ChIP Microarray (Robyr et al. 2004 Methods Enzymol. 376, 289-304)

Analysis of genome-wide histone acetylation state and enzyme binding using DNA microarrays

Daniel Robyr, Siavash K. Kurdistani, and Michael Grunstein*

Department of Biological Chemistry, UCLA School of Medicine and the Molecular Biology Institute, Boyer Hall. University of California, Los Angeles, CA 90095

* Corresponding author: Michael Grunstein

Running title: Acetylation microarray

Introduction

Nuclear events such as gene transcription, DNA replication and DNA repair must cope with local chromatin environments that may be inhibitory to these processes. Therefore, chromatin remodeling and histone post-translational modifications are an intrinsic component of the maintenance and establishment of repressive or permissive chromatin states. Histone acetylation and deacetylation play an important role in this regulation.
The small genome of Saccharomyces cerevisiae contains numerous histone deacetylases (HDACs) including Rpd3, Hda1, Hos1, Hos2, Hos3 and Sir2. Chromatin immunoprecipitation using highly specific antibodies raised against individual lysine residues (1) has established that several of these enzymes have different histone specificities. For instance Hda1 deacetylates histones H3 and H2B (2) whereas Rpd3 strongly affects all core histones sites analyzed with the notable exception of H4 lysine 16 (1). In contrast, Hos2 is required for H3 and H4 deacetylation only (3). To function at DNA regulatory elements HDAC complexes are recruited to their target promoters by specific transcriptional repressors such as Ume6 at the Rpd3-affected INO1 gene leading to the local deacetylation of about two nucleosomes around the TATA-box (4, 5). However, Rpd3 and Hda1 also deacetylate large regions of chromatin, including promoters and open reading frames, without apparent direct recruitment by DNA binding repressors (6). This global deacetylation mechanism is involved in the rapid establishment of a global repressive chromatin environment once gene transcription is turned down. Finally, Hos2 primarily deacetylates histones within the coding regions of genes. Surprisingly, unlike other histone deacetylases that are repressors, Hos2 is required directly for gene activity (3). These observations suggest that a comprehensive understanding of HDAC functions requires the analysis of large regions of chromatin, ideally at a genome wide level.
HDAC locus specificity was initially probed by analyzing the influence of each HDAC deletion on genome-wide transcription using DNA microarrays (7, 8). RPD3 disruption, however, led to more genes being repressed than up-regulated genome-wide most likely as a result of indirect effects on global gene regulation. Thus, determination of HDAC function genome-wide benefits from the additional studies of acetylation (acetylation arrays) and enzyme binding (binding arrays) as more direct tools to assess HDAC locus specificity. In this manner, a much more comprehensive analysis can determine the sites at which the enzyme binds, affects acetylation state and results directly in changes in gene activity (9, 10). We describe here methods for identifying the sites of enzyme action using acetylation microarrays and enzyme binding. We have recently used such procedures to unravel the sites of action for Rpd3 and Hos2 (3, 9, 10) and acetylation arrays for studies on Hda1, Hos1, Hos3 and Sir2 (10).

Acetylation microarrays use the combination of chromatin immunoprecipitation (11) and hybridization of DNA to microarray glass slides (Fig. 1) (12, 13). Highly specific antibodies are used to immunoprecipitate formaldehyde-crosslinked chromatin fragments enriched for a given acetylated lysine residue in cell lysates obtained from a wild type (WT) strain and its isogenic strain disrupted for the HDAC of interest. The cross-link is reversed after chromatin immunoprecipitation allowing DNA purification, amplification by PCR and DNA labeling with fluorophores. One fluorescent dye (e.g. Cy5) is used for DNA recovered from the strain carrying the deacetylase gene mutant whereas a second dye (e.g. Cy3) is used to label DNA from the WT isogenic control strain. The labeled DNA probes from both strains are combined and hybridized onto DNA microarray glass slides containing either intergenic regions or open reading frames (ORFs) or both. Glass slides are scanned for both fluorescent dyes and the normalized ratios of intensities between the deacetylase mutant and the wild type probes reflect changes in acetylation in the mutant strain.

Chromatin immunoprecipitation (ChrIP or ChIP).

1. Dilute an over night pre-culture of yeast into 50 ml YEPD medium (2% peptone, 1% yeast extract, 2% dextrose) to A600 = 0.2 and allow the cells to reach A600 ~ 1. Histones are crosslinked to DNA in vivo by adding formaldehyde (Fisher) to the culture to a final concentration of 1% (w/v). The crosslinking reaction is carried out at room temperature for 15 min with constant mild agitation and is then quenched by adding 2.5 ml 2.5 M glycine (final concentration 125 mM).

