Differentiation-dependent Alterations in Histone Methylation and Chromatin Architecture at the Inducible Chicken Lysozyme Gene*

It is now well established that locus-wide chromatin remodeling and dynamic alterations of histone modifications are required for the developmentally regulated activation of tissue-specific genes. However, little is known about the dynamics of these events during cell differentiation and how chromatin of an entire gene locus responds to signal transduction processes. To address this issue we investigated chromatin accessibility, linker histone distribution, and the histone methylation status at the macrophage-specific chicken lysozyme locus and the ubiquitously expressed gas41 locus in multipotent precursor cell lines and BM2 monoblast cells. The latter can be induced to go through macrophage maturation by treatment with phorbol-12-myristate acetate and can be further stimulated with bacterial lipopolysaccharide. We show that expression of the lysozyme gene in undifferentiated monoblasts is low and that a high level of gene expression requires both cell differentiation and lipopolysaccharide stimulation. However, depletion of the linker histone H1 is observed already in lysozyme non-expressing multipotent precursor cells. In undifferentiated monoblasts, the lysozyme regulatory regions are marked by the presence of monomethylated histone H3 lysine 4, which becomes increasingly converted into trimethylated H3 lysine K4 during cell differentiation. We also present evidence for extensive, differentiation-dependent alterations in nuclease accessibility at the lysozyme promoter without alterations of nucleosome and transcription factor occupancy.

It is now well established that locus-wide chromatin remodeling and dynamic alterations of histone modifications are required for the developmentally regulated activation of tissue-specific genes. However, little is known about the dynamics of these events during cell differentiation and how chromatin of an entire gene locus responds to signal transduction processes. To address this issue we investigated chromatin accessibility, linker histone distribution, and the histone methylation status at the macrophage-specific chicken lysozyme locus and the ubiquitously expressed gas41 locus in multipotent precursor cell lines and BM2 monoblast cells. The latter can be induced to go through macrophage maturation by treatment with phorbol-12-myristate acetate and can be further stimulated with bacterial lipopolysaccharide. We show that expression of the lysozyme gene in undifferentiated monoblasts is low and that a high level of gene expression requires both cell differentiation and lipopolysaccharide stimulation. However, depletion of the linker histone H1 is observed already in lysozyme non-expressing multipotent precursor cells. In undifferentiated monoblasts, the lysozyme regulatory regions are marked by the presence of monomethylated histone H3 lysine 4, which becomes increasingly converted into trimethylated H3 lysine K4 during cell differentiation. We also present evidence for extensive, differentiation-dependent alterations in nuclease accessibility at the lysozyme promoter without alterations of nucleosome and transcription factor occupancy.
Epigenetic processes involve the establishment of patterns of gene expression via the coordinated and heritable alterations in chromatin structure. These alterations involve a change in the covalent modification of histone N-terminal tails as well as global reorganization of chromatin architecture, all of which is driven by sequence-specific transcription factors recruiting the enzymatic machinery performing these reactions. Such histone tail modifications are not only able to alter the charge or biochemical surface of the chromatin fiber, but also serve as "docking" sites for transcription factor and chromatin remodeling complexes, thus stabilizing the interaction of such complexes with chromatin templates (1). High levels of gene ex-pression go along with an increase in the accessibility of the chromatin of extended genomic regions to nuclease digestion and hyperacetylation of the histone tails of all core histones. Another modification that is associated with active genes is the methylation of lysine 4 of histone H3 (H3K4) 1 (2)(3)(4). An interesting feature of histone methylation is the fact that, although the same enzyme modifies the histone tails, H3K4 residues can be mono-, di-, or trimethylated, and such differences are of functional consequence. It was recently shown that di-and trimethylated lysines, but not monomethylated lysines, bind the chromatin-remodeling complex ISWI with high affinity (5). In turn, it was shown that the presence of any H3K4 methyl mark inhibits the binding of heterochromatic silencing factors (6). In this context several observations are noteworthy. In yeast it was shown that the Set1 methyltransferase enzyme that modifies H3K4 is recruited by the RNA-polymerase II complex and thus leaves a trail of trimethylated histone behind even when transcription has ceased (7). This is also seen in higher eukaryotes (8). Interestingly, the K4-dimethyl mark seems to be regulated differently, because this appears also on non-transcribed genes and may indicate a poised chromatin state. Little is known about the role of H3K4 monomethylated histones at genes of higher eukaryotes and the functional significance of this modification.
