Chromosoma

, Volume 112, Issue 8, pp 389–397

Resolution of anaphase bridges in cancer cells

Authors

  • Diane R. Hoffelder
    • Department of Biological SciencesUniversity of Pittsburgh
  • Li Luo
    • Department of Biological SciencesUniversity of Pittsburgh
  • Nancy A. Burke
    • Department of Cell Biology and Molecular PhysiologyUniversity of Pittsburgh Cancer Institute
  • Simon C. Watkins
    • Department of Cell Biology and Molecular PhysiologyUniversity of Pittsburgh Cancer Institute
  • Susanne M. Gollin
    • Department of Human GeneticsUniversity of Pittsburgh Medical School
    • Department of Biological SciencesUniversity of Pittsburgh
Research Article

DOI: 10.1007/s00412-004-0284-6

Cite this article as:
Hoffelder, D.R., Luo, L., Burke, N.A. et al. Chromosoma (2004) 112: 389. doi:10.1007/s00412-004-0284-6

Abstract

Chromosomal instability is a key step in the generation of the cancer cell karyotype. An indicator of unstable chromosomes is the presence of chromatin bridges during anaphase. We examined in detail the fate of anaphase bridges in cultured oral squamous cell carcinoma cells in real-time. Surprisingly, chromosomes in bridges typically resolve by breaking into multiple fragments. Often these fragments give rise to micronuclei (MN) at the end of mitosis. The formation of MN is shown to have important consequences for the cell. We found that MN have incomplete nuclear pore complex (NPC) formation and nuclear import defects and the chromatin within has greatly reduced transcriptional activity. Thus, a major consequence of the presence of anaphase bridges is the regular sequestration of chromatin into genetically inert MN. This represents another source of ongoing genetic instability in cancer cells.

Abbreviations

GFP

Green fluorescent protein

MN

Micronucleus/micronuclei

DAPI

4′, 6-Diamidino-2-phenylindole

NPC

Nuclear pore complex

GR

Glucocorticoid receptor

UTP

Uridine 5′triphosphate

PolII

RNA polymerase II

Introduction

Chromosomal instability, defined as frequent changes in chromosome structure and number, is increasingly appreciated as a key component of tumorigenesis (Jallepalli and Lengauer 2001). A common cause and indicator of chromosomal instability is the formation of anaphase bridges—chromosomes pulled simultaneously to both poles of the microtubule spindle during chromosome segregation. Anaphase bridges were first described in maize by Barbara McClintock as a response to DNA damage (McClintock 1941). She showed that bridges resulted from fusion of broken chromosome ends leading to the formation of dicentric chromosomes that attach to both spindle poles. These bridges were observed to break under stress from the mitotic apparatus and re-form in the next interphase, giving rise to breakage–fusion–bridge cycles. In addition to DNA breaks, anaphase bridges can be caused by abnormal shortening of the telomeres (de Lange 2002; Hande et al. 1999) or by persistent chromatid cohesion (Coelho et al. 2003; Haering and Nasmyth 2003).

Anaphase bridges have been strongly linked to chromosomal instability in human tumor samples (Fouladi et al. 2000; Gisselsson 2003; Montgomery et al. 2003) and tumorigenesis in mice (Artandi et al. 2000; Rudolph et al. 2001). Recently, anaphase bridges have been suggested as a clinical diagnostic tool (Montgomery et al. 2003). At least part of the reason for the correlation between bridges and tumorigenesis may be gene loss or amplification resulting from breakage–fusion–bridge cycles altering levels of key oncogenic proteins (Daigle et al. 2001; Huang et al. 2002; Toledo et al. 1992), but other cellular consequences are also possible and are less well-documented.

Another common indicator of genomic instability is the appearance of micronuclei (MN). These small nuclear structures are observed in some types of normal tissue (Heddle 1990), but are most strongly correlated with exposure to genotoxic agents (Heddle et al. 1991; Muller-Tegethoff et al. 1997; Shelby 1988; Smith et al. 1993). Formation of MN has become one of the most widely used biomarkers for exposure to mutagens (Russo and Levis 1991) or carcinogens (Trizna and Schantz 1992; Tucker and Preston 1996). The significance of these structures can be shown by the various automated tests (Tates et al. 1990), and their optimizations and refinements (Fenech 1993; Kirsch-Volders et al. 1997; Kirsch-Volders and Fenech 2001), and internationally coordinated surveys (Fenech 1998; Surralles and Natarajan 1997). The formation of MN has become the regulatory standard for mutagenicity assessment by the United States Environmental Protection Agency (Dearfield et al. 1991).

