Nuclear envelope structural defects cause chromosomal numerical instability and aneuploidy in ovarian cancer

The current study utilized archived tumor tissues and cell lines prepared from human ovaries obtained from prophylactic surgeries. The tissues and cells lines were obtained from Fox Chase Cancer Center tumor bank. These tissues are designed for research only and contain no link to personal information of the donors. The study was reviewed by the Institutional Review Board (IRB) of both Fox Chase Cancer Center and the University of Miami Miller School of Medicine, and was approved and judged as non-human subject research.

The three ovarian tumor tissue microarrays (duplicate core of 120 tumor tissues and 5 controls) and 20 prophylactic oophorectomies were provided by the Tumor Bank of Fox Chase Cancer Center. These ovarian tissues and tumor specimens were described in more detail previously [40, 41]. Immunostaining was performed using the mouse DAKO Envision TM⁺ System and the Peroxidase (DAB) Kit (Dako Carpinteria, CA) as previously described [24‐26]. Negative controls were performed by replacing the primary antibodies with non-immunized IgG.

Human ovarian surface epithelial (HOSE) cells were established from ovaries obtained from prophylactic oophorectomies [41]. The donors were usually individuals who were either BRCA1/2 carriers or had a family history of breast or ovarian cancer but there was no indication of cancer at the time of surgery. Each fresh specimen of intact whole ovary from surgeries was immersed in medium and the ovarian surface was gently scraped with a rubber policeman to collect cells. The ovarian tissues were then subjected to histology analysis by pathologists to confirm the absence of microscopic tumors. HOSE cells were transfected with SV40 T-antigen to prepare human "immortalized" ovarian (HIO) cells. These HIO cells have a longer lifespan in culture and can be cultured for 20 to 50 passages before undergoing senescence, compared to HOSE cells that can only be maintained for 5 to 7 passages [42, 43].

Ovarian epithelial cancer cell lines were cultured in (D)MEM with 10% FBS as previously reported [42, 43]. The OVCAR lines were previously established by Thomas Hamilton [42] and the others (A1847, A2780, ES2, and SKOV-3) were obtained from American Type Culture Collection.

Probes for lamin A/C, emerin, and lamin B1 were obtained by reverse transcriptase-polymerase chain reaction (PCR) using total RNA extracts from HOSE cells at passage 2 as a template. The ascension number, sense primer (SP), and anti-sense primer (ASP) are as following: lamin B1 (BC078178): cggtg tacat cgaca aggtg (SP), cagct gttgc tgcat ttgat (ASP); BAF (BC005942); gaacc gttac gggaa ctgaa (SP), tcaaa cggat tgaaa gtgag g (ASP); and lamin A/C (NM_005572): tctgc tgaga ggaac agcaa (SP), ggtga tggag caggt catct (ASP). The PCR products were sequenced to confirm the identity.

Mouse anti-emerin, rabbit anti-p21, and goat anti-lamin B antibodies were purchased from Santa Cruz Biotechnology Inc (Santa Cruz, CA). Anti-p53 mouse IgGs were purchased from Dako (Carpinteria, CA) and mouse anti-beta-tubulin from Sigma St. Louis, MO.

siRNA oligos and shRNA plasmids directed against human lamin A/C were purchased from Santa Cruz Biotechnology and were transfected into cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Cells were retrieved 72 hours after transfection for analysis by Western blot, Northern blot, or immunofluorescence microscopy.

For sorting of live cells, 0.5 μM vybrant violet dye (Molecular Probe) was added and incubated for 30 minutes at 37°C. For fixed cell analysis, the trypsinized cells were resuspended in ice-cold ethanol/PBS (70% v/v) and kept at -20°C until ready to use. Before flow cytometry analysis, the fixed cells were resuspended in 0.5 μM vybrant violet dye or propidium iodide and RNase A solution. Cell sorting and flow analyses were performed on a LSR Fortessa driven by BD's FACS Diva software, version 6.1.3 (Becton Dickinson, San Jose, CA).

