Interphase chromosomal abnormalities and mitotic missegregation of hypomethylated sequences in ICF syndrome cells
- First Online:
- Cite this article as:
- Gisselsson, D., Shao, C., Tuck-Muller, C.M. et al. Chromosoma (2005) 114: 118. doi:10.1007/s00412-005-0343-7
- 133 Views
The immunodeficiency, centromeric region instability, facial anomalies (ICF) syndrome is a rare autosomal recessive disease. Usually, it is caused by mutations in the DNA methyltransferase 3B gene, which result in decreased methylation of satellite DNA in the juxtacentromeric heterochromatin at 1qh, 16qh, and 9qh. Satellite II-rich 1qh and 16qh display high frequencies of abnormalities in mitogen-stimulated ICF lymphocytes without these cells being prone to aneuploidy. Here we show that in lymphoblastoid cell lines from four ICF patients, there was increased colocalization of the hypomethylated 1qh and 16qh sequences in interphase, abnormal looping of pericentromeric DNA sequences at metaphase, formation of bridges at anaphase, chromosome 1 and 16 fragmentation at the telophase–interphase transition, and, in apoptotic cells, micronuclei with overrepresentation of chromosome 1 and 16 material. Another source of anaphase bridging in the ICF cells was random telomeric associations between chromosomes. Our results elucidate the mechanism of formation of ICF chromosome anomalies and suggest that 1qh–16qh associations in interphase can lead to disturbances of mitotic segregation, resulting in micronucleus formation and sometimes apoptosis. This can help explain why specific types of 1qh and 16qh rearrangements are not present at high frequencies in ICF lymphoid cells despite diverse 1qh and 16qh aberrations continuously being generated.
DNA methyltransferase 3B
fluorescence in situ hybridization
immunodeficiency, centromeric instability, facial anomalies
lymphoblastoid cell line
classical satellite II DNA
The immundeficiency, centromeric region instability, facial anomalies (ICF) syndrome is a rare autosomal recessive disease. Since it was first described by Hulten et al. (1978), about 35 cases have been reported in the literature. It has recently been shown that most ICF patients exhibit inactivating mutations of the DNA methyltransferase 3B gene (DNMT3B), leading to invariant hypomethylation of a limited number of genomic regions (Kondo et al. 2000), in particular the heterochromatic regions containing tandemly repeated sequences of classical satellite II (Sat2) and III (Sat3) (Jeanpierre et al. 1993; Hansen et al. 1999; Xu et al. 1999). Immunologically, ICF patients show hypogammaglobulinaemia but with B-cells present (Ehrlich 2003). It has been suggested that this condition is mediated by the dysregulation of lymphogenesis genes (Ehrlich et al. 2001) and defective B-cell differentiation, leading to the accumulation of immature B-cells with an increased rate of apoptosis upon in vitro activation (Blanco-Betancourt et al. 2004). Because of a strong predisposition for systemic infectious diseases, the majority of ICF patients die before reaching their teens.
Peripheral blood lymphocytes from ICF patients are cytogenetically characterized by abnormalities of the long Sat2-rich pericentromeric heterochromatin regions of chromosome 1 and 16, and, to a much lesser extent, also of the Sat3-rich or Sat2-rich heterochromatic regions of chromosomes 9, 2, and 10 (Ehrlich 2003). The abnormalities of chromosomes 1 and 16 include decondensation (stretching) of the heterochromatin, whole-arm deletions and translocations, isochromosome formations, and multiradial chromosomes. Consistent with DNA demethylation causing these abnormalities, the DNA methylation inhibitor 5-azacytidine has been shown to induce ICF-like decondensation and rearrangements targeted to the 1qh, 16qh, and 9qh regions in normal lymphoid cells (Schmid et al. 1983; Kokalj-Vokac et al. 1993; Hernandez et al. 1997). The cytogenetic changes are typically present in mitogenically stimulated B- and T-cells from peripheral blood of ICF patients, whereas they are less frequent or less complex in bone marrow cells and fibroblasts (Tiepolo et al. 1979; Carpenter et al. 1988; Maraschio et al. 1988; Turleau et al. 1989; Fasth et al. 1990; Smeets et al. 1994; Brown et al. 1995). The reason for this tissue-specificity is not known. It has been suggested that DNMT3B plays a role in 1qh and 16qh condensation by interacting with other proteins in the condensin complex and with DNA motor proteins (Geiman et al. 2004). However, early embryogenesis proceeds normally in DNMT3B knockout mice and DNMT3B is usually much less plentiful in postnatal than embryonic cells (Okano et al. 1999).
