|Home | About | Journals | Submit | Contact Us | Français|
Epigenetics refers to heritable patterns of gene expression that do not depend on alterations of the genomic DNA sequence. Nickel compounds have demonstrated carcinogenicity without any associated mutagenesis, suggesting that its mechanism of carcinogenesis is epigenetic in nature. One such potential mechanism is the heterochromatinization of chromatin within a region of the genome containing a gene sequence, inhibiting any further molecular interactions with that underlying gene sequence and effectively inactivating that gene. We report here the observation, by atomic force microscopy and circular dichroism spectropolarimetry, that nickel ion (Ni2+) condenses chromatin to a greater extent than the natural divalent cation of the cell, magnesium ion (Mg2+). In addition, we use a model experimental system that incorporates a transgene, the bacterial xanthine guanine phosphoribosyl transferase gene (gpt) differentially near to, and far away from, a heterochromatic region of the genome, in two cell lines, the Chinese hamster V79-derived G12 and G10 cells, respectively, to demonstrate by DNase I protection assay that nickel treatement protects the gpt gene sequence from DNase I exonuclease digestion in the G12 cells, but not in the G10 cells. We conclude that condensation of chromatin by nickel is a potential mechanism of nickel-mediated gene regulation.
Chromatin is the natural state of protein-associated DNA in all eukaryotes. It is a dynamic polymer of subunits called nucleosomes. The nucleosome subunit consists of approximately 146 bps of DNA wrapped around an octamer of histone proteins. The histone octamer is made up of two each of histone H2A, H2B, H3, and H4. The DNA makes one and a half turns around the octamer, with the entry and exit points of the DNA at opposite sides of the disk-shaped histone octameric complex. This complex of 146 bps and the octamer of histones is known as the nucleosome core particle. In chromatin, the exiting and entering DNA duplex strands of this nucleosome core particle extend to the next nucleosome core particles on either side of it. The intervening DNA connecting each pair of nucleosomes is termed the linker DNA. The lengths of these intervening DNA duplex strands vary from species to species and from tissue types to tissues types, as well as from one stretch between two nucleosomes to another. Linker DNA length ranges from a minimum of 25 bps to upwards of 100 bps, averaging approximately 50 to 60 bps (1). A fifth histone type, the linker histone (H1), binds to the linker DNA in a stoichiometry of 1 H1 to 1 nucleosome. The complex of the core particle (the histone octamer and 146 bps of DNA) with the additional 25 bps of DNA and the H1 molecule is called the chromatosome.
Chromatin exists in two states, corresponding to two structural forms, euchromatin and heterochromatin. Euchromatin structure has been characterized as “beads on a string”. The DNA traces a random path punctuated by nucleosomes that appear as beads, or round protrusions, along the path of the string-like DNA duplex. Heterochromatin is characterized as a thick fiber, with a diameter two to three times that of the “beads” of euchromatin, and appears to be much more rigid and less flexible than the “beads on a string” characterization of euchromatin (2). All physical data to date show that heterochromatin is much more irregular in shape and inconsistent in form than it would be if a simple rod-like fiber (3).
While the genome has been subdivided on the basis of the structural parameters of condensation and decondensation, the dynamics of chromatin are a functional phenomenon: euchromatin contains most of the transcriptionally active gene sequences, including, by and large, most of the unique DNA sequences of the genome, and these euchromatic regions decondense as the cell exits mitosis; heterochromatin, on the other hand, is made up of highly repetitive sequences of satellite DNA with a relatively small number of transcriptionally active genes interspersed. It is the program of heterochromatinization, whereby condensation spreads outwards in cis, in response to epigenetic signaling events, that gives rise to the term `facultative heterochromatin', as distinguished from `constitutive heterochromatin', which applies to chromosomal regions that apparently remain condensed in all cell types and at all stages of differentiation (4). The epigenetic signals referred to here are the DNA methylations and histone modifications that in this context can result in a stably inherited gene silencing.