Our laboratory has been investigating the interplay between developmental gene locus activation, active histone marks, and transcription factor occupancy. As a model system we have adopted the chicken lysozyme locus, which is expressed in myeloid cells. The chicken lysozyme gene is located within a generally DNase I-sensitive chromatin domain that spans 24 kb (9, 10) and contains two differentially regulated genes in close proximity: the highly tissue-specific clys gene and the ubiquitously expressed gas41 gene (11). However, a high level of general DNase I sensitivity at both genes is only observed in lysozyme-expressing cells, and the reason for this puzzling finding is not clear. Constitutive expression of the gas41 is driven by a CpG island with dual origin/promoter function. Lysozyme gene expression in macrophages is controlled by at least five tissue-specifically active cis-regulatory elements, three enhancer elements situated 6.1, 3.9, and 2.7 kb upstream of the transcription start site, a silencer element at Ϫ2.4 kb and a complex promoter (12). Priming of chromatin structure as indicated by selective demethylation of DNA, a partial accessibility to transcription factor binding, and changes in DNA topology occurs already in multipotent precursor cells, which do not yet express the gene (13)(14)(15). The mechanistic basis for this observation is unknown. Lysozyme gene expression is first observed in granulocyte-macrophage precursors and in mature macrophages is rapidly induced by pro-inflammatory stimuli such as bacterial lipopolysaccharide (LPS). As judged by DNase I-hypersensitive site (DHS) mapping, enhancer and promoter elements are only fully active in LPS-treated mature macrophages where maximum level gene expression is observed. At each successive step of the developmental activation of the lysozyme locus, levels of histone H3 (K9 and K14) acetylation at the enhancers are increased and levels of histone H3 (K9) methylation are decreased (14). However, up to date, H3K4 methylation levels were not studied and the dynamics of histone methylation levels at the cis-elements regulating the two juxtaposed gene loci is not known. In the study presented here, we addressed this issue and we also examined how histone methylation levels and chromatin architecture of the entire locus respond to the induction of differentiation and high level expression by external signals. We studied dynamic alterations in chromatin accessibility, linker histone distribution, and the histone methylation status within the lysozyme chromatin domain in lysozyme non-expressing multipotent precursor cells as well as BM2 cells representing monoblast cells, which represent committed macrophage precursor cells. The latter express the lysozyme gene at a low level, can be differentiated into monocytes by treatment with PMA, and can be further differentiated into macrophages by LPS stimulation. Both stimuli are required for maximal lysozyme gene expression. Monomethylation of histone H3K4 in the lysozyme regulatory region is found in undifferentiated monoblasts. Cell differentiation and induced gene expression leads to high levels of H3K4 trimethylation over the coding region. Interestingly, already in lysozyme non-expressing precursor cells histone H1 is depleted at the promoter and is further depleted in a locus-wide fashion in response to PMA, but not LPS treatment.
Reverse Transcription-PCR-Total RNA was prepared using TRIzol (Invitrogen) according to the manufacturer's instructions, and genomic DNA was removed by DNase I treatment. First-strand cDNA synthesis from RNA samples was carried out using oligo(dT)15 primer and Moloney murine leukemia virus reverse transcriptase (Invitrogen) in a reaction volume of 20 l. PCRs were performed using real-time quantitative PCR (ABI Prism 7700 sequence detection system, PerkinElmer Life Sciences) with SYBR green. Relative expression was calculated as a ratio of lysozyme or gas41 to ␤-actin. Primers were designed using Primer Express 1.5 software and primer sequences for the different genes are shown in Supplemental Table 2B.
