Depletion of the Ras Association Domain Family 1, Isoform A–Associated Novel Microtubule-Associated Protein, C19ORF5/MAP1S, Causes Mitotic Abnormalities

There is ever increasing evidence for the classification of Ras association domain family 1, isoform A (RASSF1A) as a tumor suppressor gene (1). In addition to being found to be epigenetically inactivated in most types of sporadic cancers (2, 3), Rassf1a knockout mice have an increased predisposition to spontaneous as well as induced tumor development (4, 5). However, the mechanism(s) by which RASSF1A can exert its growth-suppressive functions are not fully understood. RASSF1A can affect the cell cycle by inhibiting the anaphase-promoting complex/Cdc20 complex (6), association with the cyclin A modulator p120E4F (7), or by regulating cyclin D1 levels (8). RASSF1A can control apoptosis by MST (9, 10) or BAX-regulated pathways (11, 12). Furthermore, depletion of RASSF1A or deletion of Rassf1a is associated with changes in the cytoskeleton and increases phosphatidylinositol 3-kinase–dependent cellular motility (13). In fact, a main feature of RASSF1A expression has been its effect on the microtubules (14–16). We have recently identified C19ORF5 (also known as BPY2IP1 or MAP1S) as a novel RASSF1A-interacting protein in a yeast two-hybrid screen of a pretransformed human brain cDNA library where clones harboring C19ORF5 constituted the majority of the positive clones identified (14). The interaction of C19ORF5 and RASSF1A was also confirmed independently by other groups (17, 18). C19ORF5 is a ubiquitously expressed member of the MAP1A/B family of microtubule-associated proteins (19). The C19ORF5-RASSF1A interaction at the centrosome is thought to be required for the proper control of the anaphase-promoting complex/Cdc20 complex during mitosis (18). Overexpression of C19ORF5 can cause mitochondrial aggregation and genome destruction (20). Interestingly, this mechanism of cell death can explain the toxic accumulation of the mouse homologue Map8 in the Gigaxonin-null mice (21). C19ORF5, as well as its other interacting partner, LRP130, can bind nucleic acids although the function of this binding is not clear (22). Despite the existence of two conserved microtubule-binding regions in C19ORF5 (23, 24), the overexpressed protein exhibits diffused cytosolic staining patterns. C19ORF5 association with the microtubules becomes more evident when cells are treated with the microtubule-stabilizing drug paclitaxel or when C19ORF5 is coexpressed with RASSF1A (14, 25). Understanding the cellular functions of C19ORF5 will provide useful insights on how the tumor-suppressive actions of RASSF1A are exerted. Therefore, we used small interfering RNA (siRNA)–mediated knockdown of C19ORF5 in HeLa cells and HeLa cells stably expressing green fluorescent protein (GFP)–tagged histone H2B (26) to study the effect of C19ORF5 depletion on mitosis. We focused on mitosis because of the reported role of C19ORF5 in the regulation of RASSF1A-Cdc20 interaction (6), in addition to the importance of microtubules in this process.

Plasmid construction. Full-length C19ORF5 (NM_018174) was cloned by PCR from cDNA originating from the human glioma cell line H4 (ATCC HTB-148). Full-length coding region for C19ORF5 was subsequently cloned into the pGEM-T Easy vector (Promega, Southampton, United Kingdom) and the fidelity of the sequence was verified by Big Dye sequencing. C19ORF5 was subcloned into the HindIII-NotI sites of the pFLAG-CMV-4 (Sigma-Aldrich, Dorset, United Kingdom), which allowed the NH₂-termial tagging of C19ORF5 with a FLAG tag, FLAG-C19ORF5. The fragment amino acids (aa) 1 to 797 was generated by digestion of FLAG-C19ORF5 with BamHI and religating the larger fragment. The fragment amino acids 797 to 1,059 was generated by cloning the shorter BamHI fragment generated from above into pcDNA3.1-HA. HA-338-1059aa was generated by PCR cloning.