Our interest here with regard to heterochromatin lies in the capacity of the transition metal, nickel, to inactivate transcription of genes by inducing DNA methylation and chromatin compaction (5). Upon engulfment of a gene by the spreading of nearby heterochromatin, the gene is heritably silenced. Nickel-induced genotoxic effects such as DNA strand breaks, DNA-protein crosslinks, and generation of reactive oxygen species (ROS) usually occur in heterochromatic DNA, which is not genetically active and thus this genotoxicity has little mutagenic consequence (6). This squares with evidence that nickel compounds are carcinogenic yet show no mutagenicity in mammalian cell assays. It is well established that chromosomally integrated transgenes in transfected cell lines and transgenic animals are susceptible to inactivation by changes in chromatin structure, thus revealing that chromatin structural change is a potential mechanism of transcriptional regulation (7-9).
Our laboratory has previously reported findings that nickel induces DNA hypermethylation, turning off a senescence gene during the course of nickel-induced transformation of Chinese hamster embryo cells (10-11). The mechanism by which nickel might be expected to stimulate DNA methylation was puzzling to us at that time. However, in studies using two differential cell lines, one (G12) in which a reporter gene (gpt) was inserted proximal to a heterochromatic region of chromosome 1, and the other (G10) in which the reporter gene was placed far from any heterochromatic region of chromosome 6, only the gene placed in, or near, the context of heterochromatin (G12) was down-regulated in response to nickel exposure. From these studies we postulated a model of how nickel exerts its effects on the transcription of genes near heterochromatin (Figure 1; 5-6, 12).
We propose that inactivation of transcription of a gene can be caused by nickel-induced chromatin condensation at euchromatin-heterochromatin junctions: the gene becomes incorporated into regions of spreading heterochromatin, and these conditions may also favor the methylation of the DNA within this newly condensed chromatin. This would also provide an explanation of nickel's carcinogenic properties in the absence of mutagenicity. The gene is engulfed by the newly condensed heterochromatin. Condensation could somehow signal DNA methylation of CpG islands. At replication this methylation state would be inherited, an example of epigenetic inheritance. We hypothesize that DNA methylation takes place after chromatin condensation. We have not verified this experimentally, but if methylation occurs, then DNA methylation would be passed on to the next generation by the normal mechanism of DNA methylation inheritance. That mechanism has not been definitively demonstrated, and we do not pretend to know it. It has been suggested voluminously in the literature that the inheritance of cellular DNA methylation patterns is based on a CpG island-specific methylase that operates on newly replicated hemimethylated DNA (13).
Here, we have performed physical and chemical assays that demonstrate nickel's potential ability to inactivate genes by altering chromatin structure.
Purified trimmed nucleosome monomers were prepared from chicken erythrocytes as described in detail (method II) by Yager et al. (1989) (14) and used as the source of histone octamers. Briefly, nuclei were prepared from 200 ml of chicken erythrocytes before being washed 3 times by centrifuging at 1464 × g for 5 min at 4°C in 25 volumes of buffer A (0.34 M sucrose, 15 mM Tris, 60 mM KCl, 15 mM NaCl, 0.5 mM spermidine, 0.15 mM spermine, 2.5 mM EDTA, 0.5 mM EGTA, pH 7.5) containing 0.1 mM of the proteinase inhibitor phenylmethylsulphonyl fluoride (PMSF). The pelleted erythrocytes were lysed by repeated resuspension and centrifugation with the same buffer, but containing 0.5% Nonidet P-40. The nuclei were digested for 5 min at 37°C with micrococcal nuclease (30U/mg DNA). The digestion was stopped by adding EDTA and lowering the temperature to 0°C. The nuclei were washed and pelleted with a 20-min, 6800 × g centrifugation. The chromatin-containing pellet was resuspended in 10 mM Tris-HCl, 0.25 mM EDTA, and 0.35 M NaCl, pH 8.0, and the linker histones, H1/H5, were removed by incubation of the chromatin with 30 mg/mL carboxymethyl-Sephadex with stirring for 3 hr at 4°C. The cation-exchange resin was removed by centrifugation at 8000 × g for 30 min at 4°C. Supernatant was dialyzed against TE (10 mM Tris, pH 7.8, 0.25 mM EDTA, 0.1 mM PMSF) overnight at 4°C, and again the next morning in fresh buffer for two hr. (The resin is thrown out unless the linker histone is to be purified.)
