|Home | About | Journals | Submit | Contact Us | Français|
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its entirety but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions/at/oxfordjournals.org
Promoters are molecular ‘modules’, which are controlled as individual entities yet are often analyzed by nuclease digestion methodologies which, a priori, destroy this modularity. About 40% of mammalian genes contain CpG islands in their promoters and exonic regions, which are normally unmethylated. We developed a footprinting strategy to map the chromatin structure at unmethylated CpG islands by treatment of isolated nuclei with the CpG-specific DNA methyltransferase SssI (M.SssI), followed by genomic bisulfite sequencing of individual progeny DNA molecules. This gave single molecule resolution over the promoter region and allowed for the physical linkage between binding sites on individual promoter molecules to be maintained. Comparison of the p16 promoters in two human cell lines, J82 and LD419, expressing the p16 gene at 25-fold different levels showed that the two cell lines contain remarkably different, heterogeneously positioned nucleosomes over the promoter region, which were not distinguishable by standard methods using nucleases. Our high resolution approach gives a ‘digitized’ visualization of each promoter providing information regarding nucleosome occupancy and may be utilized to define transcription factor binding and chromatin remodeling.
Each promoter molecule in a cell can be considered a module which is controlled individually. Increased levels of overall gene expression can be achieved by increasing the number of promoters functioning or the frequency by which they initiate transcription. Although there is this fundamental modularity to promoter function, most approaches to chromatin structure analysis rely on methods such as nuclease digestion which give average results of all the molecules of a given promoter in a cell population and destroy the physical linkages between binding sites obscuring this individuality. (1,2).
Nucleosomal positioning, [reviewed in (3–8)] and the modification status of the histone tails (9,10), play central roles in controlling transcriptional initiation. Transcriptional co-activator complexes interact with nucleosomes (11) to induce nucleosomal rearrangements. Nucleosomes often have to unfold completely (12) or be disassembled (13) at the transcription start site, to allow for transcriptional initiation (14,15). Most studies focus on the extreme cases of the complete absence or presence of transcription, especially in yeast (16,17). In this paper, we examined nucleosomal arrangements in the p16 promoter in human cell lines, J82 and LD419, which express the p16 gene at 25-fold different levels.
We modified a previously described footprinting method (18), which now allows us to analyze how individual molecules are packaged into chromatin. We treated naked DNA, nucleosomes reconstituted in vitro, or nuclei from two cell lines expressing different levels of p16 with an excess of the robust cytosine-C5 CpG-specific DNA methyltransferase SssI (M.SssI). Although this enzyme methylates all CpG sites in purified DNA, it cannot methylate the same sites when they are assembled into nucleosomes or are associated with tight-binding factors (18,19). By cloning and sequencing DNA molecules following bisulfite treatment, we discerned patches of accessible and inaccessible sites on individual DNA molecules, which reflect packaging of, or protein binding to, each promoter molecule for a given gene (Figure 1). This differs from conventional footprinting techniques relying on nuclease digestion and primer extension (4) in that the footprints are physically linked on individual DNA molecules showing potential simultaneous occupancy of factors controlling transcription.
Our ability to assess nucleosomal positioning on individual DNA molecules reveals a remarkably heterogeneous pattern of nucleosome distribution on the p16 promoter. Since many constitutive and tissue-specific promoters contain unmethylated CpG-island promoters, this approach, alone, or combined with current chromatin-studying methodologies, will provide a useful tool to dissect chromatin structures and changes at a single molecule level.
The two cell lines, the normal fibroblasts LD419 and bladder cancer cells J82 were cultured and maintained as described previously (20).
Total RNA was extracted from the cultured LD419 and J82 cells by using RNeasy Kit (Qiagen) following manufacturer's protocols; reverse transcription and the quantification of mRNA levels was carried out by real-time PCR as described previously (21).
Total RNA was extracted as described above and 5′ ends of mRNA were detected using the RLM-RACE Kit (Ambion) according to the manufacturer's protocols. The inner 5′ RLM-RACE PCR products were cut from a 3% agarose gel, purified using the Gel Extraction Kit (Qiagen) and sequenced with a inner gene-specific primer. The primer sequences are shown in Figure 2C.