Cell lines and RNA interference. HeLa and HeLa-H2B-GFP (26) cell lines were kept in DMEM, 10% FCS, l-glutamine-penicillin-streptomycin solutions and were routinely maintained at 37°C and 5% CO₂. The double-stranded siRNA oligonucleotides targeting C19ORF5 were purchased from Ambion [Huntingdon, United Kingdom; Silencer predesigned siRNA-1 (ID 26322) and siRNA-2 (ID 45347)]. Transfection was done using the Oligofectamine reagent (Invitrogen, Paisley, United Kingdom) according to the manufacturer's recommended protocol for HeLa cells. Plasmid transfections were achieved using Lipofectamine 2000 (Invitrogen).

Antibodies, immunofluorescence, and immunoblotting. The mouse monoclonal antibody (mAb) against C19ORF5 (mAb4G1, 1:5,000 for Western blotting and 1:400 for immunofluorescence) was purchased from A&G Pharmaceuticals (Columbia, MD; ref. 20). Mouse mAbs against α-tubulin (DM1A), FLAG (M2 and M5), hemagglutinin (HA), and rabbit polyclonal anti–γ-tubulin were from Sigma-Aldrich. Rabbit polyclonal anti-HA antibody was from Santa Cruz Biotechnology (Y-11; Santa Cruz, CA). Rabbit polyclonal antibodies to pericentrin and α-tubulin were from Abcam (Cambridge, United Kingdom). Human anticentromere antibody was purchased from Antibodies, Inc. (Davis, CA); mouse anti–cyclin B1 and anti–cyclin A mAbs were obtained from the Cancer Research UK research antibodies service. Horseradish peroxidase (HRP)–conjugated rabbit anti-mouse immunoglobulin G (IgG) and HRP-conjugated goat anti-rabbit IgG secondary antibodies were purchased from DakoCytomation (Ely, United Kingdom). Alexa Fluor 594–conjugated rabbit anti-mouse IgG and Alexa Fluor 488–conjugated anti-rabbit IgG, as well as Alexa Fluor 488–conjugated anti-human IgG, were purchased from Molecular Probes (Paisley, United Kingdom).

Immunofluorescence. Staining of endogenous proteins was done on mitotic cells collected by shake-off before plating them on poly-l-lysine–coated coverslips. The cells were then preextracted with 0.2% Triton X-100 in 5% FCS in PHEM buffer (60 mmol/L PIPES, 25 mmol/L HEPES, 10 mmol/L EGTA, 2 mmol/L MgCl₂) for 2 min before fixation with methanol at −20°C for 10 min, followed by incubation with primary antibodies for 1 h at room temperature. The coverslips were washed extensively with PBS/0.01% Triton X-100 before incubation with the secondary antibodies for 45 min and washing. The coverslips were then mounted with VECTASHIELD containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Labs, Peterborough, United Kingdom). Images were visualized with a Photometrics SenSys KAF 1400-G2 charged coupled device fitted to a Zeiss Axioplan fluorescence microscope and captured using the SmartCapture software from Digital Scientific (Cambridge, United Kingdom). For confocal microscopy, slides were analyzed with Zeiss confocal microscope system and deconvoluted images were obtained by analysis with NIH ImageJ software.

Time-lapse video recording of mitosis. Approximately 1 × 10⁵ cells per well of control or C19ORF5-depleted HeLa-H2B-GFP cells were seeded in a 24-well plate 24 h after transfection. The next day, time-lapse video recording was done at an interval of 2 min using a microscope (AxioVert 200, Carl Zeiss MicroImaging, Inc., Welwyn Garden City, United Kingdom) equipped with a camera and a CO₂ incubator unit. Phase-contrast and fluorescent video images were visualized using the computer software attached to the microscope system (SimplePCI).

Microtubule regrowth assay. Cells were incubated with either 330 nmol/L or 10 μmol/L nocodazole (Sigma-Aldrich; overnight or 1 h, respectively) before washout. Nocodazole was washed out thrice using PBS followed by the addition of fresh DMEM, allowing the cells to recover at 37°C, 5% CO₂ for the times indicated. Microtubule regrowth was stopped by fixing the cells with methanol and immunofluorescence was done as above.