2. Harvest and wash cells twice in ice-cold 1x PBS (140 mM NaCl, 2.5 mM KCl, 8.1 mM Na2HPO4, 1.5 mM KH2PO4, pH7.5) by centrifugation at 4? at 2800g (Beckman J2-HC, rotor JS-4.3). The cell pellet can be frozen in liquid nitrogen and stored at –80? if needed.

3. Resuspend cells in 400 ul ice-cold lysis buffer (50 mM HEPES/KOH pH 7.5, 500 mM NaCl, 1 mM EDTA, 1% (v/v) Triton X-100, 0.1 % (v/v) SDS and 0.1 % (w/v) sodium-deoxycholate) complemented with proteases inhibitors (Complete™, Roche; 100x stock prepared in 500 ul H2O from 1 tablet). Transfer the resuspended cells into siliconized tubes (1.7 ml NoStick Hydrophobic Microtubes, GeneMate®) and add one volume of acid-washed glass beads (0.45-0.52 mm diameter, Thomas Scientific) (14). Cells are lysed on an Eppendorf shaker (model 5432) between 45 and 60 min at 4?. Puncture the bottom of the opened tubes containing the lysed cells with a red hot 25G1 needle (Becton Dickinson). Immediately close the tubes, place them on a fresh collecting 1.7 ml tube and recover the cell lysate by centrifugation (5 sec in an Eppendorf tabletop centrifuge (model 5415 C) at 14000 rpm).

4. Shear chromatin from the cell lysate by sonication down to an average fragment size of 500 bp using a Sonic dismembrator 550 (Fisher Scientific) with a Microtip‚Ñ¢ model 419 from Misonix Inc. Sonicate lysate on ice with two pulses of 15 sec each (magnitude setting of 4) and a 60 sec rest interval. Recover the clear cell lysate from the cell debris (pellet) by centrifugation at 4? in an Eppendorf centrifuge (10 min at full speed). At this point the lysate can be stored at -80? for up to a month.

5. Immunoprecipitate acetylated chromatin fragments over night at 4? on a nutator incubating 50 ul of cell lysate with 2-5 ul of a given antiserum (see below). Then, add a suspension of 25ul of 50 % (v/v) protein A sepharose‚Ñ¢ CL-4B beads (Amersham-Pharmacia), equilibrated in lysis buffer and incubate an additional 2 hr.

6. Pellet the sepharose beads for 30 sec at room temperature in an Eppendorf centrifuge (model 5415 C) at 735 g (3000 rpm). Discard the supernatant and wash successively for 5 min the beads on a nutator with 500 ul of the following solutions: twice in lysis buffer, once in deoxycholate buffer (10 mM Tris-HCl pH 8.0, 0.25 M LiCl, 0.5 % (v/v) NP-40, 0.5 % (w/v) sodium deoxycholate and 1 mM EDTA pH 8.0) and once in TE pH 8.0. Pellet the beads and discard the supernatant between each washing step as indicated above. The whole procedure is carried out at room temperature.

7. Elute the immunoprecipitated chromatin fragment from the beads and reverse crosslink overnight at 65? with 50 ul elution buffer (50 mM Tris-HCl pH 8.0, 10 mM EDTA pH 8.0 and 1% (v/v) SDS).

8. Add 50 ul of TE pH 8.0, 20 ug glycogen and treat with 20 ug proteinase K for 2 to 3 hr at 55?. Finally, extract DNA with 1 volume phenol/chloroform/isoamyl alcohol [25 : 24 : 1 (v/v)] and ethanol precipitate. Resuspend purified DNA into 50 ul TE pH 8.0 and store at –20?.

Troubleshooting.

The antibody specificity is the most critical aspect for any chromatin immunoprecipitation. The polyclonal antibodies raised against individual acetylated lysine residues were prepared in our laboratory and are available at Upstate (http://www.upstate.com). Their specificity was verified by ELISA assay and tested by ChrIP against histones mutated for the acetylatable lysine (1). The titration of the antibody amount required for an immunoprecipitation has to be determined experimentally for all antibodies.
Sonication conditions will determine the resolution of the chromatin immunoprecipitation (the higher the average fragment size is, the lower the resolution will be). Shearing efficiency will also depend on the sonicator brand. A pilot experiment should be performed in order to find the appropriate sonication settings. After sonication add 80ul elution buffer (see step 7 for recipe) to a 20 ul aliquot from the cleared cell lysate (step 4) and incubate the tube over-night at 65?. Add 100 ul TE pH8.0, 40 ug proteinase K and incubate the tubes for 2 to 3 hr at 55? as indicated above (step 8). Resuspend DNA pellet (Input DNA) in 20 ul TE pH 8.0 after DNA extraction and precipitation. Treat an aliquot (10 ul) with RNase A (10 ug) 30 min at 37? and analyze the average DNA fragment size on a 1.5 % agarose gel. Alter accordingly the sonication settings if DNA fragments are too large. Store the remaining RNase non-treated DNA at –20? for later PCR amplification if needed (see later).

