (Fig. 5). Cells were pre-extracted with 0.1% triton X-100 for 0.5 minutes before fixing with formaldehyde followed by standard immunofluorescence staining. Cells were co-stained with antibodies to either Ndc80 or Mad1 to detect kinetochores after which confocal images of metaphase or prometaphase cells were obtained. Under these conditions, C9ORF78 colocalized with Ndc80 and Mad1 indicating that this protein localizes to the kinetochore (Fig. 5A, B and C). Foci of C9ORF78 staining that do not colocalize with centromere/kinetochore markers were also observed (Fig. 5C).

Figure 5. Kinetochore localization of C9ORF78.

Asynchronously growing HeLa cells were pre-extracted with triton X-100, fixed and analyzed by immunofluorescence using a polyclonal C9ORF78 antiserum (9–78: Bethyl) and anticentromere antibodies (ACA: Antibodies INC). Alexafluor-conjugated secondary antibodies were from Invitrogen. DNA was visualized with Hoechst 33343. Confocal microscopy was used to detect fluorescent signals. Single z-planes are shown. Cells were also stained with either (A) Ndc80/Hec1 (B) or Mad1 and ACA. (C) Line scans showing C9ORF78 overlap with Mad1. *indicates presence of non-kinetochore C9ORF78 foci.

To better characterize C9ORF78 localization we calculated distance to several reference antigens. For example: we calculated distance between C9ORF78 and Hec1 and between C9ORF78 and Mad2. The calculated distance in multiple images ranged from total colocalization to 0.15μm between C9ORF78 and either Ndc80 or Mad2 (Supplemental Figure 3). This suggests that C9ORF78 localization at kinetochores may be dynamic. Next, we quantified centromere-proximal C9ORF78 pixel intensity by creating regions of interest based on ACA staining and collecting image data from both antigen channels. The C9ORF78/ACA pixel intensity ratio was somewhat variable from experiment to experiment. However, in all of four independent experiments this ratio was significantly reduced in cells transfected with RNAi targeting C9ORF78 indicating that the centromere-proximal foci detected by the C9ORF78 antibodies are specific (Fig. 6).

Figure 6. C9ORF78/ACA ratio after RNAi knockdown.

HeLa or HCT116 cells were transfected with non-targeting RNAi (scrambled) or an RNAi pool targeting C9ORF78. Cells were analyzed by immunofluorescence and the ratio of C9ORF78 to ACA staining determined by measuring pixel intensities. (A) Scrambled RNAi transfected cells had a significantly higher C9ORF78/ACA ratio than C9ORF78 depleted cells in all four experiments (p<0.05). (B) Examples of centromere regions in cells tranfected with either scrambled RNAi or RNAi pools targeting C9ORF78.

Next, we tested the effect of activating the mitotic checkpoint on C9ORF78 localization. Nocodazole significantly increased C9ORF78 intensity at kinetochores (Fig. 7A, B). As expected BubR1 accumulated at kinetochores after nocodazole treatment. C9ORF78 intensity showed a weak positive correlation with BubR1 at individual kinetochores (Fig. 7B). The total amount of C9ORF78 protein expressed in HeLa cells did not change significantly after nocodazole treatment (Fig. 7C). For better visualization zoomed in images of kinetochores in nocodazole treated cells are displayed in Fig 7D. In most cases, a single focus of C9ORF78 staining was observed next to the ACA-positive region. In some cases, multiple C9ORF78 spots were observed. These images include 2 consecutive Z-planes (500nm apart) from representative kinetochores (or pairs), (Fig. 7D).

Figure 7. Depolymerization of spindle microtubules causes increase in accumulation of C9ORF78 at kinetochores.

Cells were treated with the spindle toxin nocodazole for 8h followed by 0.5h treatment of MG132 to prevent mitotic exit. Immunofluorescence staining using antibodies to C9ORF78, BubR1, and ACA was carried with DNA visualized with Hoechst 33342. (A) Examples of single planes from confocal stacks with the antigens indicated. Scale bar 5μm. (B) Pixel intensities were obtained using ImageJ software. Regions of interest were defined using the threshold tool based on ACA staining. Each dot therefore represents a centromere, or two daughter centromeres, if they are close enough to appear as one object. Seven cells exposed to either DMSO (total 240 centromeres) or nocodazole (209 centromeres) are represented in the scatter plot. Scale bar 5μm. (C) Western blot to detect C9ORF78 in cells treated with nocodazole or vehicle control (DMSO). (D) Representative kinetochore z stacks (0.5μm apart) showing localization of C9ORF78 with respect to centromeres (ACA) and BubR1. Scale bar 0.5μm.