Preparation of metaphase spreads and centromeric fluorescence in situ hybridization. Four days after siRNA treatment, ∼2 × 10⁵ cells were treated for 1.5 to 3 h with media supplemented with colcemid at a final concentration of 100 ng/mL. Cells were harvested by trypsinization, washed with PBS, and one tenth of the cells were reserved for Western blotting. The remaining cells were processed for obtaining metaphase spreads as previously described (27). For fluorescence in situ hybridization (FISH), slides were dehydrated through 70%, 90%, and 100% ethanol, then hybridized with Cy3-labeled human pan-centromeric chromosome paint (CamBio, Cambridge, United Kingdom) according to the manufacturer's protocol.

Mitotic abnormalities following C19ORF5 depletion. To try and understand the function(s) of endogenous C19ORF5 in mammalian cells, we used siRNA-mediated silencing in HeLa cells to deplete C19ORF5 and observe the resulting phenotype(s). We could achieve significant reduction of C19ORF5 protein levels within 48 h of transfection using two independent siRNA oligos and using an siRNA oligo directed to the Luciferase gene as a control (Fig. 1A). Song et al. (18) have recently shown that a HeLa clone stably expressing short hairpin RNA constructs directed against C19ORF5 progress through mitosis in an accelerated and aberrant manner. To examine the effects of transient depletion of C19ORF5 on mitosis, we used HeLa cells stably expressing GFP-tagged histone H2B and recorded the process of mitosis by time-lapse microscopy (26). We have used the point of nuclear envelope breakdown as time 0 and the onset of anaphase as the end point. This is in contrast to the timing used by Song et al. (18) wherein they took late prometaphase as time 0. We have found that the majority of control siRNA–transfected cells undergo cell division within 50 to 60 min (Fig. 1B and C). However, the majority of C19ORF5-depleted cells took 4 h or longer for a small number of cells to eventually proceed to anaphase. During that time, cells could go through the nuclear envelope breakdown and prometaphase but they were unable to form a stable metaphase plate with all the chromosomes aligned to it (Fig. 1B). To confirm this observation in a different assay, we have measured the mitotic index following C19ORF5 depletion in the parental HeLa cells (Fig. 1D). The findings indicate an inverse relationship between C19ORF5 expression and the mitotic index, which correlated with the delay in progressing through metaphase transition. DNA histograms also indicate an accumulation of cells in the G₂-M phase of the cell cycle (Fig. 1E). Measurement of cyclin B and cyclin A levels in those cells show a modest increase in the levels of cyclin B following C19ORF5 depletion (Fig. 1F). To understand the molecular mechanisms underlying this observed phenotype, we analyzed the spindle microtubules following C19ORF5 knockdown. As shown in Fig. 2A, the spindles look abnormal and the microtubule fibers generally lack the organization observed in control-treated cells. Although the spindles are mostly bipolar, the poles are not well defined as in control cells. Staining the centromeres and the chromatin shows lack of condensation and compactness (Fig. 2A). When cells eventually proceed to anaphase and telophase, they do so erratically with a notable reduction or disorganization of the astral microtubules. We could not detect any significant changes in the fraction of monopolar spindles (Fig. 2B). In contrast, there was a 2-fold increase in the fraction of mitotic cells with multipolar spindles following C19ORF5 depletion (Fig. 2B). The largest change, however, was the large increase in the percentage of metaphase cells exhibiting metaphase plates with incomplete chromosome alignment following C19ORF5 depletion in HeLa cells (Fig. 2B).