Very briefly, a second micrococcal nuclease digestion was performed, followed by removal of the DNA by passing the sample through a hydroxylapatite (BioRad, Hercules, CA) chromatography column equilibrated with the buffer described by Simon and Felsenfeld (1979) (15). Elution fractions were monitored by Laemmli (6%/15%) SDS-PAGE (16) and peak fractions pooled. Histone octamer concentration was determined by measurements of A230 (17).
The DNA template 207-12601 consists of 12 tandem head-to-tail repeats of a 207-bp sequence derived from the selection studies of Lowary and Widom (18). The 207-12601 DNA was cloned into the pUC19 to produce plasmid pSG601 (A kind gift from S. Grigoryev), transformed and propagated in DH5α cells. The 207-12601 tandem array was excised by Xba I/Hind III double restriction digests and isolated by gel filtration through a Sephacryl S-1000, eluting from TE (10mM Tris-HCl, pH 7.8, 0.25 mM EDTA).
CD was measured at room temperature on a Jasco J-720 spectropolarimeter (Jasco, Inc., Japan). Spectra from 320 to 200 nm were collected from the room temperature samples. Ellipticities were averaged for 1 s in 1 nm steps with a 2 nm bandwidth. Scans for each sample were repeated 10 times and averaged.
1 μL liquid samples were loaded onto mica discs, and dried in a covered glass container at room temperature. Mica discs were freshly peeled to eliminate foreign particle background for each sample. The samples were scanned topographically by AFM in non-contact mode using a silicon ultralever tip (Model MPP-11123) (Veeco Instruments, Santa Barbara, CA). Samples were characterized and sized by scanning at least 4 randomly selected fields on each substrate. For quality control, a 5-μm calibration grating with 100 nm × 100 nm pits were scanned periodically over the duration of the experiments to assure image integrity. The surface scan data were processed and analyzed using Proscan v. 1.9 image software (Veeco Instruments, Santa Barbara, CA).
The G10 and G12 cells were cultured in F12 medium (Life Technologies, Carlsbad, CA) supplemented with 1 × HAT, 5% FBS, 1% penicillin-streptomycin, at 37°C in a humid 5% CO2 atmosphere for 48 hr. Twenty four hours before the start of nuclei isolation media was replaced with fresh F12 media without HAT.
Nuclei were isolated using Sigma's Nuclei Pure Prep Nuclei Isolation kit, according to the manufacturer's protocol. Metal treatments were done on ice at 106 nuclei/ml for 75 min in nuclei storage buffer (Sigma, St. Louis, MO). Nuclei were then centrifuged at 600 × g and 4°C for 15 min, and washed once with nuclei storage buffer.
Treated nuclei were resuspended in DNase I buffer (10 mM Tris-HCl, pH 7.4, 10 mM NaCl, 3mM MgCl2, 100 μM CaCl2) at 106 nuclei/ml. 195 μL of nuclei were added to each tube containing 5 μL DNase I (Roche, Indianapolis, IN). DNase I digestions were performed on ice for 20 min. 200 μL of stop solution (1% Sodium Dodecyl Sulfate, 0.3 M NaCl, 50 mM Tris-HCl, pH 8.0, 50 mM, 50 mM ethylenediaminetetraacetic acid, 20 μg RNase A) were then added to each sample, samples were mixed and placed on ice for 30 min to stop the reaction. 10 μL of a 20 mg/mL solution of Proteinase K were then added to each tube, and the tubes were mixed and incubated at 65°C for 18 hr.
Samples were extracted with phenol-chloroform and precipitated with 2.5 volumes of 100% ethanol in the presence of Polyacryl carrier (Molecular Research Center, Cincinnati, OH). Visible pellets were washed with 70% ethanol, air dried, and resuspended in 10 mM Tris-HCl, pH 8.0.