Double crosslinking with protein-protein crosslinking agents and formaldehyde

For certain proteins, such as the histone deacetylase Rpd3, formaldehyde crosslinking alone is inadequate for efficient crosslinking of the enzyme to chromatin in vivo (9). This may be due to the fact that Rpd3 is part of a large (~1 MDa) multiprotein complex and, unlike histones, may lie too far from DNA for efficient crosslinking by formaldehyde alone. The immunoprecipitation efficiency of Rpd3 is significantly improved when in addition to formaldehyde, a protein-protein crosslinking agent is also used (9). In such double crosslinking scheme, the cells are sequentially treated first with a protein-protein crosslinking agent and then with formaldehyde. We have successfully used several protein-protein crosslinking agents such as dimethyl adipimidate (DMA), dimethyl pimelimidate (DMP) or 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) prior to formaldehyde treatment to ChrIP the Rpd3 deacetylase. This new crosslinking method has been described recently in more detail for various protein-protein crosslinkers (15) and is summarized below.

1. Grow cells as indicated above but do not add formaldehyde immediately when they have reached A600 = 1.

2. Pellet and wash the cells twice with 20 ml ice-cold 1x PBS in a 50 ml Falcon tube. Prepare a fresh 15 ml solution of 10 mM dimethyl adipimidate (DMA, Pierce) in ice-cold 1x PBS containing 0.25% (v/v) dimethyl sulfoxide (DMSO). Since DMA reacts with amine groups, buffers such as Tris that contain primary amines should be avoided. Non-amine containing buffers such as Phosphate, Carbonate and Hepes are acceptable. Resuspend the cells in the DMA solution and incubate on a nutator at room temperature for 45 min. Pellet and wash the cells once again in 1x PBS and resuspend in 1x PBS containing 1% (w/v) formaldehyde. Incubate the cells on a nutator at room temperature for 10 to 11 hr. This incubation time is necessary for maximal crosslinking of Rpd3 but needs to be empirically determined for the protein under study. Stop the reaction by adding 2.5 ml glycine 2.5 M and wash the cells again twice in ice-cold 1x PBS.

3. Cell lysate is prepared as described earlier with the exception of a salt concentration of 150 mM NaCl in the lysis buffer (150mM-lysis buffer). DNA enrichment during chromatin immunoprecipitation may be further improved by partially purifying chromatin prior to sonication. To do this, pellet chromatin by centrifuging the crude extract for 3 min at 4? at 14000 rpm. Discard the supernatant and carefully resuspend the chromatin pellet in 400 ul 150mM-lysis buffer.

4. Due to the longer formaldehyde crosslinking time, increase sonication time to two pulses of 20 sec each (magnitude setting of 4) with a 60 sec rest interval for more efficient shearing of chromatin.

5. ChrIP is then carried out essentially as described earlier with the following modifications. 100 ul of the lysate (WT and rpd3?) was incubated with 5 ul of ?-Rpd3 polyclonal antibody (Upstate) and incubated on a nutator at 4°C for 2 hrs. After 1 hr incubation with 50 ul a 50% v/v suspension of protein A beads, the beads are transferred to a 0.45 um filter unit (Millipore Ultrafree-MC; cat#UFC30HV00) for the washing steps (2x in lysis buffer containing 500 mM NaCl and 2x in deoxycholate buffer, 15 min each). After each wash, the filter unit was spun at 3,000 rpm for 30 sec, and the flow through fraction was discarded To elute chromatin off the beads, 100 ul elution buffer was added to the filter unit, incubated at 65°C for 15 min and spun at 3000 rpm for 30 sec and repeated once more. The eluates were combined and incubated at 65°C overnight. DNA is then recovered as described.

Probe amplification by PCR

Low DNA yield after chromatin immunoprecipitation is not adequate for immediate labeling and hybridization onto DNA microarrays, thus requiring a DNA amplification step by PCR. This approach is adapted from Bohlander et al. (16) as described at http://www.microarrays.org/protocols.html. A similar method was recently covered by Horak and Snyder (17). Since DNA sequences in the chromatin immunoprecipitation are heterogeneous, the first step in this approach will consist in the random incorporation of degenerated oligonucleotides (nanomer) attached to a linker sequence. The latter sequence will then be used for the probe amplification by PCR.