In Vivo Footprinting Analysis-In vivo dimethyl sulfate (DMS) treatment was performed by incubating cells in 0.2% DMS in phosphate-buffered saline for 5 min. Cells were washed with phosphatebuffered saline and lysed in lysis buffer (100 mM Tris, pH 8.0, 40 mM EDTA, 2% SDS, 200 g/ml proteinase K) overnight at room temperature, and genomic DNA was purified by treatment with RNase, organic extractions, and ethanol precipitation. DMS-modified guanines were then cleaved by treatment with piperidine at 0.1 M for 10 min at 90°C, followed by three lyophilizations in water. DNase I treatment was performed as follows: suspension cells were spun down, whereas the adherent differentiated BM2 cells were first detached with trypsin and resuspended in phosphate-buffered saline containing 0.25 mM phenylmethylsulfonyl fluoride. Cells were subsequently washed with ice-cold buffer (11 mM KPO 4 , pH 7.4, 108 mM KCl, 22 mM NaCl, 5 mM MgCl 2 , 1 mM CaCl 2 , 1 mM dithiothreitol) and resuspended at 10 8 cells per ml in buffer containing 1 mM ATP. To 100 l of this cell suspension, 104 l of ice-cold buffer, consisting of 77% buffer containing 1 mM ATP, 0.38% Nonidet P-40, and various amounts of DNase I (Worthington, Lakewood, NJ) was added. After 6 min at room temperature, the reaction was stopped by the addition of 200 l of lysis buffer and incubated overnight, and DNA was purified. In vitro DNase I samples were generated by essentially the same procedure on purified genomic DNA, except that lower amounts of DNase I were used and incubation was performed on ice for 3 min. For all samples, 1 g from DMS-treated DNA and 1.5 g from DNase I-treated DNA were used for LM-PCR. LM-PCR on DMS-treated and DNase I-treated DNA was performed exactly as described previously (14,15,18). Briefly, a primer extension reaction with biotinylated primers was carried out using Vent exo-DNA polymerase (New England Biolabs). Primer extension products were then ligated to linker LP25-21, which consists of a 25-mer (GCGGTGACCCGGGAGATCT-GAATTC) annealed to a 21-mer (GAATTCAGATCTCCCGGGTCA).
After ligation, specific biotinylated products were bound to streptavidin paramagnetic beads (DYNAL) and washed as described by the manufacturer and then subjected to PCR amplification using a nested specific primer and the linker primer (LP25). Primer extension and PCR conditions were optimized for each primer set using a robotic workstation (Biomek 2000, Beckman) and a gradient PCR machine as described in precise detail in a previous study (19). PCR products were labeled by primer extension using ␥-32 P-labeled nested primers and were analyzed on 6% denaturing polyacrylamide gels. To analyze the lysozyme promoter, LM-PCR was performed using gene-specific nested primers P1-P3 and P351-P353. LM-PCR on MNase-treated material was performed on a robotic workstation as described in detail before (19). Briefly, DNA obtained as input samples for ChIP was phosphorylated with polynucleotide kinase and directly ligated to the linker, LP25-21. Primer extension was performed by using biotinylated genespecific primers. Primer extension products were bound to magnetic beads and then amplified by LM-PCR as described above. Primers P1-P3 and P351-P353 and ␤-actin promoter-specific primers are shown in Supplemental Table 2A.