C19ORF5-depleted HeLa cells were observed to have a problem in properly aligning the chromosomes to the metaphase plate, resulting in lagging chromosomes during anaphase (Figs. 1B and 2B). We therefore hypothesized that altering C19ORF5 levels may cause aneuploidy and alterations in chromosome numbers. To test this assumption, we prepared and examined metaphase spreads from control or C19ORF5-depleted HeLa cells, 96 h after transfection. The average chromosome numbers in control siRNA–treated HeLa cells were ∼64 with a median of 68 and a mode of 71. However, the average chromosome number in C19ORF5-depleted cells was 66 with a median of 68 and a mode of 72. In addition, preliminary M-Fish analysis of the metaphase spreads confirms that there is an elevated alteration in chromosomal numbers after C19ORF5 depletion. The most frequent change, however, was the increase in the incidence of sister chromatid separation (Fig. 3). In control-treated HeLa cells, chromosome arms were well resolved but the centromeres were closely associated as would be expected of normal centromeric cohesion. In contrast, chromosomes in C19ORF5-depleted HeLa cells frequently appeared as individual chromatids (Fig. 3). To confirm that these were indeed individual sister chromatids, we did pan-centromeric FISH analysis. The centromeric signals were as expected in control spreads (i.e., two closely attached chromatids). However, the centromeres from the C19ORF5-depleted cell spreads were often separated, which confirms that depletion of C19ORF5 is associated with premature separation of sister chromatids and loss of centromeric cohesion (Fig. 3).

C19ORF5 decorates the spindle microtubules. Because C19ORF5 seems to only decorate stable microtubules and its knockdown causes spindle abnormalities, we reasoned that C19ORF5 microtubule association is dynamic and requires undetermined modification of the microtubules. Therefore, we used the mAB4G1 antibody to study the localization of endogenous C19ORF5 throughout the cell cycle. First, we further validated the specificity of the antibody and mapped it to the region amino acids 388 to 797 of C19ORF5 (Fig. 4A). This is in agreement with Liu et al. (25). We have used the same antibody in immunofluorescence analysis and found that the best signal is given following methanol fixation. In interphase cells, endogenous C19ORF5 shows, similar to the overexpressed protein, a diffuse cytoplasmic staining with partial localization to the microtubules (Fig. 4B). The microtubule localization of the endogenous protein was confirmed by its disruption when treated with either nocodazole or paclitaxel (Fig. 4B). Mitotic cells, however, showed stronger staining for C19ORF5 and the protein can be clearly seen at the spindle microtubules during the different stages of mitosis (Fig. 5A). As a control, we also transfected FLAG-C19ORF5 and stained the mitotic cells with anti-FLAG antibody. As shown in Fig. 5B, FLAG-C19ORF5 also localizes to the spindle microtubules throughout mitosis (Fig. 5B). We have also determined that only the full-length C19ORF5 can localize to the spindle because deleting either the NH₂ terminus or COOH terminus disrupted this localization (data not shown).