There is still a lack of experimental data regarding several cytogenetic aspects of the ICF syndrome. It is not known when in the cell cycle the different chromosome aberrations are generated. Neither is it known whether the cytogenetic abnormalities are created independently of each other or if some of the abnormalities arise as derivatives of other abnormalities. This study is an attempt to answer these questions by monitoring the interphase configuration of hypomethylated DNA sequences in lymphoblastoid cell lines (LCLs) from four ICF syndrome patients, as well as their chromosome dynamics at mitosis and the frequency of genomic imbalances among apoptotic cells. Our findings suggest that illegitimate recombination of heterochromatic sequences at interphase due to increased pericentromeric associations of chromosomes 1 and 16 leads to severe perturbations of the mitotic process, resulting in a variety of abnormal chromosome derivatives, and an increased propensity for apoptosis among genetically unbalanced cells. Surprisingly, the mitotic abnormalities were enhanced in three of the four ICF LCLs by the presence of short dysfunctional telomeres among various chromosomes.
Materials and methods
Cell culture, chromosome dynamics, and chromosome banding
Four LCLs (ICF K, ICF C, ICF B, and ICF P5) obtained from Epstein–Barr virus (EBV)-transformed peripheral blood B-cells of ICF patients were used (Carpenter et al. 1988; Kieback et al. 1992; Smeets et al. 1994; Wijmenga et al. 2000). Two EBV-transformed LCLs from healthy individuals (AG15022 and AG14953; Coriell Institute) were the normal controls. The cells were grown in RPMI 1640 medium with HEPES buffer, supplemented with 15% (ICF K) or 20% (ICF B, ICF C, ICF P5, AG15022, AG14953) fetal bovine serum, 0.23 mg/ml l-glutamine, 100 IU/ml penicillin, and 0.2 mg/ml streptomycin. For analysis of metaphase chromosomes, cells were arrested at metaphase with 0.02 μg/ml Colcemid, harvested, and G-banded by standard procedures. The structure of heterochromatin was assessed in a blinded fashion in 50 metaphase cells from each LCL. Decondensation was scored when the entire length of 1qh or 16qh was thread-like with the width of the heterochromatin of both chromatids being no greater than that of the centromere of the same chromosome (Tuck-Muller et al. 2000). Visualization of chromosome dynamics at mitosis was performed by harvesting without metaphase arrest, washing in phosphate-buffered saline (PBS), fixation in 3:1 methanol:acetic acid, followed by hematoxylin–eosin staining. Anaphase cells showing at least one string of chromatin connecting the poles were classified as harboring an anaphase bridge. Lagging chromosomes were defined as bodies staining with hematoxylin or diamidinophenylindole (DAPI) in a fashion similar to the mitotic chromosomes in the same cell, but being clearly separated from the metaphase or anaphase plates by at least one chromosome width (approximately 1 μm). Micronuclei (MN) were defined as hematoxylin or DAPI-positive bodies adjacent to a nucleus and with the same staining as that nucleus, but with a diameter no larger than one third of the maximum nuclear diameter.
Fluorescence in situ hybridization
Fluorescence in situ hybridization (FISH) was performed as previously described (Gisselsson et al. 2000). Cell-division preparations harvested without metaphase arrest and hypotonic treatment were treated with 20 mg/ml pepsin for 10 min in 0.01 M HCl and fixed in 1% formalin/PBS prior to hybridization. Centromeric satellite sequences were detected using a Cy3-labelled human pan-alpha satellite probe (CAMBIO, Cambridge, UK). Commercially available probes (Vysis/Abbott Diagnostics, Abbott Park, IL) were used for whole-chromosome painting and detection of Sat2 sequences of chromosomes 1 or 16. Telomeric TTAGGG repeats were visualized by FISH with fluorescein-conjugated (CCCTAA)3 peptide nucleic acid probes (Landsdorp et al. 1996). The signal intensity was directly quantified by the Cytovision software (Applied Imaging, Newcastle, UK), and the number of negative chromosome termini for each metaphase cell was scored. Although FISH-negative chromosome ends may still contain up to 500 bp of telomeric repeats, this method gives a valid assessment of the protective capacity of individual telomeres (Martens et al. 2000; Gisselsson et al. 2001). Chromosomes were counterstained with DAPI. For the examination of interphase cells for colocalization of 1qh and 16qh signals with dual-labeled probes, we used only cells with nuclei that had two 1qh signals and two 16qh signals, including those with an overlapping 1qh and 16qh signal; therefore, only cells in G1 or early or middle S-phase were examined.