Chromatin Immunoprecipitation Assays and Real-time PCR Analysis-Chromatin immunoprecipitation using sonicated material was performed exactly as previously described (14). Chromatin immunoprecipitation assays using MNase treated material is described before (20). After preclearing, 50 l of the chromatin preparation was kept as control (Input), and 500 l was incubated with 1 mg of rabbit IgG (Upstate Biotechnology, 12-370) to estimate the background level for each amplicon (data not shown), anti-histone H3 C-terminal (Abcam ab1791), anti-histone H3 1me-, 2me-, or 3me-Lys-4 (Abcam ab8895, ab7766, and ab8580), histone H1 (Abcam ab7789 -250) or anti-histone H3 1me-or 2me-Lys-9 (Abcam ab9045 and ab7312) antibodies. Primers were designed using Primer Express 1.5 software and tested against a standard curve of sonicated genomic DNA. Primer sequences are shown in Supplemental Table S1. PCR signals from immunoprecipitated material were first calculated in percentage of the Input for all primers. To correct for the variation of the quality of different batches of chromatin from different cells and for varying background signals generated by different antibodies, signals were normalized to those obtained with internal control primers such as the hepatocyte-specific gene VLD-Lapo2 primers or primers specific for the coding region of the ␤-actin gene. Non-normalized data are shown in Supplemental Fig. S2 for five different primers, including primers specific for the control genes.

Maximal Expression of the Chicken Lysozyme Gene Requires
Both PMA-mediated Monocytic Differentiation and LPS Signaling-In previous studies we measured transcription factor occupancy and chromatin structure in retrovirally transformed cell lines representing different stages of myeloid differentiation (16,17,21) and identified the transcription factors binding to the different cis-regulatory elements by chromatin immunoprecipitation (ChIP) assays and in vivo footprinting experiments. The results of these experiments are summarized in Fig. 1A. Fig. 1B depicts the cell models used in this study. These include multipotent progenitor cells (HD50 MEP) as well as HD37 representing a non-macrophage cell type (erythroblasts). HD50 myl cells represent bi-potent granulocyte-macrophage precursors. BM2 cells are v-myb-transformed committed macrophage precursors (monoblasts). BM2 cells grow in suspension and differentiate into adherent monocytes after treatment with PMA and can be further differentiated to activated macrophages after treatment with LPS. We measured lysozyme and gas41 mRNA expression in the different cell types by real-time PCR (Fig. 1C). gas41 was expressed at low level in all cell types analyzed. In contrast, lysozyme mRNA was not detected in HD37 erythroblasts and HD50 MEP progenitor cells. Un-stimulated BM2 cells showed a low level of expression, which was not significantly increased by treatment with PMA alone, whereas stimulation with LPS alone up-regulated mRNA levels. Treatment with both agonists leads to a synergistic increase in mRNA levels, indicating that both signals are required for high levels of lysozyme gene expression.
Developmental Regulation of Histone H3 Lysine 9 and Lysine 4 Methylation Patterns-We next examined the distribution of histone H3K9 methylation and H3K4 methylation across the lysozyme chromatin domain. Using sonicated chromatin, we had previously measured the distribution of Histone H3K9 methylation by ChIP using an antibody that did not distinguish between mono (1me)-, di (2me)-, or tri (3me)-methylated K9. These experiments showed that H3K9 methylation levels were high in HD37 erythroblasts and HD50 MEP progenitor cells, but dramatically decreased in mature macrophages (14). Here we asked the question at which stage of macrophage differentiation the H3K9 methyl mark was lost, and we also examined whether H3K9 methylation levels were altered during the differentiation of BM2 monoblasts to activated macrophages. This question is of importance, because the H3K9 methyl mark is very stable (22) and PMA-treated differentiating cells cease proliferating (data not shown). To answer these questions we used antibodies against 1me, 2me, and 3me Lys-9. To increase resolution without having to purify mononucleosomes, we prepared cross-linked chromatin from the indicated cell types (HD37, HD50 MEP, untreated BM2 cells, and BM2 cells treated with LPS, PMA, and PMA plus LPS) and then digested such chromatin preparations with micrococcal nuclease. It has been shown before that this avoids nucleosome movement during the preparation of native chromatin while generating chromatin fragments of mostly nucleosome length (23) (an example for such a digestion profile can be seen later in Fig. 5D). To be able to normalize signals and to precisely measure the relative representation of each amplicon, this material (input) was subjected to a real-time PCR analysis using 13 primer pairs across the lysozyme chromatin domain, plus different internal control primers. The results of a ChIP experiment with an antibody against H3 1meK9 are depicted in Fig. 2A and show that histones in the lysozyme cis-regulatory elements have already lost most of the H3 1meK9 mark in undifferentiated BM2 cells. The same results were obtained with an antibody against the H3 2meK9 (data not shown), and methylation levels remained low during further differentiation. Significant histone H3K9 methylation was still observed between cis-elements at all stages of differentiation. We could not detect any H3 3meK9 anywhere across the locus (data not shown). No histone H3K9 methylation was seen at the gas41 CpG island, whereas the histones at the promoter of the liver-specific VLDLapo2 gene, which is inactive in all examined cell types, were highly meth- ylated at H3K9 (Supplemental Fig. S2D).