Despite the widespread use of this biomarker, the basic biology of MN formation and the impact of MN on the cell are relatively unexplored (Jenssen 1982; Norppa and Falck 2003; Smith et al. 1993). An early indication that MN themselves may have an adverse effect on the cell was shown by studies that revealed that irradiated Chinese hamster V79 cells were less likely to form colonies in culture if they contained MN (Grote et al. 1981). Formation of MN has been proposed to contribute to the premature death of mouse embryos (Titenko and Evsikov 1977), and erythrocytes with MN are preferentially eliminated in the spleen (Schlegel and MacGregor 1984). At the cellular level, any deleterious consequences of MN are less clear. Micronuclei induced in human Burkitt lymphoma cells by X-ray exposure were found to be mostly transcriptionally inhibited, based on incorporation of tritiated UTP (Kato and Sandberg 1968), suggesting that incorporation of genes into MN would negatively impact the cell. However, more recently, MN from PtK1 cells, both isolated and in situ, were shown to be able to take up tritiated UTP at 65–70% the levels of intact nuclei (Labidi et al. 1987). Thus, the impact on the cell of having DNA in MN remains uncertain.

It is widely accepted that MN arise from the assembly of a nuclear envelope around a chromosome, or chromosomal fragment, that is separated from the main group of chromosomes at the end of mitosis. There are at least two general mechanisms by which chromatin segments could become separated. One is a defect in the spindle apparatus, such that some chromosomes fail to segregate and lag behind the rest of the chromosomes during anaphase. The second general mechanism is a structural defect in the chromosome itself, for example, the formation of acentric fragments that lack a microtubule binding site. Micronuclei have been strongly correlated with the presence of anaphase bridges (Gisselsson et al. 2001; Thomas et al. 2003; Umegaki and Fenech 2000); however, MN formation from bridges has not yet been demonstrated.

To examine the fate of anaphase bridges in cancer cells, we utilized green fluorescent protein (GFP)-tagged histone (Kanda et al. 1998) in cultured oral squamous cell carcinoma cells to image bridge resolution in real-time. Our analysis showed that in at least 50% of the imaged samples, chromatin bridges fragmented during anaphase, giving rise to MN. The MN remained distinct from the interphase nucleus. The chromatin that became encapsulated in the MN was apparently lost from the transcriptionally active gene pool, based on nucleotide incorporation and RNA polymerase II (PolII) activity. The transcriptional defect may be to due to a failure of nuclear import. Nuclear pore complexes (NPCs) were greatly reduced in number or absent in MN, and import of a GFP-tagged glucocorticoid receptor (GR) was sharply inhibited. These results reveal for the first time the frequency and significance of micronucleation occurring during anaphase bridge resolution.

Materials and methods

Cell culture, transfection and tissue

The UPCI:SCC oral squamous cell carcinoma cell lines are heterogeneous populations of keratinocytes grown from tumor tissue. Patients were not treated with chemotherapy or radiation before surgery. The cells were grown in minimal essential medium (MEM) supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 0.05 mg/ml Gentamicin, and 1% MEM non-essential amino acids (all from Gibco BRL, Grand Island, N.Y.). Diploid human fibroblast cells (GM03349B, Coriell Institute, Camden, N.J.) were cultured in MEM supplemented with 15% FBS. JAR cells were a gift from Dr. Urvashi Surti, University of Pittsburgh School of Medicine, and were grown in MEM supplemented with 10% FBS. Normal oral cells were grown in KGM-2 medium (Bio-Whittaker, Md.) as primary cultures derived from uvulopalatopharyngoplasty (UP3) specimens. All cultures were grown at 37°C with 5% CO2.