Detection of apoptotic cells
Phosphatidylserine on the outer leaflet of the cytoplasmic membrane was detected by Annexin V staining (Vybrant Apoptosis Kit, Molecular Probes, Eugene, OR). Fragmented DNA was fluorescently labeled using terminal transferase (TUNEL assay; In Situ Cell Death Detection Kit, Roche, Mannheim, Germany).
Chromosome aberrations at metaphase
Chromosome banding data and telomere status
Percentage of metaphases (actual ratio)b
TTAGGG-negative termini: mean valuec
1qh or 16qh decondensation
1qh or 16qh rearrangement
Non-clonal chromosome losses
Non-clonal chromosome gains
Chromosome dynamics at mitosis
Abnormalities of cell division morphology
Percentage of cells (actual ratio)
Hybridization to interphase nuclei with whole-chromosome painting probes for chromosomes 1 and 16 in the ICF C LCL revealed that 33 and 22% of chromatin bridges, respectively, contained material from these chromosomes. Neither of these chromosomes was found in telomeric fusions in this cell line, suggesting that many of the chromosome bridges in the ICF LCLs arose from abnormalities in the pericentromeric heterochromatin of chromosomes 1 and 16. This was corroborated by FISH with probes for 1qh and 16qh satellite sequences to anaphase bridges in three of the LCLs, showing that a proportion of bridges were positive with at least one of these probes (Fig. 2b; 1/6 bridges in ICF B; 2/5 bridges in ICF C; and 2/4 bridges in ICF K).
The ICF LCLs also exhibited lagging of chromosome material at metaphase in 5–10% of the cells (Fig. 1f), whereas this was rare in the control LCLs (Table 2). However, at anaphase, cells from one of the control LCLs as well as the ICF LCLs exhibited lagging chromosome material. FISH with an alpha satellite probe for all human centromeres demonstrated that only 42–68% of the metaphase laggards in the ICF LCLs contained centromeric alpha satellite sequences, indicating that the laggards could originate from acentric fragments as well as from whole chromosomes that had detached from the mitotic machinery. A small fraction of the ICF metaphase cells (approximately 1–2%) exhibited loops of extended chromatin (Figs. 1g and 2c). Segments of these loops were invariably positive by FISH with a pan-alpha satellite probe, indicating that they contained centromeric sequences, and, therefore, might contain adjacent juxtacentromeric heterochromatin. In ICF C, FISH accordingly demonstrated looping chromatin positive for 1qh in one metaphase cell (data not shown).
A likely outcome of mitotic chromosome lagging is loss of genetic material from the nuclear genome through the formation of MN. Earlier studies revealed the frequent presence of chromosome 1 and 16 material in MN of ICF lymphocytes (Maraschio et al. 1992; Sawyer et al. 1995). We therefore evaluated the contribution of these two chromosomes to genomic loss in the ICF C LCLs. Upon hybridization with whole-chromosome painting probes for chromosomes 1 and 16, 33% (8/29) of MN in this cell line had signals from either one or the other of these chromosomes, and approximately 10% (3/29) contained material from both of these chromosomes (Fig. 2d and e).
1qh and 16qh at interphase compared to metaphase
Frequency of 1qh–16qh association
Percentage of cells (actual ratio)
To test whether the loss of chromosomal material might increase the apoptosis rate, cells from ICF C and AG14953 were stained with annexin V and DAPI. Annexin V staining revealed an apototic cell fraction of approximately 10% in both LCLs. Annexin V-positive cells lacking nuclear fragmentation, corresponding to early stages of apoptosis, were then assessed separately in ICF C for genomic losses seen as the presence of a MN. Cells with MN were clearly overrepresented (31/52) in annexin V-positive cells compared to the non-apoptotic interphase cell population (21/204; P<0.001; Fig. 2h,i). Similar results (23/44 vs. 21/204, for the early apoptotic vs. non-apoptotic ICF C cell populations, respectively) were obtained using the TUNEL assay. In comparison, early apoptotic cells in the control AG14953, exhibited very few MN (1/101). This indicated that missegregation leading to loss of chromosome material through MN formation might be one factor contributing to cell death in ICF syndrome LCLs. This was supported by the finding of subpopulations of cells with both apoptotic nuclear morphology and chromatin strings (Fig. 1h and i) in all four ICF LCLs. Similar phenomena were not found in either of the control LCLs.