The digested DNAs were subjected to PCR using a specific gpt PCR primer set 5'TATTGTAACCCGCCTGAAGTTAAA3' and 5'AACACTTTTTAAGCCGTAGATAAA3', annealing at 55°C. The primers encompass a 581-bp DNA sequence containing the coding region of the gpt gene.
We reconstituted oligonucleosomal tandem arrays, as described in Experimental Procedures, that were of homogeneous constitution and with subunits (the nucleosomes) that positioned along the 12-mer tandem array with perfect regularity. We did this by using the 207-12601 DNA based on the Widom 601 DNA which had been cloned to make a 12-mer repeat of the 207-bp DNA sequence (21). This DNA positions one histone octamer uniquely and with high affinity at each of the 207 bp repeats. Sedimentation velocity analysis verified that these nucleosomal arrays were homogeneous in structure (data not shown), with one histone octamer bound per 207 bp tandem repeat of the 207-12601 DNA, and 12 core octamers for each 12-mer DNA 207 bp repeat.
Figure 2 shows an atomic force microscopic image of one of these reconstituted oligonucleosomes in an extended form in the absence of divalent cation and at very low salt (10 mM Tris-Cl). It is well known and it has been well documented that salt, either monovalent (NaCl) or divalent (for example MgCl2), condenses the chromatin structure in solution (22-24). The data show that the ionic environment has a great influence on the extent of condensation from the 10-nm polynucleosome chain to the so-called “30-nm fiber” of condensed higher order chromatin structure. Although the in vivo state of chromatin is influenced by a mixture of both mono- and divalent cations, divalent cations perturb chromatin structure at concentrations that are nearly three orders of magnitude less than those of monovalent cations. When considering the following biophysical data presented, it would be helpful to keep in mind that the concentrations of mono- and divalent cations in nucleo are high enough to maintain most of the chromatin in a highly condensed state. Both monovalent and divalent cations are capable of holding the chromatin fiber at comparable levels of condensation, but at concentrations that are orders of magnitude less in the case of divalent cations, indicating that condensation of the fiber is not simply due to changes in ionic strength. The special interaction of divalent cations with negative charges of the DNA phosphates is probably what leads to their particular effectiveness in the compaction process. Also, because the 30-nm fiber is a functional as well as a physical entity, it is probably most appropriate to consider it a structurally heterogeneous and dynamically transmuting complex at any given instant.
Reconstituted oligonucleosome samples were exposed to various concentrations of the divalent cation magnesium in the form of the soluble MgCl2, and CD measurements were carried out to examine the organization of DNA in these oligonucleosomal preparations. Figure 3A shows that there is a significant difference between the CD spectra of oligonucleosomes in the presence and in the absence of the divalent cation. The maximum molar ellipticities value at 272 nm decreases from about 6000 for oligonucleosomes in the absence of divalent cation to 5000 in the presence of 0.6 mM Mg2+. As shown in figure 3B, the decrease in maximum molar ellipticities value at 272 nm is significantly greater in the presence of equivalent amounts (0.6 mM) of Ni2+, reduced to about 3000. These concentrations of divalent ion were those that brought on the greatest increment of change in molar ellipticities.
Of particular interest is the formation of two well resolved peaks at 282 and 272 nm in the positive band which are absent in the samples that have not been exposed to divalent cation (Figure 3, A and B, compare red to blue lines in each). These are considered a characteristic feature of the intranucleosomal DNA-histone interactions of the core particle that are absent in the looser configuration of core particles in the “beads-on-a-string” structure (25-27). Note that this characteristic appears much more clearly in the case of nickel-exposed oligonucleosomes than in that of magnesium-exposed ones. This is a clear indication of high-affinity molecular interactions characteristic of DNA-histone molecule binding in the context of higher-order chromatin structure, observed in nickel-treated oligonucleosomes but not observed as significantly in magnesium-treated oligomers.