Cells were fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100 for 5 minutes, blocked with 3% BSA in PBS containing 0.1% Tween-20 for 30 minutes, incubated for 1 hour at 37°C with primary antibodies, and subsequently with secondary antibodies conjugated with AlexaFluor 488 (green fluorescence) or AlexaFluor 594 (red fluorescence) (Molecular Probes/Invitrogen, Carlsbad, CA). The nuclei were stained with Hoechst 33342 (Molecular Probes). Cells were mounted/sealed in anti-fade reagent containing 0.1 M n-propyl gallate (pH 7.4) and 90% glycerol in PBS. A Zeiss microscope and AxioCam Camera and AxioVision software 4.8 were used for image acquisition.

Chromosome number counting and fluorescence in situ hybridization (FISH) analysis were performed by the University of Miami Cytogenetic Core Facility. Briefly, the cells were exposed to colcemid (demecolcine) (0.015 mg/ml) for 12 hours, resuspended in 10 ml hypotonic buffer (75 mM KCl) for 20 minutes, and then kept in 1 ml of chilled fixative (methanol/acetic acid, 3/1). The cell suspension was dropped on wet slides and heated (70°C) overnight, dyed with Wright stain, and analyzed to count 50 metaphases for each sample in duplicate. For FISH analysis, an alpha-satellite DNA probe for chromosome X centromeres (Xp11.1-q11.1) and a probe for chromosome Y satalite III (Yq12) were used in interphase cells. The pre-labeled FISH probes were purchased from Abbott Molecular (Abbott Park, Illinois, USA).

Since the nuclear envelope structural protein emerin was found lost in a fraction (38%) of ovarian cancer [39], we examined if the expression of other nuclear envelope proteins might account for nuclear morphological deformation of emerin-expressing ovarian cancer cells. First, the expression of several nuclear envelope proteins was investigated by Northern blot in a panel of ovarian surface epithelial cells and cancer cell lines (Figure 1A). mRNA for lamin B1 and emerin was found to be generally higher in ovarian cancer and immortalized (HIO) cells than primary human ovarian surface epithelial (HOSE) cells (Figure 1A). Lamin A/C mRNA is expressed in some but is reduced in other (OVCAR2 and A2780) ovarian cancer cells. We noticed that lamin A but not lamin C mRNA is missing in HOSE981 cells, though information is not available for us to correlate with potential disease or phenotype of the individual. For proteins, emerin is often missing or reduced in ovarian cancer lines as reported previously [39], and noticeably, lamin A/C proteins are generally reduced in ovarian cancer cells (Figure 1B). It appears that emerin mRNA but not protein level is often increased in "immortalized" ovarian epithelial and ovarian cancer cells, and the expression of emerin mRNA and protein levels do not correlate. Thus, we conclude that in ovarian cancer cells emerin is lost/reduced at the protein but not the mRNA level; and lamin A and lamin C are reduced both at the mRNA and protein levels.

Next, we examined human ovarian tumors for lamin A/C expression by immunohistochemistry. Normal ovarian tissues show strong and uniform lamin A/C staining around the nuclear envelope of all ovarian surface epithelial cells (Figure 2A). Out of 108 informative ovarian carcinomas [see Additional file 1: Table S1], we found that 51 (47%) lack lamin A/C expression in epithelial tumor cells, while on the same field of the slide some stromal cells are positive, as shown in two examples (Figure 2B). Additionally, the majority of ovarian carcinomas that were catalogued as lamin A/C positive, contain a fraction of lamin A/C-negative cells intermingling with lamin A/C-positive cells [see Additional file 1: Table S1], as shown by an example (Figure 2C). Another case is shown for an area containing both normal ovarian surface epithelial cells and neoplastic cells, which shows that the loss of lamin A/C correlates closely with neoplastic transformation and an increase in nuclear size (Figure 2D). Also, emerin staining in the same area from an adjacent section becomes cytoplasmic instead of at the nuclear envelope, accompanying the loss of lamin A/C (Figure 2E). The correlation between loss of lamin A/C expression and emerin cellular distribution is consistent with the fact that lamin A is essential for nuclear envelope localization of emerin [44]. We conclude that the loss or abnormal distribution of lamin A/C occurs in a significant percentage (47%) of human ovarian cancers, and the rest show a variable degree of loss of lamin A/C expression in a fraction of tumor cells [see Additional file 1: Table S1]. Thus, in addition to the previous reports for leukemia [32] and B-cell lymphomas [33], colon cancer [34, 35], prostatic cancer [9], lung cancer [36], and gastric cancer [37], lamin A/C is also often lost or reduced in ovarian cancer.