C19ORF5 expression is required for anchoring the microtubules to the centrosome. C19ORF5 is a microtubule-associated protein that is unusual in its inability to protect the microtubules against nocodazole-induced depolymerization. To further elucidate the role of C19ORF5 in the microtubule organization, we treated control or C19ORF5-depleted HeLa-H2B-GFP cells with 330 nmol/L nocodazole for 16 h to block the majority of cells at prometaphase before washout and video recorded mitosis by time-lapse microscopy (Fig. 6; Supplementary movies). This treatment with nocodazole completely depolymerizes the microtubules as shown in Fig. 8B. Whereas the control-transfected cells recovered and proceeded to anaphase within 50 min of nocodazole washout, the majority of C19ORF5-depleted cells took significantly longer time to divide after nocodazole washout (Fig. 6). This indicated that C19ORF5 depletion may have caused a delay in microtubule repolymerization or it may have caused other related defects. To test this, we did the microtubule regrowth assay following C19ORF5 knockdown and stained the mitotic cells with antibodies against C19ORF5 and α-tubulin/γ-tubulin or pericentrin (Fig. 7). At time 0 after nocodazole washout, there was complete depolymerization of the microtubules and C19ORF5 was detected as diffused cytosolic staining (data not shown). Following recovery for 10 min, in control-treated cells, the microtubule-organizing centers, stained with α-tubulin/γ-tubulin antibodies, appear as expected at the spindle poles, which colocalize with endogenous C19ORF5 that appears as a concentrated signal at the microtubule-organizing center. In contrast, there was a considerable lack of a recognizable microtubule-organizing center in the C19ORF5-depleted cells, as indicated by a lack of C19ORF5 expression, where α-tubulin seems to be radiating from multiple sites and not from the spindle poles (Fig. 7A). As well as in mitotic cells, interphase cells treated in a similar fashion exhibited similar phenotypes (data not shown). Song et al. (18) reported that C19ORF5 depletion causes centrosome fragmentation as shown by γ-tubulin staining. We could detect similar levels of γ-tubulin fragmentation in interphase cells (data not shown). In addition, following recovery from nocodazole, fibers of γ-tubulin can be seen enveloping the chromatin instead of the spindle poles; fragments of random sizes of γ-tubulin can also be seen where C19ORF5 is depleted. To test whether C19ORF5 depletion disrupts the anchoring and organization of the centrosomes or exerts its effects through anchoring of γ-tubulin and microtubule nucleation, we stained the C19ORF5-depleted cells with antibodies against pericentrin and C19ORF5. Surprisingly, C19ORF5 depletion did not affect pericentrin localization either in untreated cells or following nocodazole washout (Fig. 7D). This may indicate that the centrosomal abnormalities seen following C19ORF5 depletion reflect γ-tubulin fragmentation and lack of stable anchoring to the centrioles.

The NH₂ terminus of C19ORF5 is required for microtubules anchoring. To establish which part of C19ORF5 protein is required for localizing to the microtubule-organizing center following nocodazole washout, we transfected HeLa cells with full-length FLAG-C19ORF5, FLAG-C19ORF5-1-797aa, HA-C19ORF5-388-1059aa, or HA-C19ORF5-797-1059aa (Fig. 8A) and did the microtubule nucleation assay as above but staining the cells with either FLAG-HA and γ-tubulin antibodies (Figs. 8 and 4A for expression levels). As shown in Fig. 8C, only the full-length C19ORF5 and amino acid 1 to 797 constructs localized to the microtubule-organizing center whereas the other two constructs did not. The amino acid 1 to 797 deletion lacks the COOH-terminal putative light chain whereas the amino acid 388 to 1,059 construct has a truncated putative heavy chain (Fig. 8A). This shows that the NH₂ terminus putative heavy chain is required for microtubule-organizing center localization and is potentially important for anchoring the microtubule-organizing center.

There is an ever increasing amount of evidence for considering RASSF1A as a bona fide tumor suppressor gene (1). It remains, however, difficult to ascertain how RASSF1A can exert its tumor suppressive functions. Understanding the functions of the RASSF1A-interacting microtubule-associated protein C19ORF5 will yield greater understanding of the RASSF1A-regulated signaling pathways and ultimately will benefit our understanding of the multiple steps of carcinogenesis. Song et al. (18) suggested that C19ORF5 (or RABP1 as they refer to it in their article) is a centrosomal protein responsible for recruiting RASSF1A to the spindle poles where it can exert inhibitory actions on the activation of the anaphase-promoting complex/Cdc20 complex and block the prometaphase to metaphase progression. The authors have also reported that depletion of C19ORF5/RABP1 accelerates mitotic progression and increases the frequency of abnormal mitosis. We extended on these observations, but instead of using short hairpin RNA–mediated stable knockdown, we have opted for the transient treatment with siRNA oligos. Using two independent siRNA oligos directed against C19ORF5 and a control oligo directed against the luciferase gene, we show that C19ORF5 depletion in HeLa cells actually delays mitosis. The C19ORF5-depleted cells are able to form the metaphase plate but that happens transiently and in an unstable fashion. The end result is an increase in the mitotic index and the frequency of metaphase cells with unaligned chromosomes. The spindle is also activated because we could detect BUBR1 at the kinetochores of C19ORF5-depleted cells (data not shown). The discrepancy between these findings and the findings reported by Song et al. (18) maybe attributed to differences between transient and stable depletion of C19ORF5. In addition, we have timed mitosis from the point of nuclear envelope breakdown to the onset of anaphase, instead of late prometaphase to anaphase as previously reported (18).