Histone H2B-GFP transfection

UPCI:SCC040 cells were transfected with pBOS-H2BGFP vector (Pharmingen) using Lipofectamine (Gibco BRL) following the manufacturer’s instructions. As transfection efficiency was less than 10%, the culture was enriched for GFP-expressing cells by selection with 5 μg/ml blasticidin S (Calbiochem) for 4 days and then 2 μg/ml for 3 weeks. After selection, cells were maintained in antibiotic-free medium with ∼75% of the cells retaining label.

Transfection of GR-GFP

The GR-GFP plasmid was a gift from Donald DeFranco (University of Pittsburgh). Cells were transfected using Lipofectamine 2000 (Gibco BRL) in OPTI-MEM (Gibco BRL).

Tumor tissue was sectioned from an oral squamous cell carcinoma, T3-size lesion (AJCC Cancer Staging Manual, 6th Edition, 2003) without lymph node metastasis. Fresh tumor tissue was spread on slides and then fixed with −20°C methanol for 5 min and allowed to air dry. (Frozen tissue was insufficiently immunoreactive to use.) The samples were then processed for immunofluorescence. Under these conditions significant nuclear fragmentation occurred and nuclear lamin staining was used to distinguish MN from chromosomal fragments.

Antibodies

The following primary antibodies were used in this study: anti-lamin B (Oncogene, NA12), anti-nucleoporins (Abcam, ab6440), anti-PolII (Covance, MMS-129R), anti-β-tubulin (gift from C. Walsh, University of Pittsburgh), anti-phosphorylated histone H2A.X (Upstate Cell Signaling Solutions, Lake Placid, N.Y.), and CREST human auto-immune serum (gift from C. Feghali, University of Pittsburgh). Secondary antibodies and stains included: Alexa Fluor488 and Alexa Fluor546 (Molecular Probes), anti-Human Cy3 (Sigma) and 4′,6-diamidino-2-phenylindole (DAPI, Sigma).

Immunofluorescent labeling

Cells were grown on glass coverslips and fixed in methanol for 20 min at −20°C. Cells labeled with anti-PolII were permeabilized in 1% Triton X-100 for 10 min at 4°C and fixed in 1.75% paraformaldehyde for 20 min at 4°C (Bregman et al. 1994). Methanol-fixed cells were rehydrated in PBS and blocked with 1% BSA, PBS. Primary antibodies were diluted in 1% BSA, PBS and incubated on cells in a humid chamber for 1 h at 37°C. Secondary antibodies were applied for 1 h at 37°C. Cells were counterstained with DAPI and then mounted onto slides using antifade mounting solution containing 1 mg/ml p-phenylene diamine in glycerol. Cells were viewed with an Olympus BX60 microscope and imaged with a Hamamatsu CCD camera controlled by QED imaging software (Pittsburgh, Pa.).

Live cell imaging

Live cell imaging was performed at the Center for Biological Imaging, University of Pittsburgh Medical Center. UPCI:SCC cells expressing H2B-GFP were maintained at 37°C, without supplemental CO2, in a heated Bioptechs perfusion chamber. To provide more consistent pH levels, 25 mM HEPES buffer was added to the medium. Time-lapse images were collected on a Zeiss Axiovert 135 microscope. Data acquisition and analysis was achieved using Metamorph imaging software (Universal Imaging, Downingtown, Pa.).

Incorporation of UTP

Chromatide Alexa Fluor488-5-UTP (Molecular Probes) was incorporated into sites of transcription as described previously (Bregman et al. 1994). Cells were grown on coverslips in their appropriate medium supplemented with 25 mM HEPES for 24 h. Fifty microliters of medium containing either 0.50 or 500 μg/ml α-amanitin (Sigma) was then incubated on cells for 30 min at 37°C. Twenty microliters of 0.05 mg/ml AlexaUTP was then added to the cells and 100 μm glass beads (Sigma) were used to permeabilize the plasma membrane (McNeil 1989, no. 1166). After 3 min, the coverslips were washed in MEM and then incubated for the indicated time intervals. Cells were fixed in 95% ethanol for 5 min at room temperature and stained with DAPI.