Furthermore, the tumors grouped as lamin A/C-positive usually also contain a fraction of lamin A/C-negative cells, ranging from 10% to 40% [see Additional file 1: Table S1]. In contrast, all ovarian epithelial cells in normal tissues have strong lamin A/C staining around the nuclear envelope (Figure 2A). Thus, nearly all ovarian cancer cases have some degrees of nuclear envelope structural defect due to the loss of lamin A/C (reported here) or emerin [39] protein in a fraction to all of the tumor cell population.

Using immunofluorescence microscopy, we examined the presence of lamin A/C in individual cells in culture. In primary HOSE cells, all the cells (100%) stained strongly for lamin A/C, showing a smooth and oval-shaped nuclear morphology (Figure 3A). In all lines of cancer cells examined, lamin A/C is lost in some lines, or the expression is heterogeneous in the cell population. For example, in OVCAR5 and A2780 ovarian cancer cells, lamin A/C is detectable but weak by Western blot (Figure 1B); however, most of the cells are negative and only a small fraction of positive cells are present, as detected by immunofluorescence microscopy (Figure 3).

To explore the potential role of the loss of lamin A/C in ovarian cancer, we investigated the consequence of lamin A/C suppression in HOSE cells. Suppression of lamin A/C proteins with siRNA oligonucleotides was achieved as indicated by Western blot analysis (Figure 4A). We noticed a dramatic occurrence of cells with large and irregularly shaped nuclei two to three days after targeting lamin A/C, but not lamin B, with siRNA (Figure 4B). Nuclei with chromatin bridges, conjoined, and heart-shape appearance are common in lamin A/C-suppressed cells (Figure 4C). In several experiments using multiple HOSE cell preparations, lamin A/C down regulation consistently resulted in around 50% of the cells having large and atypical nuclear morphology. Conversely, cells transfected with control siRNA exhibited mostly smooth, round or oval shaped nuclei with less than 4% of cells having atypical nuclei. The extent of lamin A/C down regulation on nuclear size and morphology was quantified by computer-assisted imaging analysis, and the results indicated that lamin A/C down regulation caused a drastic change in nuclear shape (Figure 4D). In comparison, down regulation of emerin caused nuclear morphological deformation in around 20% of cells, consistent with our previous report [39].

The nuclear sizes of the cell population also changed upon lamin A/C suppression (Figure 4E), as determined using computer software (AxioVision software 4). The diameter of the nucleus had a range of 13 to 29 μm in HOSE cells and shifted up to 16 to 38 μm in the lamin A/C-suppressed cells (Figure 4E). The average size increased from 18 to 29 μm upon lamin A/C suppression, and the change was found statistically significant (P < 0.005). Following lamin A/C suppression in HOSE cells, the nuclear size distribution of the cell population resembles OVCAR5 ovarian cancer cells (Figure 4E). The variation in nuclear size may be caused by chromatin organization, condensation, and DNA content. OVCAR5 has 51 to 57 chromosomes per cell http://​www.​ncbi.​nlm.​nih.​gov/​sky/​skyweb.​cgi?​form_​type=​submitters[13]. Thus, the experiments showed that the loss of lamin A/C leads to atypical nuclear morphology and increased nuclear size in HOSE cells, resembling those of cancer cells. Similar to OVCAR5 cells, the increased nuclear size distribution in lamin A/C-suppressed HOSE cells was speculated to be caused by increased DNA content and changes in chromatin organization. These experiments have been repeated three times over a one-year period using three independent preparations of HOSE cells, and the reproducibility of the results was very good.