1. Use small 0.2 ml PCR tubes and perform all reactions in a PCR thermocycler. Add 2ul 5X sequenase buffer (200 mM Tris-HCl pH 7.5, 100 mM MgCl2 and 250 mM NaCl) and 1 ul (40 picomoles) oligonucleotide (5'-GTTTCCCAGTCACGATCNNNNNNNNN-3') to 7ul immunoprecipitated DNA. Denature DNA 2 min at 94?, cool the tubes down to 8? and incubate at 8? for an additional 2 min. Pause the PCR machine and add 5 ul of reaction mix (1x Sequenase buffer, 0.9 mM dNTPs, 15 mM DTT, 0.75 ug BSA and 4 U Sequenase 2.0). The oligonucleotide is annealed to DNA by slowly raising the temperature from 8? to 37? over a period of 8 min. DNA synthesis is allowed to proceed for an extra 8 min at 37?. Repeat once the whole denaturation-annealing-elongation process. However, since the DNA polymerase is heat sensitive, add 4U Sequenase after the denaturation step at 94?. Finally, stop the reaction by diluting the reaction with 35 ul TE pH 8.0. DNA (step 1 DNA) can be stored at –20? or used immediately for the PCR reaction described below.

2. The following step of the DNA amplification procedure is a regular PCR reaction using the fixed linker oligonucleotide sequence (5'-GTTTCCCAGTCACGATC-3') and DNA from step1 as a template. Transfer an aliquot of 15 ul step 1 DNA into a fresh 0.5 ml PCR tube and carry the reaction in a final volume of 100 ul (20 mM Tris-HCl pH 8.4, 50 mM KCl, 2 mM MgCl2, 1.25 nanomoles oligonucleotide, 0.25 mM dNTPs and 5 U recombinant Taq polymerase (Invitrogen‚Ñ¢)). Denature DNA at 92?C for 30 sec, anneal with two consecutive short 30 sec steps at 40? and 50? respectively and elongate at 72? during 90 sec. Repeat this cycle 24 times and allow the final elongation to proceed for another 10 min at 72?.

3. Purify the PCR reaction through columns (QIAquick PCR purification kit, Qiagen Inc.) as indicated by the manufacturer and elute DNA with 50 ul 10 mM Tris-HCl pH 8.5.

4. Estimate DNA yield under UV-light by spotting serial dilutions of DNA in a Petri dish (0.5 ul DNA mixed with 0.5 ul EtBr (10 ug/ml) along with a standard of known concentration (2 fold serial dilution from 100 ng/ul to 6.25 ng/ul) and test the average size of DNA (5 ul aliquot) after PCR on a 1.5 % (w/v) agarose gel. The probe average size is reduced from 500 bp (after sonication) down to 300 bp. This is due to the random incorporation of the oligonucleotide into DNA during step 1.

Troubleshooting

We routinely obtain between 3 to 5 ug DNA after amplification by PCR. It is not absolutely necessary to have such a DNA yield since the labeling reaction described below requires 500 ng DNA. However, if the amount of DNA is not satisfactory, try to start the amplification procedure with more material or scale-up the chromatin immunoprecipitation. We do not advise increasing the number of PCR cycles since it is kept relatively low in order to ensure amplification linearity.
The final PCR purification through columns (QIAquick PCR purification kit, Qiagen Inc.) is critical since unincorporated oligonucleotides and dNTPs may interfere with the subsequent labeling reaction efficiency.
For some microarray applications, it may be necessary to compare immunoprecipitated DNA with input DNA (i.e. study of a wild-type strain acetylation profile instead of analyzing the changes of acetylation between a wild-type and a histone deacetylase mutant). The Input DNA is prepared from 20 ul cleared lysate as indicated earlier (troubleshooting section for Chromatin immunoprecipitation (ChrIP or ChIP)). Dilute 150 fold an aliquot of the Input DNA prior to start the amplification procedure (step 1).

Klenow labeling of the probe and hybridization.

We favor labeling of the probe by Klenow random priming over a PCR-based method described elsewhere (17). We have noticed that the fluorescent labels are not efficiently incorporated into DNA using a Taq DNA polymerase. Moreover, our first PCR amplification step described above yield enough DNA for direct labeling by Klenow random priming, thus bypassing the need for a second PCR reaction. It is advised to switch the fluorescent dye between the two probes when repeating the experiment to correct for the incorporation efficiency of each dye into DNA.