All procedures for nuclei isolation were performed at 4°C. Actively growing cells (108) were trypsinized and washed twice with cold phosphate-buffered saline. Cells were resuspended in 9 ml RSB buffer (10 mM Tris–HCl, pH 7.4, 10 mM NaCl and 3 mM MgCl2) and kept on ice for 10 min. Following this incubation, 1 ml of 10% NP-40 detergent was added and cells homogenized with 10–15 strokes of the tight pestle of a Dounce homogenizer. Nuclei were washed twice with RSB buffer to eliminate the detergent, followed by one wash with M.SssI buffer (10 mM Tris–HCl, pH 7.9, 50 mM NaCl, 10 mM MgCl2, 1 mM DTT and 0.3 M sucrose). NaCl (0.4 M) was used to remove most non-histone-bound proteins from chromatin (22). Nuclei were washed once with RSB buffer and once with medium salt buffer (20 mM Tris–HCl, pH 7.4, 0.4 M NaCl, 1 mM EDTA and 5% Glycerol) followed by one wash with M.SssI buffer. Nuclei or salt-washed nuclei were resuspended in 1× M.SssI buffer to a concentration of 107 nuclei/ml. The quality of the salt-washed nuclei was tested by SDS–PAGE.
Core histone octamers were purified from chicken erythrocyte cells as described previously (23). Supercoiled plasmid DNA pSVO-CAT, containing an 870 bp DNA fragment of the p16 promoter region (24), was purified using Wizard Plus Midipreps DNA Purification System (Promega). Reconstitutions were carried out by the salt dialysis method at 4°C as described previously (25). Core histone octamers were mixed with supercoiled plasmid DNA at equivalent amounts.
Extracted nuclei or reconstituted nucleosomes were treated with M.SssI immediately after preparation for 15 min at 37°C. The methylation reactions were carried out in 1× M.SssI buffer with 160 µM SAM (supplied with M.SssI by New England Biolabs). Nuclei from 106 cells (≈6 µg DNA) in a total reaction volume of 150 µl or 1.5 µg reconstituted nucleosomes in a total reaction volume of 50 µl were treated with 60 and 15 U M.SssI, respectively. The activity of M.SssI was also tested using 6 µg of purified genomic DNA or 1.5 µg of plasmid DNA, respectively, under the same conditions as described above. Reactions were stopped by the addition of an equal volume of stop solution (20 mM Tris–HCl, pH 7.9, 600 mM NaCl, 1% SDS, 10 mM EDTA and 400 µg/ml proteinase K). Samples were incubated at 55°C for 16 h and DNA was purified by phenol/chloroform extraction and ethanol precipitation.
Bisulfite genomic sequencing was used to analyze the methylation patterns of individual DNA molecules. The p16 promoter was PCR amplified using DNA that had undergone M.SssI treatment followed by sodium bisulfite conversion (26). PCR conditions and the primer sequences for the p16 promoter have been previously described (27). PCR products were then cloned into the pCR2.1 vector provided by TOPO-TA Cloning Kit (Invitrogen) following the manufacturer's instructions. The positive screened colonies contained the unique sequence of one individual DNA molecule. The plasmid DNA from the selected positive colonies containing vectors with the p16 promoter inserts was purified using the Qiagen plasmid Mini Kit. The purified plasmids were sequenced at the region of interest. Between 30 and 50 individual sequences were performed for each nuclear preparation. In order to exclude a sampling error we performed Ms-SNuPE analysis on six individual CpG sites within the p16 region using the same PCR products which were later cloned. The methylation levels obtained by Ms-SNuPE were comparable with the total methylation of these sites for all sequenced molecules (data not shown). Sequencing was performed at the USC/Norris Microchemical Sequencing Core Facilities. A flow diagram of the method is depicted in Figure 1.
Nuclei from the cultured LD419 and J82 cells were isolated as described above. Nuclei were resuspended in 1× MNase buffer (10 mM Tris–HCl, pH 8.0, 100 mM NaCl and 1 mM EDTA) after two washes with cold RSB buffer to a DNA concentration of ≈50 µg/ml. A total of 400 µl nuclei aliquots were used for MNase digestion in a total reaction volume of 450 µl containing 2 mM CaCl2. Nuclei were digested with increasing concentrations of MNase (0, 0.25–28 U/ml) for 15 min at 37°C. Reactions were stopped by the addition of 50 µl of 10× stop solution (1% SDS, 100 mM EDTA). Freshly dissolved proteinase K was added to a concentration of 200 µg/ml and samples were incubated at 55°C for several hours or overnight. DNA was purified by phenol/chloroform extraction and ethanol precipitation. Qualities of MNase digestions were checked on a 1.5% agarose gel and the appropriate digestion products were selected for Southern hybridization. Partial MNase digested DNA was cut by DraI and subsequently fractioned on a 1.5% agarose gel with a molecular weight marker and then Southern blotted. The blot was hybridized with a 250 bp PCR amplified DraI-probe, which is located 896 bp apart and upstream of the most downstream transcription initiation site (TIS) of p16/CDKN2A. Labeling of the probe was performed with [α-32P]dCTP using High Prime (Roche), and for hybridization the ExpressHyb Hybridization Solution (BD Biosciences) was used according to the manufacturer's instructions. The PCR primer sequences for the probe are sense: 5′-AAACTCACAACAACCCCTATAAAGC-3′ and antisense: 5′-TAGGCAGATAGGAAAATGGGG-3′. Visualization of the bands was achieved by exposure of the blot to Kodak BioMax MR film at −80°C for 7 days.