We next analyzed H3K4 methylation and performed ChIP assays with antibodies against 1me, 2me, and 3me H3K4. No H3 1meK4 was observed anywhere across clys in lysozyme non-expressing cells (Fig. 2B), but it was present across the entire 5Ј regulatory region in lysozyme-expressing cells (even at the far-upstream amplicon at Ϫ10 kb), and increased slightly upon cell differentiation. H3 1meK4 levels in the lysozyme coding region were low, and it was absent in the gas41 gene and absent in the 5Ј coding region of the ␤-actin gene (data not shown). H3 2meK4 and 3meK4 patterns were similar to each other, but different to the H3 1meK4 patterns (Fig. 2C). Interestingly, some H3 3meK4 could be found in HD50 MEP cells, but not in HD37 cells, and levels increased during cell differentiation at and in between the cis-elements. H3 3meK4 levels in the coding regions increased parallel to the increase in mRNA levels, which were consistently higher in cells treated with LPS (Fig. 1C). Irrespective of the cell type, H3 3meK4 levels at the gas41 CpG island were significantly higher than in the lysozyme gene and within the lysozyme gene were highest in the coding region.
Macrophage Maturation Leads to Site-specific Alterations in Chromatin Architecture and a Gene-wide Depletion of Histone H1-In macrophages, the chicken lysozyme locus is organized in a DNase I-sensitive domain that also harbors the gas41 gene (10). We have previously mapped nucleosome positions at low resolution using micrococcal nuclease digestion and have observed a significant reorganization of nucleosomal architecture at specific cis-regulatory elements after the activation of gene expression (24). In the experiments described here we wanted to further investigate this finding and also test whether the increase in general DNase I sensitivity goes along with a loss of the linker histone H1. The results of these experiments are summarized in Fig. 3.
In the ChIP experiments described above we noticed that not all amplicons were equally represented in the input fraction, indicating significant differences in MNase accessibility and chromatin architecture across the lysozyme locus. Fig. 3A depicts an analysis of the relative representation of input material of cross-linked and MNase-digested chromatin as measured by real-time PCR. From the results depicted in Fig. 3A it is obvious that sequences within the DNase I-hypersensitive cores of the enhancers and the promoter, but not the flanking or coding regions, were becoming progressively under-represented in the Input material the higher the gene was expressed (compare amplicons depicted in white or black to amplicons depicted in gray). The promoter was highly MNase-sensitive over more than 100 bp as measured by looking at two juxtaposed amplicons (Ϫ75 to Ϫ156 bp (black) and Ϫ190 to Ϫ255 bp (white)). Interestingly, this under-representation was not seen at the gas41 CpG island, even though this element harbors a constitutive DHS. Control experiments using amplicons in the inactive VLDLapo2 promoter or the active ␤-actin gene did not yield differences in relative amplicon representation between cell types (Supplemental Fig. S2A).