Colocalization of 1qh and 16qh in interphase in ICF cells
Because of the prevalence of metaphase abnormalities in 1qh and 16qh in mitogen-stimulated lymphocytes and LCLs from ICF patients (Jeanpierre et al. 1993; Smeets et al. 1994), it was important to elucidate the nature of these genomic regions during the cell cycle in ICF cells. Previously we showed that multiradials composed of arms of both chromosomes 1 and 16 were favored over homologous associations and suggested that they represent unresolved Holliday junctions (Tuck-Muller et al. 2000). By comparing four B-cell lines from ICF patients and two control B-cell lines, we now show that ICF cells exhibit co-localization of the juxtacentromeric heterochromatin of chromosomes 1 and 16 (1qh and 16qh) in interphase nuclei at a significantly higher frequency than control cells and than in the corresponding ICF metaphases. These findings extend previous indirect evidence for more frequent somatic paring of 1qh and 16qh in ICF lymphocytes than in control lymphocytes by counting interphase FISH signals from a single 1qh/16qh probe (Maraschio et al. 1992). Although we did not quantitate 1qh:1qh or 16qh:16qh associations (due to the inability to distinguish monosomy from these associations and the greater technical difficulty of resolving two close signals of the same color from two overlapping signals of different colors in dual-color interphase FISH), we noted for the four ICF LCLs that cells with one 1qh or one 16qh signal were much less frequent than those with overlapping 1qh and 16qh signals. This may be related to our previous finding that ICF metaphase quadriradials with mixed chromosomes 1 and 16 arms predominated over those with just chromosome 1 arms or only chromosome 16 arms (Tuck-Muller et al. 2000).
Although it was difficult to quantitate precisely, our study provides evidence for increased stretching of 1qh and 16qh even in interphase. Frequent decondensation of 1qh and 16qh in interphase could favor associations via the extensive homology between Sat2 at 1qh and 16qh. This could be fostered by possible difficulties in replicating decondensed, undermethylated Sat2-rich heterochromatin, which could lead to DNA damage, exposure of single-stranded sections of DNA, and enhanced pairing of 1qh and 16qh at interphase.
Chromosome 1 and 16 stretching and breakage during mitotic movement
Probably some, but not all, of the colocalization of 1qh and 16qh FISH signals at interphase arises from covalent cross-links between these chromosomes. In our previously reported analysis of 1,500 ICF metaphases, the predominant rearrangement containing both chromosome 1 and 16 arms was the quadriradial(1;16)(pq;pq), which appears to consist of one chromosome 1 and one chromosome 16 homologue covalently attached in the pericentromeric region (Brown et al. 1995). Covalent cross-links should be under considerable mechanical strain during the process of chromosomal congression at the interphase–metaphase transition. This could explain the presence of despiralized DNA strings and DNA fragments in metaphase cells, found in all ICF syndrome LCLs but not in either of the control LCLs. The one ICF LCL studied in the most detail, ICF C, displayed both looping chromatin at metaphase that was positive for 1qh and DAPI-positive chromosome fragments that hybridized with 1qh or 16qh probes. There were also anaphase chromatin bridges positive for chromosome 1 and 16 heterochromatin. Overall in the ICF cells about one third of these bridges were from chromosomes 1 and 16. These were found even in the ICF LCL that did not exhibit elevated levels of telomere associations (ICF K). Therefore, some of the ICF-specific chromosome anomalies seen at metaphase persist to form bridges at anaphase. These anomalies include multiradial chromosomes, which consist predominantly of just three or more chromosome 1 arms, three or more chromosome 16 arms, or both chromosomes 1 and 16 arms (Tuck-Muller et al. 2000). Moreover, some of these 1qh- or 16qh-positive bridges persisted even into the next interphase. However, more commonly, loss of chromosome material in micronuclei seemed to occur. These findings are consistent with our previous hypothesis for how chromosome 1 and/or 16 multiradial chromosomes, other than balanced quadriradials, form. Multiradial ICF chromosomes other than the balanced quadriradials always were associated with gains or losses of chromosome 1 or 16 arms seen at metaphase, which we proposed to result from chromatid breakage and missegregation of the much more prevalent balanced quadriradials (Tuck-Muller et al. 2000). Another even larger source of anaphase bridges and chromosome lagging is probably the chromosome 1 and 16 homologues not associated with each other but with extensive decondensation at 1qh or 16qh. These individual chromosomes, as well as multiradial chromosomes, which should be subject to great strain during anaphase chromosome movement, probably contribute to the frequent pericentromeric breaks and whole-arm deletions of chromosomes 1 and 16 observed in mitogen-stimulated ICF lymphocytes and LCLs (Tuck-Muller et al. 2000).