This distinction is corroborated by the atomic force microscopic images of these same samples, as seen in Figure 4, unfixed and mounted, without vacuum, in air. Figure 4B shows oligonucleosomal arrays in the presence of magnesium, and Figure 4C shows the arrays in the presence of nickel. While the magnesium-exposed oligonucleosomes in Figure 4B show significant compaction, the nickel-exposed oligonucleosomes in Figure 4C are still more condensed, and there are fewer small bead-like structures, indicating that few nucleosomes have escaped the condensation process. We tested with various concentrations of divalent cations, but the concentrations of divalent cation as shown here were those that had the greatest, most significant and dramatic effects observed.
We have previously reported that nickel yields very high levels of 6-thioguanine (6TG) resistance at the bacterial xanthine guanine phosphoribosyl transferase (GPT) locus (gpt) in Chinese hamster V79-derived G12 cells (28-29). The G12 cell line carries a single copy of gpt, the gene for a bacterial nucleotide salvage enzyme that is functionally analogous to the endogenous mammalian hypoxanthine phosphoribosyl transferase (HPRT) gene (hprt). Inactivation of xanthine guanine phosphoribosyl transferase (gpt-) is selected by resistance to 6TG, and gene reactivation (gpt+) is selected by resistance to a mixture of 100 μM hypoxanthine, 1 μM aminopterin, and 100 μM thymidine (HAT) (30-31).
In later work we compared the Chinese hamster V79-derived G12 cell line, whose transgenic gpt plasmid-borne gene has been stably inserted into a site on chromosome 1 near an interstitial subtelomeric band of dense constitutive heterochromatin identified in V79 karyotypes (32), with the Chinese hamster V79-derived G10 cell line, whose transgenic gpt plasmid-borne gene appears to have been inserted, in contrast, in a euchromatic region of chromosome 6 (5). Thus, the G12 transgene is likely to lie within condensed heterochromatin while the G10 transgenic gpt is far from any condensed chromatin, and likely to remain accessible to DNA processes such as transcription, DNA repair, and replication. The results of this comparison led to our proposal that it was the chromosomal integration of the G12's transgene near the heterochromatin of chromosome 1, and the lack of heterochromatinization at the euchromatic integration site of the gpt transgene in the G10 cell line that explains the inactivation of gpt in G12 and its activation in G10. DNase I protection assays, often used as an indicator of chromatin condensation state, showed resistance of the heterochromatic region of the gpt gene in G12 to DNase I, while that of the G10 cell line gave no such resistance to DNase I cleavage.
Therefore, we wished to reevaluate the extent of DNase I protection of G12 and G10 cell line-derived chromatin induced by nickel, and compare the protection expected to be induced by nickel with that of other divalent cations, namely, magnesium and cobalt. Figure 5 is the result of a DNase I footprint assay for nickel, magnesium, and cobalt. Although nickel-treated chromatin from the G12 cell line was entirely refractory to DNase I cleavage, magnesium showed weak protection at a concentration 20 times that of the nickel, and cobalt showed no indication of protecting G12 cell line-derived chromatin from cleavage. As expected, G10 cell line-derived chromatin was cleaved whether untreated or nickel-, cobalt-, or magnesium-treated.
Our previously reported findings that nickel-induced inactivation of gpt expression in a transgenic gpt+ G12 cell line occurs without mutagenesis or deletion of the transgene (5) led to our hypothesis that nickel silenced the gene by an epigenetic mechanism. In the present work we have shown that 1) nickel (Ni2+) condenses a model chromatin, and that the resulting higher-order structure is more compact than the same model chromatin treated with magnesium (Mg2+), suggesting that nickel is a powerful competitor capable of displacing magnesium from a protein- DNA complex, and that 2) when nickel exerts its effect on a chromatin region that contains an active gene, namely the transgene gpt, nickel prevents accessibility to that gene, thus changing the functional state of the underlying DNA epigenetically, since the sequence of the DNA has not changed. It is observed that the model gpt transgene is regulated by nickel-induced condensation only when it is inserted in the region proximal to the dense, heterochromatic chromatin of chromosome 1 of the V79-derived cell line G12, and is not thusly regulated when inserted in the euchromatic regions of chromosome 6 of the V79-derived G10 cell line. In combination with our previous report (5) that nickel induces DNA hypermethylation, and thereby silences a gene during the course of nickel-induced transformation of Chinese hamster cells, we can now propose a potential epigenetic mechanism of nickel-induced carcinogenesis.