Because it has been documented that MNase can generate single strand cuts on a nucleosomal template (25), the question arose whether this loss of material in the input fraction was due to an increased accessibility within nucleosomes or to nucleosome loss. To this end, we prepared chromatin by sonication and performed a ChIP assay with an antibody against the C terminus of histone H3 (Fig. 3B). Here, input signals for specific fragments were similar for all cell types (Supplemental Fig. S1). This analysis yielded a different picture to that of MNase-digested chromatin. Progressive and significant nucleosome loss as indicated by a decrease in precipitated chromatin with increased expression level was observed at the Ϫ6.1-kb enhancer and the Ϫ2.7-kb enhancer. This was less prominent at the other cis-elements or was not observed at all (including the VLDLapo2 promoter (Supplemental Fig. S2B)). Interestingly, although we observed a strong reduction of input signal at the promoter after MNase digestion (Fig. 3A), with the same amplicon no difference in H3 content was observed between cell types. This indicated that, although histone H3 was partly removed from a subset of enhancers, the promoter and other elements retained their nucleosomal structure. The reduction in input signal at these elements therefore had to originate from an enhanced accessibility of these nucleosomes to nuclease digestion.
Our analysis of linker histone distribution in the different cells types is described in Fig. 3C. These experiments showed that H1 levels at the gas41 CpG island were consistently low. Except for the far-upstream amplicon, all regions of the lysozyme locus showed a progressive reduction of H1 signals in lysozyme-expressing cells with cell differentiation, which was most prominent over the cis-regulatory elements, and greatest at the promoter. However, except for the lysozyme promoter, which lost H1 occupancy very early in differentiation, it was clear that efficient linker histone removal required PMA treatment. The signals in cells treated with LPS alone were consistently higher than those from cells treated with PMA only. Another interesting observation was that, although HD50 MEP progenitor cells do not express the lysozyme gene, they have already clearly reduced levels of H1 across the locus. This was particularly prominent over the promoter, but the same trend was visible at other cis-regulatory elements as well.
The Stimulation of Lysozyme Gene Expression Does Not Alter the Degree of Transcription Factor Binding at the Lysozyme Promoter but Leads to Significant Changes in Chromatin Architecture-We next wanted to uncover the reason for the increased MNase accessibility at the lysozyme promoter and assayed transcription factor occupancy and nuclease accessibility by in vivo footprinting. DMS in vivo footprinting showed no difference in transcription factor occupancy between unstimulated and stimulated BM2 cells (Fig. 4A). As shown previously (13), HD37 cells and HD50MEP cells gave a pattern similar to naked DNA, indicating that no transcription factors were bound to the promoter or anywhere else (data not shown). The DNase I in vivo footprints show little difference between HD37 and MD50 MEP cells, although the pattern differs from naked DNA (Fig. 4B). In contrast, unstimulated and fully differentiated PMA-and LPS-treated BM2 cells showed a number of dramatic alterations in the chromatin structure at the promoter. In concordance with our DHS mapping studies the high resolution experiments demonstrated that promoter chromatin in lysozyme-expressing cells became highly nuclease-accessible. Strong DNase I-hypersensitive sites and protections from DNase I cleavage were observed at the known transcription factor binding sites. Again, pattern and intensity of footprints did not differ between unstimulated and stimulated BM2 cells. Our data therefore show that there was no alteration in the degree of transcription factor occupancy at the promoter as a result of differentiation of BM2 cells.