In mitotic cells, C9ORF78 staining was not confined to the centromere. Distinct foci of C9ORF78 staining could be observed in regions without ACA staining. As one measure of overlap, we used ImageJ thresholding to identify either ACA objects or C9ORF78 objects in the same confocal image plane and then assessed overlap (Fig. 8A). In untreated HeLa cells, ~16% of ACA puncta show C9ORF78 staining 2 fold above background, while ~58% are 1.5 fold above the C9ORF78 background (Fig. 8B). In a reciprocal analysis we observed that ~20% of C9ORF78 objects show ACA staining 2 fold above the ACA background and ~23% are 1.5 fold above the ACA background (Fig. 8B). Furthermore, in a randomly selected image plane, we detect 116 C9ORF78 objects and 55 ACA objects with less than half of these object overlapping each other (Fig. 8B). Upon treatment with nocodazole, 90% of ACA regions show C9ORF78 staining 1.5 fold above background while 53% are 2 fold above the C9ORF78 background (Fig. 8B). This analysis corroborates our finding that nocodazole treatment enhances co-localization of C9ORF78 with centromeres.

Figure 8. Analysis of C9ORF78 and ACA staining overlap.

Confocal images were analyzed using the ImageJ threshold tool. Objects were identified based on ACA intensity and then C9ORF78 pixel intensity within those objects determined. Conversly, ACA intensity was determined in objects identified based on C9ORF78 intensity. (A) Examples of antigen staining and corresponding objects identified using ImageJ. (B) Relative intensity of either C9ORF78 or ACA within the objects indicated.

Chromosome mis-segregation and micronuclei formation upon C9ORF78 knockdown.

Initial observations suggested that HeLa cells transfected with RNAi targeting C9ORF78 exhibited chromosome alignment defects. To assess this effect cells were fixed and stained with antibodies to ACA, and tubulin to visualize the mitotic spindle. Sample identities were shielded so that chromosome mis-segregation could be assessed in a blinded manner. Mis-alignment was defined by the presence of chromosomes that had not congressed to the metaphase plate. C9ORF78 knockdown was associated with a significant increase in chromosome mis-segregation (Fig. 9A, B). Chromosomes that are not properly segregated at mitosis often form micronuclei in cells that progress to interphase (26). We measured the occurrence of micronuclei in at-least 200 interphase nuclei and observed ~5.0% of cells in C9ORF78 knock-down cells showing micronuclei as compared to 2.4% in mock transfected HeLa cells (Fig. 9C,D).

Figure 9. Chromosome mis-segregation and micronuclei formation upon C9ORF78 knockdown.

HeLa cells were transfected with RNAi targeting C9ORF78 (9–78) or non-targeting control (SCR). Phenotypes were quantified in a blinded manner using samples analyzed by immunofluorescence staining. (A) Example of a HeLa cell with mis-aligned chromosomes. (B) Quantification of mis-alignment phenotype. (C) Percent of cells containing micronuclei were quantified microscopically. Average and standard deviation are shown. (D) Western blotting carried out in parallel to assess C9ORF78 knockdown.

C9ORF78 interacts with splicing factor PRPF8 but its function is not strictly conserved.

In a previous study, C9ORF78 was implicated in regulating the splicing of telomere associated genes in fission yeast (27). In a proteomics screen in human cells, C9ORF78 was associated with splicing factor PRPF8 (14, 15). To investigate the splicing function of C9ORF78 in human cells, we first tested binding to PRPF8. C9ORF78 was GFP tagged at either the N-terminus or C-terminus, expressed in HeLa cells and isolated using GFP trap magnetic beads (Fig. 10A). PRPF8 was associated with both GFP-tagged forms of GFP but not GFP alone (Figure 10A).

Figure 10. Splicing function of C9ORF78 may not be conserved between fission yeast and humans.

Association with and regulation of spliceosome components by C9ORF78 was analyzed in HeLa M cells. (A) pre-mRNA processing factor 8 (PRPF8) interacts with C9ORF78 tagged with GFP at N terminus or C-terminus. GFP-C9ORF78 was isolated from cells HeLaM cells stably expressing the protein using GFP-TRAP magnetic beads. Western blotting was used to detect associated proteins. For C-terminal GFP fusion we used CRISPR and short-homology directed repair to generate a GFP-knockin at the endogenous C9ORF78 locus. (B) RNAi mediated knockdown of C9ORF78 did not significantly affect expression of either TERF2IP or POT1 in HeLa cells. HeLaM cells transfected with RNAi pools directed against C9ORF78 were analyzed by western blotting. Scrambled (SCR) pools were used as a negative control. Triplicate experiments were quantified in (C) and (D). We observed a slight elevation of POT1 however this was not significantly different. Fission yeast orthologues of both TERF2IP and POT1 are significantly decreased after TLS1 (fission yeast C9ORF78) loss of function (27).