1. The labeling reaction makes use of reagents (random octamer and concentrated Klenow) from the Bioprime® DNA labeling system (Gibco-BRL) as described at http://www.microarrays.org/protocols.html. Amplified DNA (500 ng) is mixed with 20 ul 2.5 X random primer solution (125 mM Tris-HCl pH 6.8, 12.5 mM MgCl2, 25mM 2-mercaptoethanol and 750 ug/ml random octamers). Complement the solution with water to a final reaction volume of 41 ul. Boil the samples for 5 min and transfer the tubes on ice immediately. Add 5 ul dNTPs mix (1.2 mM dATP, 1.2 mM dTTP, 1.2 mM dGTP and 0.6 mM dCTP in TE pH 8.0), 3 ul of either Cyanine 3-dCTP or Cyanine 5-dCTP (Renaissance®, NEN®) and 1 ul of concentrated Klenow fragment (40 U/ul). Incubate the reaction in the dark for at least 2 hrs at 37?.

2. The probes are purified and concentrated through a microcon-30 filter (Amicon®) as follows. Combine the Cy3- and Cy5-labeled probes and add 450 ul TE pH 8.0 along with tRNA (10 ug) and salmon sperm DNA (10 ug). The addition of nucleic acid competitors (tRNA and salmon sperm DNA) at this stage presents the advantage of increasing the maximum probe volume (up to 4.6 ul) that can be subsequently mixed with the hybridization solution. Place the column on top of a collecting tube and centrifuge at room temperature for 7 min (Eppendorf tabletop centrifuge (model 5415 C) at 12000 rpm). The labeled DNA (violet-purple color) should be clearly visible at the bottom of the column. Discard the flow-through, add another 450 ul TE pH 8.0 in the column and centrifuge again for 7 min. Check the remaining probe volume in the column and centrifuge by 1 min increments until it is concentrated down to 4.6 ul or less. Invert the column and recover the combined probes in a collecting tube by centrifugation (1 min).

3. Prepare the hybridization mix as follow: mix the concentrated probe with 5 ul 20X SSC (3 M NaCl, 0.3 M trisodium citrate, pH 7.0), 10 ul 100% formamide and 0.4 ul 10% (v/v) SDS. Complement the solution with water up to a final volume of 20 ul if necessary (Final concentration: 5X SSC, 50 % formamide, 0.2 % SDS, 0.5 mg/ml tRNA and 0.5 mg/ml salmon sperm DNA). Denature the hybridization mix for 5 min at 95?. The probe can be cooled on ice briefly (few seconds) but should not remain there in order to prevent SDS precipitation. Pipet the hybridization mix on the glass cover slip and cover with the microarray slide. This is favored to the alternative (i.e. pipet the probe on the slide and cover with the cover slip), as it reduces the formation of bubbles. The slide is placed in a hybridization chamber (Corning Inc.) which in turn is arranged in a humid chamber consisting of a closed plastic box (Tupperware®) containing paper soaked in water and a little stand. Hybridization is carried out over-night at 44? in a hybridization oven.

4. The cover slip is removed after a brief immersion in 400 ml 2X SSC at room temperature. The slides are then washed at room temperature for 5 min in 0.1xSSC/0.1% (v/v) SDS and twice in 0.1 x SSC. All washing steps are done in staining jars with 400 ml solution and under mild shaking. Slides are dried by centrifugation in microtiter plate carrier for 5 min at room temperature at 500 rpm (57 g in Beckman J2-HC, rotor JS-4.3). Drainage of liquid from the slides is helped by placing a piece of folded paper towel between the slide holder and the microtiter plate carrier. Microarrays are scanned for both fluorescent dyes using a GMS 418 Array Scanner (Genetic Micro Systems).

Troubleshooting

In our hands, the labeling procedure is reliable and successful most of the time. The color of the purified probe is the best indicator for the procedure success. The combination of the two probes should give a dark purple color after concentration. Do not discard the flow-through if the probe is colorless after the first 7 min centrifugation step through the microcon-30 filter, as it is possible that the filter is defective. Simply reload the flow-through onto another filter and centrifuge again. If the problem persists, this is a clear indication that the labeling was not efficient and the probe should not be used for hybridization. Similarly, a pink (Cy3-labeling) or a blue (Cy5-labeling) color indicates that one of the probes was not labeled properly. Change the vial of dye if you suspect the lack of efficient labeling might stem from a bad batch. The dNTP mix is also an important factor for labeling. We have consistently observed a reduction in its efficiency when using several times the same dNTP mix stock. We strongly advise to prepare it fresh shortly before the labeling reaction.

We have observed that pre-hybridization of the microarray slide is not required when used for acetylation microarrays. Some antibodies used for immunoprecipitation of epitope-tagged protein (i.e. anti-hemagglutinin) appear to create unacceptable hybridization background levels. The following pre-hybridization conditions can be applied in order to alleviate or reduce the background problem. Incubate the slides in 3.5X SSC, 0.1% (v/v) SDS and 10 mg/ml BSA in a staining jar for 20 min at 44?. Rinse the slides briefly first with water then with isopropanol. Dry the slides by centrifugation on a microtiter plate carrier as indicated earlier (step 4).