Our low resolution MNase mapping experiments had previously shown the presence of an array of regular MNase cuts over the promoter and downstream (one such MNase cleavage site is indicated in Fig. 5B). This pattern did not significantly change during cell differentiation, but we had observed an increased cleavage frequency (24). This agrees with the H3 C terminus ChIP assays described here (Fig. 3B). To examine nucleosome boundaries at nucleotide resolution, we used the same MNase-treated material as for ChIP assays (Fig. 5D) to  FIG. 3. A, representation of specific amplicons in input DNA after MNase treatment to prepare the chromatin assayed in Fig. 2. Relative representation of amplicons were calculated relative to the control amplicons depicted in Supplemental Fig. S2A. ChIP assay with sonicated chromatin measuring histone H3 using the H3 C terminus antibody (B) and histone H1 (C) during myeloid differentiation, and results were expressed as IP/Input versus IPc/Inputc. MNase-treated DNA used as a control or used antibody are indicated on the left. The relative representation of amplicons in the Input material from sonicated chromatin is shown in Supplemental Fig. S1. Results represent the average of at least two independent chromatin preparations assayed in triplicate. For all other details see legend of Fig. 2. perform a double strand-specific LM-PCR. The results of this experiment are depicted in Fig. 5 (A and B). As a control for equal digestion efficiency we assayed the ␤-actin promoter using the same material (Fig. 5C). It is obvious that the distribution of MNase double strand cuts differs between lysozyme non-expressing cells and -expressing cells. In lysozyme nonexpressing cells the main transcriptional start sites are significantly protected from MNase digestion, indicating the presence of a specifically positioned nucleosome with a boundary at the Sp1 site at Ϫ80 bp. The pattern was different with lysozyme-expressing cells and did not change during cell differentiation. It is clear that DNA over an extended region becomes highly accessible to MNase cleavage once transcription factors have stably bound, which coincides with the DNase I-hypersensitive region. Taken together, these experiments suggest that the promoter contains a nucleosome that is extensively remodeled and highly accessible to nuclease digestion. DISCUSSION Our studies demonstrate extensive alterations in chromatin modification, chromatin composition, and chromatin accessibility within the lysozyme gene, but not the gas41 gene, during macrophage differentiation, some of which are induced by external signaling. This indicates that, although in very close vicinity, both regulatory domains are quite separate and do not interact with each other. The mechanism mediating this independence in regulation is currently under investigation. However, in one respect the two genes appear to influence each other: There is no change in the intra-nuclear localization of the lysozyme locus with respect to the localization of centromeric heterochromatin after gene activation, which is most likely caused by the close vicinity of the active gas41 locus (26).
Depletion of Histone H1 Is an Early Mark for clys Activation, and Macrophage Differentiation Leads to Extended Alterations in Chromatin Architecture and Nuclease Accessibility-We have previously shown that the chromatin in multipotent progenitor cell lines (HD50 MEP cells) is partially reorganized and transiently accessible to the binding of transcription factors, although in these cells no stable transcription factor complexes and DHS are formed, and they do not contain hyperacetylated histones (13,14). Here we show that such cells contain a significantly reduced amount of histone H1 at the lysozyme promoter as compared with cells from an alternative lineage (HD37), and the same trend, although less dramatic, is observed at the other cis-regulatory elements as well (Fig. 3C). This result indicates a reduction in chromatin compaction and could account for some of the changes in DNA topology we have seen before (13). It is known that certain transcription factors, such as the retinoic acid receptor or the winged-helix factor HNF3 (a FOX family member) can lead to a displacement of histone H1 at specific sequences (27,28). It is therefore tempting to speculate that the transient interaction of transcription factors in myeloid precursor cells leads to a reduced chromatin compaction via a partial displacement of H1.
We also found that PMA-induced macrophage differentiation, but not LPS induction alone leads to a significant further depletion of H1 from lysozyme cis-regulatory elements, although for maximum depletion both signals are needed. This and the fact that some reduction of H1 is already seen in multipotent precursor cells, indicates that it is the differentiation process itself and not high level gene expression that is responsible for H1 depletion. More elaborate experiments outside of the scope of this study using specific signal transduction inhibitors to block specific signal transduction pathways combined with ChIP assays will hopefully indicate which transcription factors drive this process.