Fission yeast homolog of C9ORF78, Tls1, is required for efficient splicing of shelterin components, rap1+ and poz1+. Wang J et al observed that downregulation of Tls1 directly affected splicing and lowered expression of rap1+ and poz1+ and hence assembly of telomeric heterochromatin (27). To test whether this splicing function is conserved in humans, C9ORF78 was knocked down in HeLa cells. Then the levels of human orthologs of rap1+ and poz1+ (TERF2IP and POT1 respectively) were assessed by western blotting. Surprisingly the levels of TERF2IP or POT1 were slightly, but not significantly, elevated suggesting splicing regulation of these targets by C9ORF78 is not a conserved function (Figure 10B,C and D).

Discussion

C9ORF78 is a poorly characterized protein with an ancient evolutionary origin. Orthologues exists in most of the eukaryotic kingdoms (NCI-BLAST). Limited studies suggest a potential role in splicing or regulation of ubiquitination depending on the species under consideration. For example, a potential plant orthologue, CSU2, affects photomorphogenesis by regulating the E3 ligase COP1 (28). On the other hand, the putative orthologue in fission yeast, named TLS1, regulates the splicing of rap1+ and poz1+, components of the shelterin complex that protects telomeres (27). Also, C9ORF78 was identified as a substoichiometric component of complex C of the spliceosome, however, the functional relevance of this association was not investigated (14). Using GFP-trap pull-down in cells expression GFP-C9ORF78, we confirmed that C9ORF78 associates with the complex C protein Pre-mRNA processing Factor 8 (PRPF8). However, C9ORF78 did not significantly affect the expression levels of shelterin components as observed for the fission yeast orthologue. Therefore, if C9ORF78 regulates splicing, its targets are yet to be defined.

In humans, there is evidence that C9ORF78 may be important for proliferation. For example, C9ORF78 was previously named hepatocellular antigen (HCA)59 based on the finding of antibodies in patients with that particular cancer (29). In that study, HCA59 was reported to be overexpressed in several cancer cell lines, a finding we have also corroborated. Furthermore, several genome wide CRISPR-Cas9 screens suggest that C9ORF78 could be involved in cell proliferation and oncogenesis (22, 30, 31). Our studies uncover additional aspects of C9ORF78 regulation consistent with a role in proliferation, particularly during mitosis. Whether this cell cycle function is related to regulation of splicing or ubiquitination is not known.

Genes known or likely to be involved in proliferation show distinct upregulation after serum stimulation of quiescent cells. Upregulation of genes involved in cell cycle progression such as Cyclins, CDKs, DNA replication factors and several transcription factors after serum stimulation is well documented (32). Our studies show that C9ORF78 is a serum-induced protein which is at least partly explained by E2F1 and Myc-dependent regulation. Immunofluorescence analysis showed that C9ORF78 was dispersed throughout the nucleus during interphase. However, in mitosis, a population of C9ORF78 was observed proximal to centromeres/kinetochores. This centromere/kinetochore pool of C9ORF78 was elevated when the mitotic checkpoint was activated by exposing cells to nocodazole. Knocking down expression of C9ORF78 increased the frequency of mitotic cells with non-congressed chromosomes. Also, a higher percent of interphase cells contained micronuclei after C9ORF78 knock-down, which might have arisen due to defects in mitosis.

There are numerous reports linking various aspects of RNA metabolism to kinetochore function. For example, transcription of ncRNAs at the centromere is essential for proper kinetochore assembly and function (33). In one study, splicing of these ncRNAs was also implicated in kinetochore function (34). Indeed, several components of splicing machinery localize to mitotic structures such as spindle, kinetochores and spindle poles (35, 36). Furthermore, depletion of PRFP8 caused defects in splicing of transcripts of several crucial proteins required for a faithful mitotic division. Mitotic defects such as delayed mitosis, mis-segregated chromosomes, anaphase chromatin bridges and multipolar divisions were observed in U2OS cells as a result of alternate splicing in PRPF8 depleted cells (37). Therefore, it is possible that C9ORF78 localizes to kinetochores to regulate splicing of ncRNAs. Alternatively, the role of C9ORF78 at kinetochores maybe independent of its interaction with complex C of the spliceosome. In summary, we provide evidence for a novel role of C9ORF78 in chromosome segregation and uncover a kinetochore-localized population of the protein that may fulfill this role.

ABBREVIATIONS:

ACA

Anti-centromeric antibody

APC/C

Anaphase promoting complex/cyclosome

BSA

Bovine serum albumin

CENP-A

centromeric protein A

CPC

Chromosome Passenger Complex

INCENP

Inner centromere protein

MCC

mitotic checkpoint complex

PBS

Phosphate buffered saline

C9ORF78

human Chromosome 9 Open Reading Frame 78

TERF2IP

Telomeric Repeat binding Factor 2 Interacting protein

POT1

Protection Of Telomeres 1

PRPF8

Pre-mRNA processing Factor 8