Telomere associations in ICF cells and anaphase abnormalities
Telomere associations were observed often in ICF, but not control, LCLs in a previous study (Tuck-Muller et al. 2000). In the present study these were seen in three of the four ICF LCLs. It is likely that telomeric associations leading to functionally dicentric chromosomes often form bridges at anaphase. A similar sequence of events has been noticed in many cancers and in immortalized cells in vitro (Gisselsson 2003). In tumor cells exhibiting such chromosomal instability, the telomeric fusions are typically initiated by shortening of the terminal TTAGGG repeats. In the present study, the three ICF LCLs with a high frequency of telomeric associations did indeed show an abnormally high number of TTAGGG-negative chromosome ends compared to controls, indicating that abnormal shortening of the tandem telomeric repeat sequences is the most likely explanation for many of these chromosomal fusions. An overrepresentation of telomeric associations has not been reported in primary lymphocyte cultures from ICF syndrome patients, and telomerase is typically not expressed in diploid LCLs (Okubo et al. 2001). We therefore favor the explanation that the extensive shortening of telomeres in the three ICF LCLs reflects short, but normal, telomeres at the establishment of these lines, which reached subnormal TTAGGG repeat lengths during subsequent proliferation in vitro. Whether this abnormality results indirectly from decreased DNMT3B enzymatic activity in ICF cells or is due to altered interaction of DNMT3B with chromatin proteins remains to be determined. However, many or all of these protein interactions involve parts of the enzyme not affected by ICF-causing DNMT3B mutations, and so should be unaffected by these mutations (Gowher and Jeltsch 2002; Geiman et al. 2004).
ICF chromosome damage and apoptosis
The high frequency of damage to chromosomes 1 and 16 in ICF lymphoid cells is manifest in interphase as nuclear protrusions and MN, as described in previous studies of ICF lymphocytes and LCLs (Sawyer et al. 1995; Stacey et al. 1995). A similar association between 1qh/16qh instability and MN formation was observed in the present study. ICF lymphocytes exhibit an increased sensitivity to B-cell apoptosis after in vitro stimulation (Blanco-Betancourt et al. 2004). They are also abnormally prone to apoptosis and non-apoptotic cell death following radiation-induced chromosome breakage (Narayan et al. 2000). We found that the one tested ICF LCL, ICF C, exhibited a significant overrepresentation of cells with MN among the annexin V-positive, early apoptotic cells. Previous studies have shown that, after irradiation, cells with MN are less likely to form colonies in culture than those without MN (Grote et al. 1981). That survival is rare after chromosome loss or breakage in ICF syndrome cells is further supported by the low frequency of clonal 1qh and/or 16qh abnormalities in the LCLs, despite the high frequency of non-clonal anomalies involving these regions. This indicates that cells with sustained chromosomal breakage or chromosomal loss were not capable of mitotic regeneration. This is in stark contrast to SV40- or HPV-transformed cells with short telomeres, in which anaphase bridging and subsequent loss of genomic material in MN may cause a progression towards highly complex karyotypes (Toouli et al. 2002; Gisselsson et al. 2005). The difference in permissiveness for clonal evolution might be related to the fact that EBV-transformed LCLs have well-preserved DNA damage/cell cycle checkpoints, as also has been shown for ICF LCLs (Narayan et al. 2000) compared to cell lines in which the TP53- and RB1-checkpoints have been disrupted through either the T-antigen of SV40 or the e6–e7 antigens of HPV. The apparently intact DNA damage response machinery in ICF lymphocytes might thus help explain why lymphatic malignancies have not been reported in ICF patients compared to patients with several other chromosomal instability syndromes with short life expectancies and low occurrence in the human population (Seidemann et al. 2000). The association between loss of genomic material in MN and apoptotic cell death in ICF LCLs further supports the notion that chromosomal damage may contribute to the defective development of B-cells in ICF syndrome patients. However, because ICF patients can have very large decreases in specific serum immunoglobulins despite normal B-cell levels (Ehrlich 2003), transcriptional dysregulation still appears to be the predominant cause of their immunodeficiency.
We thank Drs. G.K. Hinkel and K. Sperling for generously sharing the ICF K cell line with us. This work was supported by the Swedish Children’s Cancer Foundation, the Swedish Medical Society, and the Sharon B. Lund Foundation of the American Cancer Society to DG and NIH grant no. CA 81506 to ME.