We have previously defined and characterized the p16/CDKN2A promoter (24), which together with exon 1 of this gene is located in a CpG island that is unmethylated in both LD419 fibroblasts and J82 bladder cancer cells (28). Quantitative real-time PCR showed that the J82 bladder cancer cell line expressed ~25 times more p16 mRNA at the steady-state level than the normal fibroblasts (Figure 2A). Since both cell types have little or no cytosine methylation in the promoter region, the different levels of expression cannot be caused by DNA cytosine methylation. However, it might be due to altered states of chromatin structure which may cause, or be a consequence of, transcription factor recruitment to the promoter. In common with several CpG island promoters, RNase protection analysis has shown evidence of multiple p16 TISs (29); however, the locations of these TISs were not identified precisely and the cell types used differed from our study. We applied 5′-RACE to identify the 5′ ends of p16 mRNAs in LD419 and J82 cell lines. Both cell lines showed three TISs, two of which were shared and one was unique to each cell type (Figure 2B and C). The results of the sequencing of the 5′-RACE-generated transcripts are shown in Figure 2C. The three 5′-RACE-generated mRNAs in the LD419 cell were of equal abundance; however, in J82 cells the shorter transcript (173 bp PCR product) appeared stronger than the other two mRNAs (Figure 2B).
The apparent ‘positioning’ of nucleosomes can be determined by nuclease digestion; however, this approach gives an average for all promoter molecules and does not directly show whether all promoters are organized in the same manner. To assess the average distribution of nucleosomes, we next treated nuclei isolated from both cell types with micrococcal nuclease (MNase) and analyzed the digestion products by Southern blotting (Figure 3). In LD419 cells at least three positioned nucleosomes were apparent over the promoter region (II, III and IV), of which the nucleosomal intervals were ~200 bp (II and III) and 180 bp (IV), consistent with an approximate 190 bp of bulk nucleosomal repeat length. Nucleosome I may also be positioned; however, its upstream boundary could not be defined as the naked DNA control also showed a hypersensitive site in this region. Interestingly, nucleosomes I and II encompassed the TISs and GC boxes. This might potentially restrict access of this region to RNA polymerase II and transcription factors and might contribute to the lower levels of expression in these cells relative to the J82 bladder cancer cells.
Uniformly positioned nucleosomes at the more active p16 promoter in J82 cells were difficult to define by the MNase assay, possibly due to enhanced nucleosomal mobility (11,30). This suggests that the two cell lines may have quite different chromatin structures at this promoter. Furthermore, a distinct MNase hypersensitive region was detected in the region of nucleosome I, which was not detected in either the LD419 cells or the naked DNA control. This seems to be consistent with the observation that many promoter molecules in J82 cells were free of nucleosomes in this region in the M.SssI analysis (see below).
These results encouraged us to modify the method pioneered by Kladde and Simpson (18) for probing chromatin structure at a single molecule resolution.
We took advantage of previous studies which have shown that nucleosomes or DNA binding proteins can block methylation by the CpG-specific DNA methylase M.SssI (18,19). Unlike these earlier studies in which the bulk of the bisulfite treatment reaction products were sequenced, we cloned individual DNA molecules before sequencing to probe for the position of nucleosomes at single DNA molecule resolution.
Initially, we determined the kinetics of methylation of DNA purified from human sperm by M.SssI to establish a baseline upon which the accessibility of CpG sites in genomic DNA and chromatin could be compared. The level of methylation in untreated sperm DNA was negligible at the p16 promoter (Figure 4). The enzyme modified 71% of the CpG sites at the p16 promoter after 7 min and methylation levels reached near saturation after 15 min (data not shown). While there was some slight preference for methylation of individual CpG sites at the intermediate times of incubation, these did not differ much from each other at 15 min (data not shown). Analysis of the reaction products of chromatin fractions incubated with M.SssI might therefore provide a footprinting method, which would give single molecule resolution.