Nuclear import

Dexamethasone was added at 10−6 M to the medium of cells expressing GR-GFP fusion protein to induce nuclear transport. Cells were fixed in 95% ethanol for 5 min at room temperature and viewed for GFP fluorescence.

DNA damage

For H2A.X immunofluorescence, cells were fixed in 2% paraformaldehyde, PBS for 30 min at room temperature and permeabilized in 1% Triton X-100 for 5 min at 4°C. Goat serum (4%, PBS) was used for 15 min at room temperature for blocking before processing for immunofluorescence as above.

The TUNEL assay was performed using the in situ cell death detection kit (Roche, Mannheim, Germany). Briefly, cells were air dried and then fixed in 4% paraformaldehyde, PBS for 1 h at room temperature. Permeabilization in 0.1% Triton X-100, 0.1% sodium citrate in PBS was done for 2 min on ice. The TUNEL reaction mixture was prepared as described in in the manufacturer’s instructions, added to fixed cells, and incubated for 1 h at 37°C.

DNA replication

5-Bromo-2’-deoxyuridine (BrdU; 10 μM) (Aldrich) was added to cells on coverslips. After 2 h incubation, cells were fixed in 2% paraformaldehyde for 15 min. Permeabilization was then done in 0.2% Triton X-100 for 5 min. To denature the DNA, 4 M HCl was used for 10 min at room temperature. Tween 20 (0.1%), PBS washes neutralized the acid before blocking in 5% BSA, 0.1% Tween 20, PBS. Anti-BrdU AlexaFluor488 conjugate (A-21303, Molecular Probes) was incubated on the cells for 1.5 h at 37°C and then cells were counterstained with DAPI.

Results

Anaphase bridges lead to MN formation

To image anaphase bridge resolution the oral squamous cell carcinoma lines UPCI:SCC040 and UPCI:SCC103 (S.M.G., unpublished) were transfected with a plasmid expressing GFP-tagged histone H2B (Kanda et al. 1998). We first determined whether the presence of the tagged histone affected cell division. Population doubling time, mitotic frequency, and the occurrence of segregational defects were similar between the transfected and untransfected cells (data not shown), indicating that the presence of the labeled histone did not noticeably interfere with division (Kanda et al. 1998).

To determine the fate of bridges, labeled interphase cells were identified and imaged at 2–20 min intervals, for up to 48 h. Bridges were observed in 46% of the anaphase cells. This compares with 58% of anaphases in fixed non-GFP-expressing cells stained with DAPI. Of cells that formed chromatin bridges during anaphase, 45 could be followed for sufficient numbers of frames to determine the fate of the bridges. In 40/45 cases, the bridge was quickly broken during anaphase into at least two major fragments. This is consistent with the previous analysis of bridge resolution in plants (Bajer 1963). In none of the 45 cells containing an anaphase bridge was the bridge observed to be pulled intact into one of the daughter cells. The cells that formed a bridge also did not undergo cell cycle arrest while under observation. In 5/45 cells with an anaphase bridge, the bridge persisted into interphase as has been observed previously (Bajer 1963). Of the remaining 40 cells the anaphase bridge broke during mitosis. A MN formed in 70% of the cells with broken anaphase bridges (Fig. 1). Micronuclei were defined as small fragments of chromatin that remained physically distinct from the main nucleus. We distinguish MN from nuclear blebs whose interior volume appears to remain connected to the nucleus. In ∼40% of the cases, two or more MN were observed to form from a single bridge (Fig. 2). The formation of multiple large nuclear fragments (>4) was rarely observed and was assumed to result from apoptosis or injury and was not counted as micronucleation. When two MN were observed after bridge resolution, there was typically one MN in each daughter cell.
Fig. 1

Chromatin bridges often give rise to micronuclei (MN). An oral squamous cell carcinoma keratinocyte (UPCI:SCC040) expressing green fluorescent protein (GFP)-tagged histone H2B forms a chromatin bridge during anaphase. The bridge breaks and an MN (arrow) soon forms after the onset of anaphase. Time in minutes from an arbitrary start is indicated. Bar represents 10 μm