All lysozyme cis-regulatory elements (but not the flanking regions) became progressively more nuclease accessible with increasing gene expression, as indicated by an under-representation of specific amplicons in the input fractions. For some elements, but not all, this correlated with the destabilization of nucleosome binding and nucleosome loss as measured by ChIP analysis with a H3 C-terminal antibody using sonicated chromatin. Such nucleosome loss is a genome-wide phenomenon (29), which has been carefully studied (30,31) and indicates that care has to be taken when interpreting negative ChIP data. It also indicates that what is precipitated may be a transiently modified nucleosome population that is destined for eviction, as indicated previously (32). The lysozyme promoter clearly behaves differently. We show that although there are no detectable differences in nucleosome content (Fig. 3B), promoter structure becomes more and more nuclease-accessible with increasing transcription as measured by double strandspecific LM-PCR and real-time PCR, which indicates also the presence of single strand cuts (Fig. 3A). Curiously, our previous low resolution mapping experiments did not indicate major differences in nucleosome positions at the promoter between lysozyme expressing and non-expressing cells (24). However, although we can make statements of the precise position of nucleosomal linker regions in lysozyme non-expressing cells, this enhanced accessibility makes it very difficult to map precise nucleosomal boundaries in lysozyme-expressing cells by high resolution methods. The lysozyme promoter is highly inducible, and it is possible that weakened nucleosome-DNA contacts are a prerequisite for the rapid induction of gene expression, very similar to what has been observed with other inducible systems (33)(34)(35). and DNase I reactivity are indicated as gray bars. Only alterations that were seen in more than one experiment are highlighted. In B multiple alterations in the DNase I digestion patterns were observed in lysozymeexpressing cells, which are not further highlighted.
cis-elements, but remain relatively high between the elements. The inverse is observed with H3 1meK4, H3 2meK4, and H3 3meK4 levels. Two observations are noteworthy: the active lysozyme locus contains all three types of H3K4 methylation. In contrast, the gas41 CpG island only contains di-and trimethylated H3K4, which are also at much higher level compared with the lysozyme locus. Second, there are differences in the distribution of mono-and di/trimethylated K4 across the lysozyme gene. Monomethylated H3K4 is mainly seen in the regulatory region, whereas di/trimethylated H3K4 increases with transcription and is found in both the coding and the regulatory region. We speculate that this distribution also reflects the inducible nature of the lysozyme gene. We have previously shown that all enhancer complexes interact with the histone acetyltransferase CBP in lysozyme-expressing cells and that LPS induction leads to an increase in histone H3 acetylation without an increase in histone acetylase recruitment (14), indicating that outside signals are required for maximum histone modification. This strict correlation of histone acetylation and lysozyme mRNA levels is also true for differentiating BM2 cells (data not shown). Because the binding of active chromatin remodeling complexes is only promoted by H3 2me and 3meK4 and not by H3 1meK4 (5), our data support a model in which the transcription factors already bound to the cis-regulatory elements recruit a H3K4 methyltransferase activity that marks the lysozyme locus in undifferentiated monoblasts. This would block gene silencing (6), but would not promote chromatin remodeling. High nuclease accessibility and significant nucleosome eviction are only found in fully stimulated cells.  (24) and in this study are indicated by different size brackets. The position of the promoter nucleosome in lysozyme non-expressing cells is indicated as half-oval shape. The main transcription start sites and their relative use as mapped in (36) are indicated as arrows. C, control LM-PCR of the ␤-actin promoter on the MNase-treated material. Note that the gel is not shown in scale: the gel had the same original size as the gels shown in B, showing that there is no bias against longer reaction products in lysozyme-expressing cells, but the image was squashed flat to fit it into the figure. The relative sizes of the fragments are indicated. D, DNA purified from MNase-treated nuclei prepared from the indicated cell types loaded on ethidium bromide stained agarose gel. M, marker.