Interactions of nickel divalent cations (Ni2+) with chromatin facilitate chromatin compaction, creating dense, heterochromatic regions which spread at either, or both, end(s) of the chromatin fiber. Previously we have shown that the gpt gene in G12 cells is regulated by epigenetic processes, including chromatin condensation and DNA methylation of CpG islands within the gpt transgene (5), and that these processes were reversible. Thus, it is possible that heterochromatinization is followed by DNA methylation, and the resulting silencing of the underlying gene is likely inherited by subsequent generations through this epigenetic mechanism. If these nascent silenced chromatin regions contained tumor suppressor or senescence-related genes the outcome might be cell transformation, tumor progression, and ultimately, oncogenesis. In fact, this may be a more prominent epigenetic mechanism of nickel carcinogenesis than any other propounded to date.
There is substantial evidence that nickel interacts preferentially with the heterochromatic regions of chromosomes (33), and the selective interactions of nickel with heterochromatin may explain why nickel compounds are not generally mutagenic. Nickel ion (Ni2+) was shown here by CD spectropolarimetry and atomic force microscopy (AFM) to increase chromatin condensation much more effectively than Mg2+. This is in agreement with results obtained by other laboratories using other biophysical and hydrodynamic techniques (23). In previous studies (5), we have shown that DNA hypermethylation was associated with decreased DNase I sensitivity, and in those studies, the silencing of the gpt gene was readily reversible by the addition of azacytidine, a reagent which induces hypomethylation. It is clear from this earlier work that nickel can induce hypermethylation, and we propose that both heterochromatinization and DNA hypermethylation are potential mechanistic processes in nickel carcinogenesis. We believe that DNA hypermethylation has the capacity to stabilize the nascent spreading of chromatin condensation in cis that can be induced by nickel ion's displacement of Mg2+ in heterochromatic complexes, as represented in Figure 1, with the potential to perpetuate a heritable epigenetic state through subsequent generations.
Here we have shown supporting evidence that nickel interacts preferentially with heterochromatin rather than euchromatin, since nickel exposure increased DNase I protection in the heterochromatic regions of the V79-derived G12 cell line, while exhibiting no protective effect in euchromatic regions of the V79-derived G10 cell line. This is not surprising as nickel ions bind proteins with much greater affinity than they do DNA. For example, the binding constant of nickel for DNA is K = 7.3 × 102 M-1, while typical binding constants of nickel for amino acids such as cysteine and histidine are 4.37 × 109 M-1 and 1.9 × 109 M-1, respectively (33). In earlier work we speculated that nickel may not condense chromatin as much as does magnesium (33). The results here counter such a speculation, as nickel condensed chromatin more than did magnesium. Clearly there is a structural condensation of chromatin induced by nickel. We believe it is more likely that chromatin condensation is followed by DNA methylation than vice versa because nickel ions should have greater avidity, if not affinity, for chromatin than for specific DNA methylase enzymes.
The molecular interactions shown here between divalent nickel ion (Ni2+) and a defined molecular model of chromatin, and the structural contribution of that interaction to a potential nickel-induced epigenetic mechanism of gene regulation shown here in the context of chromatin associated with a specific functional transgene (gpt), support a model of a potential epigenetic mechanism by which chromatin exposure to nickel leads to carcinogenesis through a structural modulation of chromatin, such as heterochromatinization. This may be one of the most prominent mechanisms of nickel-induced carcinogenesis.
We thank Dr. Michael Schimmerlik (Oregon State University, Corvallis, OR) for the use of the CD spectropolarimeter, and Sergei Grigoryev (Pennsylvania State University, University Park, PA) for the kind gift of the pSG601 DNA.
This work was supported by grants ES014454, ES005512, and ES000260 from the National Institute of Environmental Health Sciences (NIEHS), and grant number CA16087 from the National Cancer Institute (to M. C.) and training grant number 5 T32 ES07324-08 (to T. P. E.).