Fig. 2

Anaphase bridges can resolve into multiple fragments. In the cell shown in this figure a bridge breaks into two to three smaller fragments that remain distinct from the nucleus as the cell enters interphase. Time in minutes from an arbitrary start is indicated. Bar represents 10 μm

For further analysis of MN we continued our studies in cells from the oral cancer cell line UPCI:SCC103, which has a higher frequency of MN. If MN come from bridges breaking into more than two pieces they should often contain chromatin lacking centromeres. Immunofluorescence with autoimmune anti-kinetochore antibodies revealed that 57% (n=200) of the MN were immunonegative, consistent with chromosomal fragmentation during bridge resolution. Figure 3a,b shows a series of cells with two MN marked by arrows. Both fail to react with anti-kinetochore antibodies. However, previous studies have shown that immunolabeling of kinetochores does not always identify centromere-containing MN (Miller et al. 1991). As a further test of DNA fragmentation during MN formation we tested with immunostaining for γ-phosphorylated histone H2AX foci. H2AX is a histone 2A variant that becomes transiently phosphorylated in response to DNA double-stand breaks (Rogakou et al. 1998). Of 51 MN, 55% were immunopositive for γ-H2AX labeling. Figure 3c shows the same cells as in Fig. 3a,b and both MN are immunoreactive, indicating the presence of DNA breaks. As an additional test for chromosomal breakage, we used the TUNEL assay for DNA nicks and breaks (Gavrieli et al. 1992). Of 100 MN, 64% stained positive with this reagent, indicating the presence of broken DNA strands (Fig. 3d,e). For a positive control, cells were incubated with DNase I, resulting in labeling of all MN (not shown). These results are consistent with many but not all of the MN in these cells forming via breakage of anaphase bridges.
Fig. 3a–g

Micronuclei (MN) from anaphase bridge resolution often contain fragmented chromatin. a,d,f UPCI:SCC103 cells were fixed and labeled with 4′, 6-diamidino-2-phenylindole (DAPI). b Staining with autoimmune antibodies to kinetochores. Micronuclei are labeled with arrows. Some MN fail to react with the anti-kinetochore antibodies, suggesting the formation of centromere-less fragments, consistent with formation through DNA breaks. c,e Cells were also stained with antibodies to phosphorylated γ-H2AX (c) or subject to the TUNEL assay to detect DNA breaks (e). These results show that the MN often contain fragmented chromosomes, consistent with formation through breakage of anaphase bridges. Immunolabeling of nuclear lamins in g shows that most MN contain intact nuclear envelopes. The same field of cells is shown in a, b and c; d and e; and f and g

Micronuclei generally have transcriptional and nuclear import defects

Because MN seem to be a major consequence of bridge formation, we examined them in more detail to determine whether the chromatin in MN could still contribute to the general growth and metabolism of the cell. First, we followed these structures in real-time to see whether they rejoined the main nucleus during interphase. The MN were observed to be very dynamic and appeared to move freely in the cytoplasm relative to the nucleus (data not shown). In many cases, they moved behind the nucleus and out of view, only to re-emerge at a later time. However, at no time did we observe an MN rejoin the nucleus during interphase. Therefore, the genetic material in the MN appears to remain spatially separated from the nucleus for most of the cell cycle. Once the next mitosis began, we were unable to distinguish the chromatin in MN from the rest of the chromosomes and could not reliably determine the fate of these structures during division.

The MN are physically separated from the nucleus, but may continue to contribute transcriptionally to the metabolism of the cell. Two tests were used to determine whether the chromatin in the MN was transcriptionally active. The first was to immunolabel the cells with antibodies specific to the elongating form of the PolII holoenzyme (Patturajan et al. 1998; Fig. 4a,b). These antibodies labeled all nuclei, but only 13% (n=307) of MN. To confirm the specificity of this assay, we showed that the PolII transcriptional inhibitor α-amanitin (Ochs 1998) completely blocked the immunolabeling (Fig. 4c,d). To confirm the reduction in transcription within the MN, we tested for the incorporation of fluorescently labeled UTP into newly synthesized mRNA in situ (Fig. 5). The nucleus was brightly labeled with the added UTP, but most MN (24/25) were completely negative. (Note that the efficiency of uptake of the UTP label was ∼5% and only a limited number of MN could be assayed.) To distinguish incorporation versus passive accumulation of the UTP, we again treated the cells with α-amanitin (Fig. 5e,f). After treatment with inhibitor, only the nucleoli were labeled. This is expected since nucleolar polymerases are α-amanitin resistant (Wieland and Faulstich 1991) and shows the specificity of the UTP labeling. Thus, we believe that MN in the cell are a mixture of transcriptionally active and inactive forms, but that in the majority of cases the MN show undetectable levels of transcriptional activity. It follows that the genes contained within the inactive MN are functionally lost from the genome and unavailable for the general maintenance of the cell.
Fig. 4a–f