Next, we analyzed cellular DNA content in HOSE cells following lamin A/C suppression. First, we determined the DNA content of the cell populations by flow cytometry (Figure 5A). Lamin A/C-suppressed HOSE cells exhibited a slight but recognizable shift in DNA content compared to HOSE cells of both 2n and 4n populations (Figure 5A). The subtle increases in cellular DNA content upon lamin A/C suppression were comparable to those of OVCAR5 (51 to 57 chromosomes per cell) and OVCAR3 (52 to 70 chromosomes per cell) cells, performed in the flow cytometry as controls (Figure 5A). Thus, we conclude that the lamin A/C-suppressed cells achieved aneuploidy and a chromosomal distribution similar to the aneuploid ovarian cancer cells (OVCAR5 and OVCAR3).

These cells were further analyzed by FISH using an X centromere probe. Typically, the control cells showed two X-chromosome markers per cell, indicating a diploid ploidy (Figure 5B). Rare (less than one per 100 nucleus examined) tetraploid cells were observed, as shown in an example (Figure 5B). In contrast, three to four X-chromosomes per nucleus were detected in 37% of the lamin A/C-suppressed HOSE cells, as shown in examples from six images (Figure 5C). Thus, suppression of lamin A/C leads to the formation of tetraploid and aneuploid ovarian epithelial cells. The wide distribution of nuclear sizes and the broad range of DNA contents in the lamin A/C-suppressed cells is consistent with a variation of chromosomal numbers and aneuploidy of the cells; however, we were not able to determine the exact chromosomal number in each cell due to difficulty in obtaining a sufficient number of mitotic cells (discussed below).

In several attempts to analyze the exact chromosome number by cytogenetic approach, we failed to collect a sufficient number of mitotic nuclei in lamin A/C-suppressed cells, while control cells readily formed metaphase spreads. This observation suggests that suppression of lamin A/C may interfere with cell proliferation and normal mitosis. We performed flow cytometry to analyze simultaneously DNA content (Figure 6A) and bromodeoxyuridine (BrdU) incorporation as an indicator of DNA synthesis (Figure 6B) in the same cell population. Indeed, in lamin A/C-suppressed HOSE cells, 75% of the cells have DNA content higher than 2n, but only 5.1% of the cells had high BrdU incorporation. In comparison, 23% of the control cells had DNA content higher than 2n, and 15% of the cells had high BrdU uptake. These results indicate that the lamin A/C-suppressed cells were aneuploid or polyploid, and were not in S-phase or actively undergoing replication.

Although the primary HOSE cells proliferated slowly in culture (compared to cancer cells), we observed that about 12% of the control cells in culture were undergoing mitosis using beta-tubulin staining to mark the mitotic spindles (Figure 6C). In contrast, mitotic spindles were rare (< 1%) in lamin A/C-suppressed cells (Figure 6D), indicating a reduced cell division following the loss of lamin A/C and aneuploidy. It is known that the p53/p21 pathway is activated to mediate growth arrest of aneuploid cells [45]. Indeed, we found that p53 and p21 proteins were increased in lamin A/C-suppressed, aneuploid cells (Figure 6E). The induced cell cycle inhibitor p21 following lamin A/C-suppression is p53-dependent, since simultaneous suppression of both lamin A/C and p53 did not stimulate an increased p21 expression. It was reasoned that the lamin A/C-suppressed and aneuploid cells might overcome growth arrest once p53 function is lost, and future experiments are to determine the long-term consequence of lamin A/C suppression in the absence of p53.