In this study, we show that C19ORF5 localization is not restricted to the centrosomes and it does decorate the spindle microtubules during mitosis. We have attempted a variety of fixation methods and cell lines, but we consistently observed that the majority of either endogenous or exogenous C19ORF5 localized to the spindle with some present in the cytosol. Additionally, we could detect C19ORF5 concentrating at the microtubule-organizing center during microtubule regrowth. We cannot exclude the possibility of centrosomal localization because the antibody used in ref. 18 maps to different regions of C19ORF5, which, as a result of potential proteolytic cleavage, may concentrate at the centrosomes. Due to the predicted multiple layers of posttranslational modifications exerted on C19ORF5, we cannot exclude the possibility that an undetectable fraction of the protein may reside at the kinetochores functioning in microtubule-kinetochores attachments.

Based on C19ORF5 localization and ability to bind to the microtubules (20, 23), we hypothesized that the mitotic problems associated with C19ORF5 depletion result from aberrant microtubule dynamics. We have found that the spindle is abnormal throughout mitosis in C19ORF5-depleted cells. In addition, we have completely depolymerized the microtubules in HeLa-H2B-GFP cells, which caused the cells to be blocked at prometaphase. The prolonged time that the C19ORF5-depleted cells took to proceed to anaphase following nocodazole washout strongly indicated that these cells are slower in repolymerizing the microtubules and forming the spindle. We do not have evidence for a slower microtubule repolymerization caused by C19ORF5 depletion. Instead, we have identified a novel role for C19ORF5 in anchoring α- and γ-tubulin to the centrosomes during microtubule nucleation and regrowth. C19ORF5, whether endogenous or overexpressed, concentrates at the microtubule-organizing centers within 1 min following nocodazole recovery (data not shown). When C19ORF5 is depleted in mitotic cells, the microtubules appear to be nucleated from multiple sites. C19ORF5 depletion does not seem to affect the centriole position at the centrosome because pericentrin does not seem to be affected, although γ-tubulin appear fragmented, giving the impression of a fragmented or duplicated centrosome.

The mechanism by which the sister chromatids separate prematurely when C19ORF5 is depleted is unknown. It could be argued that, based on Song et al. (18), C19ORF5 depletion could perturb the intricate regulation of the anaphase-promoting complex activity and therefore cause the premature degradation of securin. An alternative mechanism is that altered spindle dynamics causes changes to the spindle pulling and pushing forces on the kinetochores, thus perhaps affecting chromatids cohesion at the centromeres. It has recently been shown that depletion of the microtubule-associated proteins, anaphase-promoting complex, and EB1 increases the frequency of chromosome misalignment and lagging (28).

We have shown in this study that expression of C19ORF5 is required for mitotic fidelity, functioning as an anchoring protein for γ-tubulin and the microtubule-organizing center in mitotic cells. This adds to its function as a docking agent for RASSF1A to the centrosome. Considering the emerging active role microtubules play in apoptosis (29), as well as their established role in the spindle formation during mitosis, the RASSF1A-C19ORF5 interaction is an intriguing link between the microtubules, cell cycle control, and apoptosis. The disruption of this pathway, as evident in several types of human cancers, may have a large effect on the process of tumorigenesis. In fact, truncating mutations in C19ORF5 have recently been identified in colorectal tumor cell lines (30). Incidentally, colorectal cancers exhibit a relatively low percentage of RASSF1A methylation inactivation (1).

Grant support: Association for International Cancer Research, Sport Aiding Medical Research for Kids, Cancer Research UK, Breast Cancer Campaign, and Birmingham Children's Hospital Research Foundation (F. Latif).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Dr. G. Parsonage for help with time-lapse microscopy.