The majority of micronuclei (MN) do not contain detectable levels of transcribing RNA polymerase II (PolII). UPCI:SCC103 cells were stained with 4′, 6-diamidino-2-phenylindole (DAPI) (a,c,e) and antibodies to the elongating form of PolII (b,d,f). Nuclei are uniformly positive for transcribing PolII. Arrows in a,b show MN that are not immunoreactive for PolII. c,d The PolII inhibitor α-amanitin eliminates immunoreactivity in the nuclei, showing the specificity of the anti-PolII antibodies. In contrast to MN, nuclear blebs (arrows in e,f) are always positive for PolII immunoreactivity. Bars represent 10 μm. Images in a and b; c and d; and e and f are of the same cells

Fig. 5

Micronuclei mostly fail to incorporate labeled UTP. Fluorescently labeled UTP was bead-loaded into UPCI:SCC103 cells and incorporation into RNA visualized by epifluorescent microscopy. UTP fluorescence is shown in a,c,e. DAPI staining is shown in b,d and phase contrast in f. White arrows show MN that fail to incorporate UTP, indicating a lack of transcriptional activity. α-Amanitin eliminates UTP incorporation outside of the nucleoli (shown with black arrows in f), indicating the specificity of the UTP labeling. Bars represent 10 μm. Images in a and b; c and d; and e and f are of the same cells

We next investigated why MN showed reduced transcriptional activity. One trivial explanation is that the size of these structures somehow physically blocks transcription or our ability to detect it. To test this, we compared the labeling of MN with small nuclear blebs or out-pockets of the nuclear envelope. Unlike MN, blebs were always positive for both transcriptional labels (Fig. 4e,f). Thus, the reduction of transcription in MN is not directly due to their small size. Similarly, we saw no correlation between MN size and transcriptional activity (data not shown).

We next hypothesized that the MN might have reduced transcriptional activity owing to restricted nuclear trafficking. To test this, we first looked for NPCs in the nuclear envelope of MN. Immunofluorescence with a monoclonal antibody that recognizes a variety of related nucleoporins (Davis and Blobel 1986) intensely labeled nuclei in essentially all cases, but MN were positive only 12% (n=135) of the time (Fig. 6a,b). In contrast, nuclear blebs were always positive (Fig. 6c,d). To confirm that the loss of NPCs in MN was not an artifact of cell culture we examined both frozen sections and fresh touch preparations of tumor material. In these samples, significant nuclear fragmentation was observed. To confine our analysis to MN, we co-stained with anti-lamin antibodies to distinguish nuclear membrane-bound structures from free chromatin. In the relatively few unambiguous examples of MN identified in tumor tissue, all were negative for NPCs (Fig. 6e,f,g). These results show that MN are defective in the assembly of immunoreactive nuclear pores.
Fig. 6a–g

Most micronuclei (MN) show no labeling with antibodies to nuclear pores. UPCI:SCC103 cells were stained with 4′, 6-diamidino-2-phenylindole (DAPI) (a,c) and antibodies to nuclear pore complexes (NPCs) (b,d). Nuclei were intensely labeled with the anti-NPC antibody, but MN (arrows in a,b) showed generally low levels of labeling. Nuclear blebs (arrows in c,d) were always positive for NPC staining. e,f,g Micronuclei in touch preparations of oral squamous cell carcinoma tumor tissue were also negative for NPCs, showing the defect is not a result of cell culturing. e DAPI and g anti-nuclear pore labeling. f In tumor samples, anti-lamin antibodies were used to distinguish MN from chromosomal fragments. Bars represent 10 μm. Images in a and b; c and d; and e, f, g are of the same cells