Nucleosomes reconstituted on a cloned p16 promoter were treated with M.SssI for 15 min and individual DNA molecules sequenced to define whether nucleosomal DNA was inaccessible to M.SssI at the p16 promoter (Figure 5). As a control, naked supercoiled plasmid DNA was treated with M.SssI as well. Most cloned DNA molecules derived from reconstituted nucleosomes contained patches of CpG sites, which were inaccessible to the DNA methyltransferase; however, the naked plasmid DNA was completely methylated (data not shown). Interestingly, the analysis of single promoter sequences showed a substantial number of inaccessible regions with overlapping positions. The presence of inaccessible patches in the reconstituted nucleosomes, which were not present when purified DNA was used as the substrate, suggests that this method has applicability toward studying native chromatin structures.
We prepared nuclei from LD419 and J82 cells and analyzed the accessibility of the p16 promoter regions to M.SssI (Figure 6). Virtually no background methylation was present in this region in either cell type before M.SssI treatment as reported previously (28).
Similar to the result obtained with the reconstituted nucleosomes, clear patches of inaccessibility to M.SssI were present on the promoters from both LD419 and J82 cells, supporting the concept that the method is detecting the positions of individual nucleosomes (Figure 6A and B). In both cell types, completely unmethylated or scarcely methylated molecules were observed. This might reflect a fraction of the promoters which may exist in a less accessible configuration such as the 30 nm fiber, or the inability of the methylase to access the chromatin due to technical limitations.
For determination of nucleosome positions, we compared three definitions of a nucleosomal patch. These were defined as at least two consecutively unmethylated sites flanked by 1, 2 or 3 consecutively methylated CpG sites on both sides (Figure 7). All three definitions gave a maximum peak at 120–180 bp corresponding to the expected size of a nucleosome. However, using the 1–1 definition, another peak at the 60–80 bp was observed. This may reflect stochastic methylation by M.SssI, or occasional accessibility of some CpG sites within the core particles, as shown by others (18,19). Both the 2–2 and 3–3 definitions reduced the number of small-sized patches. However, the 3–3 definition was very stringent and resulted in loss of positive data (41 compared with 100 patches). Therefore, the 2–2 definition was further used for analyses of nucleosomal positioning.
Interestingly, the analysis of single promoter sequences showed a clear difference in the positioning and possibly in the number of nucleosomes that could be seen in the two cell types (Figure 6). In LD419 nuclei, many individual promoters had one nucleosome clearly positioned and additional two inaccessible regions spanning the two ends of the analyzed region, suggesting the presence of more nucleosomes. Regions of accessibility were limited to small areas of ~50–70 bp most probably correlating with linker DNA. In contrast, most promoter molecules in the highly p16 expressing J82 cells contained only one nucleosome. The areas accessible to M.SssI were larger in these cells compared with LD419 fibroblasts. Specifically, the region upstream of the TISs was highly accessible in 45% of the promoters in these cells, indicating the lack of positioned nucleosomes correlating with the high levels of transcription from this promoter. Pre-wash of the nuclei with 0.4 M NaCl before M.SssI treatment, which removes most non-nucleosomal proteins from chromatin (22), caused a small reduction in size of the protected patches (Figure 6B). However, it did not markedly alter their number and overall organization, thus validating that these are indeed nucleosomes. Interestingly, while the translational nucleosome positioning in LD419 seemed to vary only to a minor extent, the nucleosomes in J82 cells were much more heterogeneously positioned, possibly reflecting the dynamics of a more active promoter. Thus, this method allowed us to see that in both cell types, each promoter had a unique arrangement of nucleosomes showing that their ‘positioning’ is not absolute. When the percentage of methylation was calculated for each CpG site in both washed and unwashed nuclei (Figure 6C), the overall pattern of protection recapitulated the chromatin organization mapped by the MNase assay (compare with Figure 2), correlating to nucleosomes I, II and III, in LD419 cells and to the hypersensitive region in J82 cells. Thus, averaging of the results derived from our method detects similar results obtained by the commonly used methods. However, its single molecule resolution enables one to see the dynamic nature of the promoter.