Data quantitation, normalization and analysis

There are several commercial and freely available software packages that can be used for data quantitation. Fluorescence intensities are quantified in our laboratory using Imagene‚Ñ¢ software (version 4.1) from BioDiscovery, Inc (http://www.biodiscovery.com). The following site contains a non-exhaustive list of various data quantitation and analysis tools (http://genome-www5.stanford.edu/MicroArray/SMD/restech.html). Refer to their respective manual to learn how to use them.
Normalization is probably one of the greatest challenges in data analysis. Indeed, the dye emission intensities are not equal and probes are not labeled similarly due mostly to the different DNA incorporation efficiency of Cy3 and Cy5. A broadly used normalization method makes the assumption that the sums of all intensities of both dyes are equal across the entire array. In other words, any acetylation enrichment should be compensated by a similar decrease in acetylation somewhere else in the genome. If this were true, one could use the ratio of the total intensities between both probes as a normalization factor. While this approach is used for expression arrays, it is not suitable for acetylation arrays since deletion of a histone deacetylase will lead to an increase of general histone acetylation which will not be compensated by a global decrease. Therefore, we strongly suggest the use of internal controls for normalization of acetylation microarrays. Yeast telomeres are hypoacetylated and are not affected by most of the deacetylases (Rpd3, Hda1, Hos1, Hos2 and Hos3) with the notable exception of the telomeric Sir2 which deacetylates histone H4 lysine 16 (10, 18). Telomeric sequences from chromosome 6R (coordinates 269505 to 270095, also labeled Tel6R in the intergenic array) are appropriate for a crude pre-normalization of the data. If possible, this DNA fragment should be spotted many times throughout the glass DNA microarray in order to be reliably used for normalization. Once the data are pre-normalized, identify other regions in the array (between 6 or 10 such regions) whose ratio of intensity between the two probes is very close to Tel6R (ratio of about 1). Confirm that the acetylation state of these newly identified regions is not affected in your experiment using a standard PCR analysis of the original ChrIP (14). Use these regions to finalize normalization of the entire data set.

Different antibodies have inherently different affinities for their own epitope leading to variations in immunoprecipitation efficiency. For instance a highly specific antibody raised against histone H3 lysine 18 may not be as efficient as another antibody recognizing histone H3 lysine 9. This is particularly important when comparing data obtained for different acetylation sites. Data from different experiments can be scaled by calculating the percentile ranking (using Microsoft Excel) of their respective enrichment ratio. This allowed us to compare acetylation with expression data (Fig. 2) (10).
Basic data analysis does not require fancy and expensive piece of software but can readily be performed using Microsoft Access, which is part of the Microsoft office bundle. Access allows the creation and handling of databases. We have for instance created our own database containing gene annotation, published microarray data sets from other laboratories and gene functional categories as defined by MIPS (Munich information center for protein structure at http://mips.gsf.de/proj/yeast/CYGD/db/pathway_index.html). Further data analysis (clustering, sequence analysis, etc…) will need software such as GeneXPress that is freely available for academics (http://genexpress.stanford.edu/).

Most of the data analysis software however were designed for ORF arrays and not for intergenic arrays. Promoter-containing intergenic fragments will have first to be assigned to their respective ORF especially when comparing acetylation status at a promoter with the transcription level of the ORF it regulates. Some intergenic regions are located between two divergent ORFs and are by default assigned to both ORFs. Other intergenic regions are located more than one kb away from any ORF. These are considered orphan and are not assigned to any particular ORF.

DNA intergenic microarray preparation.

Yeast intergenic microarrays includes all sequences located between ORFs including telomeric regions, rDNA, tRNA, centromeres and transposable elements. A full description of the method to prepare microarray slides is not the scope of this review and was thoroughly discussed earlier (19). Rather, we will briefly comment on the method we are currently using in our laboratory. We have amplified by PCR about 6700 intergenic regions from yeast genomic DNA using primer pairs available at ResGen‚Ñ¢ (http://www.resgen.com/). The PCR amplification is performed in 96-well plates and is described at http://www.microarrays.org/protocols.html. The size of every single PCR product was checked by agarose gel electrophoresis. It is possible to coat glass slides in the laboratory with poly-L-lysine (http://cmgm.stanford.edu/pbrown/protocols/1_slides.html). However we strongly recommend the use of amino-silane-coated slides from Corning Inc. (CMT-GAPS‚Ñ¢ coated slides) since they lead to better printing and reduced background during hybridization. PCR fragments (100 ng/ul in 3xSSC) were transferred onto 384-well plates for printing using 16 pins from Array-It (TeleChem International) and an arrayer built according to specifications provided at http://cmgm.stanford.edu/pbrown/mguide/index.html. Printed slides are processed using succinic anhydride blocking according to a method provided by the slide manufacturer. Such a blocking procedure will dramatically reduce hybridization background. The preparation of ORF microarrays is identical. We suggest the printing of ORF and intergenic regions on the same slide if required for your application. Information concerning microarray arrayers and slides manufacturers is hosted at http://www.biologie.ens.fr/en/genetiqu/puces/links.html.