The reduced pore density in the MN suggests a defect in nuclear import. To test this directly, we measured nuclear uptake of exogenous glucocorticoid receptor-GFP (GR-GFP) (Carey et al. 1996). Prior to addition of dexamethasone the GR-GFP is primarily cytoplasmic (Fig. 7a). After addition of the ligand, the receptor is found in the nucleus (Fig. 7b). Receptor uptake was essentially always observed for nuclei, but markedly diminished for MN (Fig. 7c,d,e,f). Only 10% (n=108) of MN appeared competent to take up this nuclear protein to detectable levels. Nuclear blebs were able to import the labeled GR (Fig. 7c, small arrow). These data indicate that MN are defective for nuclear transport and are consistent with the conclusion that the MN have reduced NPCs. Micronuclei are generally positive for nuclear lamin staining, showing that they have at least some of the structural features of nuclei and are not grossly defective for all resident nuclear proteins (Fig. 3f,g). Some MN were negative for even this basic structural component of the nuclear envelope (arrow in Fig. 3g).
Fig. 7a–f

Micronuclei are defective for nuclear import. UPCI:SCC103 cells were transfected with a plasmid containing glucocorticoid receptor-green fluorescent protein (GR-GFP). GR-GFP labeling is shown in a,c,e, DAPI staining in b,d,f. Cells in a are untreated, and in b treated with dexamethasone, showing nuclear import of this receptor is triggered by addition of dexamethasone. Cells in c,d,e,f were all treated with dexamethasone. Import into the nucleus and nuclear blebs (thin arrow in c,d) was essentially always observed, but import into the MN (thick arrows) was much less common. Bars represent 10 μm. Cells in a and b; c and d; and e and f are from the same field

We next measured these variables in other tumor-derived cell lines (Fig. 8a). Micronuclei showed generally poor PolII labeling, nuclear pore density and nuclear transport for all of the tested lines. However, the levels varied markedly between different cell lines. In the trophoblastic choriocarcinoma JAR cells, 68% (n=50) of the MN were positive for NPCs, but in one oral squamous carcinoma line (UPCI:SCC040), 0% of MN (n=70) had NPC immunostaining. Further analysis will be required to determine whether these variations we see in the cell lines reflect differences in the tumor tissue of origin or other aspects of tumorigenesis.
Fig. 8a,b

Comparison of features of micronuclei (MN) of cancer cells, normal cells and cells exposed to the DNA-damaging agent H2O2. a Comparison of MN frequency as a percentage of in the indicated cell lines. UPCI cells are human oral cancer cell lines, JAR are human trophoblastic choriocarcinoma cells, and UP3 are normal oral epithelial primary cells. An asterisk indicates that the assay was not done for that cell line. b Comparison of features of H2O2-induced and untreated control MN in HeLa cells. (GR-GFP glucocorticoid receptor-GFP, mn freq frequency of micronuclei, NPC nuclear pore complex, PolII RNA polymerase II)

As discussed in the Introduction, MN are measured in many studies to identify and classify various genotoxic compounds (Heddle et al. 1991; Muller-Tegethoff et al. 1997; Shelby 1988; Smith et al. 1993). To see whether these conclusions from cancer cells apply to MN induced by genotoxic agents, we examined the same variables following exposure to the DNA-damaging agent H2O2 (Termini 2000). Chemically induced MN in HeLa cells were found to have the same defects in nuclear import, NPCs and PolII transcriptional activity as spontaneously occurring MN (Fig. 8b). Similar results were observed for the colorectal tumor-derived line HCT116 (not shown). We next asked whether these features of MN were unique to tumor or tumor-derived cells. In normal oral keratinocytes, 65% (n=234) of the MN were negative for NPCs and 98% (n=131) were negative for active PolII (UP3 in Fig. 8a). Thus, both induced and spontaneous MN in normal and tumor-derived cells are mostly negative for NPCs and PolII transcriptional activity.