Many detailed studies on the roles of chromatin structure in transcriptional regulation have compared active and inactive promoters in simple eukaryotic organisms, such as yeast (16,17,31). However, the mechanisms by which endogenous human CpG island promoters in mammalian cells are regulated is less well understood (32,33). This paucity of information led us to study the composition of the CpG island promoters of p16/CDKN2A.
Conventional approaches are not sensitive enough to discern the organization of individual promoter molecules. While high resolution techniques involving primer extension showed multiple nucleosome translational frames (4) as was found in the present work, these still rely on nuclease digestion. By definition, this means that a promoter could not be studied as a single intact entity until now. The method presented here to footprint DNA–protein interactions at single DNA molecule resolution relies on the known resistance of nucleosomal DNA to M.SssI modification (18,19) coupled with the exquisite sensitivity afforded by the cloning and sequencing of individual progeny molecules.
Several lines of evidence suggest that the 120–180 bp inaccessible patches we observed in the p16/CDKN2A promoter correspond to the presence of nucleosomes. First, their mean size reflects the expected size of a core nucleosome. These patches were also present on nucleosomes reconstituted using purified histones. The differences in patch sizes may be attributed to varying degrees of accessibility of M.SssI to the DNA as it approaches the nucleosomal pseudodyad (18,19). Second, the average patch positions correspond to the patterns detected by traditional MNase digestion. Finally, pre-washing with 0.4 M salt, which is known to remove most non-histone binding proteins, did not markedly alter the patch distribution. However, regions upstream of the TISs became slightly more accessible to M.SssI, especially in high p16 expressing J82 cells (Figure 6C), indicating that DNA binding proteins may also restrict M.SssI accessibility (34,35). In some of the promoters in which we find patches which are significantly larger than 120–170 bp, we believe these to reflect compaction of nucleosomes in a way which did not enable M.SssI access to the linker DNA. Thus, we conclude that the patches correlate with the approximate positions of nucleosomes on single molecules. This new approach substantially increases the resolution of analysis of protein–DNA interactions. It is limited by the fact that it relies on the density of CpG sites found in the area analyzed and that only sites unmethylated before M.SssI treatment can be analyzed. However, while the mammalian genome is generally CpG repressed (36), ~40% of mammalian genes have promoters and exonic regions containing CpG islands (37), which are normally unmethylated (38), meaning that they can be analyzed by this approach. Since organisms such as Saccharomyces cerevisiae and Drosophila melanogaster are not CpG depleted (36), this method should allow for a high resolution analysis of nucleosomal positioning in these species. Another limitation is that while the outer limits of patch size can be determined, the exact sizes cannot. This problem may be overcome in the future by combining CpG-specific (M.SssI) and GpC-specific methyltransferases (M.CviPI) (39), to increase the density of methylated sites.
Application of this method, which reflects the state of a single promoter molecule ‘frozen’ in time, revealed different dynamics of the promoters of two cell types with markedly different levels of p16 gene expression. Many of the promoters in the low expressing LD419 cells had nucleosomes organized in a quite conserved pattern which correlated to nucleosomes I, II and III mapped by the MNase assay. However, in the highly expressing J82 cells only one or two nucleosomes were detected, which were more dynamically organized. Since nucleosomes are often moved or disassembled (13) to allow the transcriptional machinery access (3,5,8,31,40) this could reflect nucleosomal remodeling which may enable the increased level of p16 expression.
Apart from analyzing nucleosome positioning this method may be useful to detect footprints of regulatory proteins acting on various promoters. Furthermore, the combination of cross linking prior to treatment with M.SssI might provide an even higher detailed profile of the promoters analyzed (E. Nili Gal-Yam, S. Jeong, P.A. Jones, unpublished data).
In summary, we have modified a technique by Kladde and Simpson (18) to analyze promoters at single molecule resolution. The method does not rely on cutting with nucleases, so that the relevant nucleosomes and transcription factor footprints are maintained on the replica DNA molecules allowing us to see how the promoter functions as a unit. This approach provides a powerful tool to investigate dynamic changes involved in nucleosome remodeling and transcriptional activation.
The authors thank Drs Judd Rice, Woojin An and Deborah L. Johnson for critical comments on the manuscript, S. Ramu for helping with genomic bisulfite sequencing. This project was supported by the National Institute of Health grant CA 82422 and 83867 and the Max Kade foundation (to G.E.). Funding to pay the Open Access publication charges for this article was provided by NIH Grant CA 83867 and CA 82422.
Conflict of interest statement. None declared.