Concluding remarks

Acetylation microarrays have proved to be extremely useful in discovering new functions for the yeast HDACs, functions that would have been much more difficult to identify using classical genetic or molecular biology techniques (10). Although the approach described here focuses on acetylation, it can be extended to other histone modifications as well. The key to a successful analysis depends on the availability of highly specific antibodies and an appropriate crosslinker. We have used a similar approach to study genome-wide binding of Rpd3 (9) using a double-crosslinking approach with a protein-protein crosslinking agent (DMA) and formaldehyde. The latter genome wide binding study clearly showed that Rpd3 binds to many chromatin loci where it has no detectable effect on acetylation when deleted in logarithmically growing cells. These regions would be hidden in acetylation microarrays under the same condition (Fig.3), emphasizing the need for a combination of different type of arrays (acetylation, binding and expression) for full understanding of Rpd3’s function(s).

While acetylation microarrays are extremely powerful, they have some limitations. First, the relative low resolution of the microarrays (up to 1 kb large fragment spotted on the microarray) may not reveal regions where only one nucleosome is modified if nearby nucleosomes are not affected. High resolution is obtained by analyzing a region of interest by standard ChrIP (14) or by printing custom microarrays containing 100 to 200 bp DNA fragments along the regions of interest. Second, histone post-translational modifications create interacting surface for histone binding proteins (e.g. bromodomains of Gcn5 and Snf2 (20)) which may potentially interfere with antibody recognition for the same site. Finally, aneuploidy between two strains (21) has to be corrected in order to ascertain that increase binding, acetylation or expression is not the result of gene duplication in the analyzed strain versus the control. A simple hybridization of INPUT genomic DNA (obtained from the sonicated cell extract but not incubated with the antibody) from the experiment and the control strains can indicate potential aneuploidy problems.

The study of other histone modifications using similar approach will in the future provide important insights not only on their genome-wide respective patterns, but also on how they may relate and influence each others and correlate with transcription activity, unraveling the roles of histone modifications in chromosome functions.

1. Suka N, Suka Y, Carmen AA, Wu J, Grunstein M. New antibodies for sites of acetylation determine novel histone site usage in heterochromatin and euchromatin. Mol Cell 2001;8:473-479.

2. Wu J, Suka N, Carlson M, Grunstein M. TUP1 utilizes histone H3/H2B-specific HDA1 deacetylase to repress gene activity in yeast. Mol Cell 2001;7:117-126.

3. Wang A, Kurdistani SK, Grunstein M. Requirement of Hos2 histone deacetylase for gene activity in yeast. Science 2002;298(5597):1412-1414.

4. Kadosh D, Struhl K. Cell 1997;89(3):365-371.

5. Rundlett SE, Carmen AA, Suka N, Turner BM, Grunstein M. Nature 998;392(6678):831-835.

6. Vogelauer M, Wu J, Suka N, Grunstein M. Global histone acetylation and deacetylation in yeast. Nature 2000;408:495-498.

7. Bernstein BE, Tong JK, Schreiber SL. Genomewide studies of histone deacetylase
function in yeast. Proc Natl Acad Sci USA 2000;97:13708-13713.

8. Hughes TR, Marton MJ, Jones AR, Roberts CJ, Stoughton R, Armour CD, Bennett HA, Coffey E, Dai H, He YD, Kidd MJ, King AM, Meyer MR, Slade D, Lum PY, Stepaniants SB, Shoemaker DD, Gachotte D, Chakraburtty K, Simon J, Bard M, Friend SH. Functional discovery via a compendium of expression profiles. Cell 2000;102:109-126.

9. Kurdistani SK, Robyr D, Tavazoie S, Grunstein M. Genome-wide binding map of the histone deacetylase RPD3 in yeast. Nat Genet 2002;31:248-254.

10. Robyr D, Suka Y, Xenarios I, Kurdistani SK, Wang A, Suka N, Grunstein M. Microarray deacetylation maps determine genome-wide functions for yeast histone deacetylases. Cell 2002;109:437-446.
11. Hecht A, Strahl-Bolsinger S, Grunstein M. Spreading of transcriptional repressor SIR3 from telomeric heterochromatin. Nature 1996;383:92-96.