Discussion

We have examined the resolution of anaphase chromatin bridges in cancer cells labeled with GFP-tagged histone. The bridges frequently break into multiple fragments that become incorporated into MN in the next interphase. This high level of MN formation from anaphase bridges is unexpected and indicates that bridge resolution is a major source of MN formation. Many of the MN were negative for anti-centromere immunolabeling while positive for γ-H2AX staining and TUNEL staining. Based on these results, we propose that many of the MN observed in response to genotoxic agents are due to the formation and resolution of anaphase bridges.

The formation of more than two fragments during resolution of ∼40% of anaphase bridges suggests that forces acting to break the bridge may not simply involve tension on the bridge caused by the spindle poles. Once a single break occurs on the chromosome, tension from the bipolar attachment should be lost. Multiple breaks argue for another, as yet unknown, mechanism resolving these bridges. Anaphase bridges have been observed to be broken by phragmoplast formation in the plant Haemanthus katherinae (Bajer 1963, 1964), and an analogous mechanism may be operating in mammalian cells. In some circumstances, chromatin bridge breakage in our hands was coincident with cytokinesis, but we as yet have no direct evidence that the cleavage furrow contributes to resolution of the bridge. A second potential mechanism may be the activity of DNA-modifying enzymes like topoisomerase II (Swedlow and Hirano 2003). Further analysis will be required to determine whether these mechanisms play a role in bridge fragmentation during anaphase in cancer cells.

Most MN appear to be inactive for transcription based on immunolabeling with antibody to the elongating form of PolII and uptake of fluorescently labeled UTP. Previous analysis of this issue has been inconclusive (Kato and Sandberg 1968; Labidi et al. 1987). Our results are most consistent with those of Kato et al. and suggest that most MN are transcriptionally inactive and that the DNA sequences contained within are at least temporarily unavailable to the transcriptional machinery. While the subnuclear lamin appears normal in the MN, the NPCs and nuclear import of GR are both markedly reduced. Therefore, we suggest that a defect in nuclear trafficking may explain the lack of transcriptional activity in the MN. One possible explanation for the reduced nuclear pore numbers is suggested by the observation that the zinc finger motifs of NPC proteins can bind chromatin or DNA directly (Sukegawa and Blobel 1993). We hypothesize that in the absence of these DNA-binding sites the nuclear pores may fail to form or form incorrectly. This model could explain the variation we see between different MN; those that contain the putative binding regions would retain the NPCs, the rest would not. Testing of this hypothesis will require the identification of the DNA sequences required for nuclear pore binding.

Previous analyses of chemically induced MN in Syrian hamster embryonic cells suggested that they had normal ultrastructure as judged by electron microscopy, including the presence of nuclear pores, nucleoli, nuclear lamin and the juxtaposition of heterochromatin against the nuclear envelope (Schiffmann and De Boni 1991). However, in these ultrastructural studies, it was not reported how frequently the pores were seen relative to normal cells, or whether the observed pores were active for transport. Thus, the MN observed in this study may be nonfunctional or present at a low frequency.

In summary, our results show that anaphase bridges usually give rise to transcriptionally inactive MN, thereby restricting the genetic capacity of the cell. These MN usually lack nuclear pores and have reduced nuclear import. These observations emphasize yet another mechanism whereby anaphase bridges can alter and destabilize the cancer cell karyotype.

Acknowledgements

The authors gratefully acknowledge: Dr. D. DeFranco for the gift of the GFP-GR, Dr. U. Surti for the JAR cells, Dr. C. Walsh for the anti-tubulin antibodies, Dr. C. Feghali for the anti-centromere antibodies, and the tissue bank of the Oral Cancer Center of Discovery of the University of Pittsburgh Medical Center and Dr. Jennifer Hunt for the tumor samples and the University of Pittsburgh Cancer Institute for material support. We also thank Dr. Nicholas Quintyne and Janet Reing for critical review of the manuscript. The work was supported by National Institutes of Health Grant P60DE13059 to Dr. Eugene N. Myers.

Supplementary material

supp.pdf (78 kb)
(PDF 79 KB)

Copyright information

© Springer-Verlag 2004