12. Iyer VR, Horak CE, Scafe CS, Botstein D, Snyder M, Brown PO. Genomic binding sites of
the yeast cell-cycle transcription factors SBF and MBF. Nature 2001;409:533-538.

13. Ren B, Robert F, Wyrick JJ, Aparicio O, Jennings EG, Simon I, Zeitlinger J, Schreiber J, Hannett N, Kanin E, Volkert TL, Wilson C, Bell SP, Young RA. Genome-wide location and function of DNA binding proteins. Science 2000;290:2306-2309.

14. Hecht A, Strahl-Bolsinger S, Grunstein M. Mapping DNA interaction sites of chromosomal proteins. Methods Mol Biol 1999;119:469-479.

15. Kurdistani SK, Grunstein M. In vivo protein-protein and protein-DNA crosslinking for genomewide binding microarray. Methods 2003;In Press.

16. Bohlander SK, Espinosa Rr, Le Beau MM, Rowley JD, Diaz MO. A method for the rapid sequence-independent amplification of microdissected chromosomal material. Genomics 1992;13:1322-1324.

17. Horak CE, Snyder M. ChIP-chip: A genomic approach for identifying transcription factor binding sites. Methods Enzymol 2002;350:469-483.

18. Suka N, Luo K, Grunstein M. Sir2p and Sas2p opposingly regulate acetylation of histone H4 lysine16 and spreading of heterochromatin. Nat Genet 2002;32:378-383.

19. Eisen MB, Brown PO. DNA arrays for analysis of gene expression. Methods Enzymol 1999;303:179-205.

20. Zeng L, Zhou MM. Bromodomain: an acetyl-lysine binding domain. FEBS Lett 2002;513:124-128.

21. Hughes TR, Roberts CJ, Dai H, Jones AR, Meyer MR, Slade D, Buchard J, Dow S, Ward TR, Kidd MJ, Friend SH, Marton MJ. Widespread aneuploidy revealed by DNA microarray expression profiling. Nat Genet 2000;25:333-337.

Figure legends

Figure 1

Acetylation microarrays. Chromatin fragments from crosslinked mutant cells (rpd3?) and their isogenic wild-type counterparts (WT) were immunoprecipitated using highly specific antibodies raised against acetylated histone sites. DNA from enriched chromatin fragments was purified, amplified by PCR and labeled with a fluorophore (Cy3 or Cy5). Probes from both sets of labeled DNA were then combined and hybridized to a DNA microarray containing either intergenic regions, open reading frames or both. For a given region on the microarray, the ratio of the normalized fluorescent intensities between the two probes indicates whether the analyzed lysine residue is hypo- or hyper- acetylated in the experiment strain.

Figure 2

Histone acetylation sites correlate differently with transcription resulting from rpd3?. This figure illustrates the scaling of different acetylation data sets using percentile ranking in order to compare them with a transcription data set. Due to inherent noise in microarray data, absolute correlation analysis between acetylation and transcription is not very informative. A moving average analysis that greatly reduces noise can be used to extract general trends. The moving average (window size, 100 data point; step, 1 data point) percentile rank of acetylation enrichment is plotted as a function of transcription increase resulting from rpd3? (B.E. Bernstein, J.K. Tong and S.L. Schreiber, Proc. Natl. Acad. Sci. USA 97, 13708 (2000)). Acetylation data are plotted for H4 K5 (dark blue), H4 K12 (red), H4 K16 (orange) and H3 K18 (light blue). Control corresponds to a comparison of two sets of probes amplified from the immunoprecipitation of acetylated H4 K12 in the WT strain and labeled separately with Cy-3 and Cy-5 prior to hybridization. Data show that increased acetylation at histone H4 K5 and K12 is associated most directly with increased gene transcription in rpd3?. H4 K16 show the poorest correlation with gene activity (RPD3 disruption has no significant effects on the status of H4 K16 acetylation) (D. Robyr, Y. Suka, I. Xenarios, S.K. Kurdistani, A. Wang, N. Suka and M. Grunstein, Cell 109, 437 (2002)).

Figure 3

Binding arrays are complementary to acetylation arrays. The moving average (window size, 100 data point; step, 1 data point) of Rpd3 enrichment (binding) over intergenic regions is plotted as a function of percentile rank of H3 K18 acetylation in rpd3?. Data sets with (+RP) and without (-RP) ribosomal protein genes are plotted as indicated. Rpd3 binds strongly at the promoter of ribosomal protein genes in logarithmically growing cells where these genes are highly active but under the same conditions, has little or no effect on acetylation of these promoters (S.K. Kurdistani, D. Robyr, S. Tavazoie and M. Grunstein, Nat. Genet. 31, 248 (2002); D. Robyr, Y. Suka, I. Xenarios, S.K. Kurdistani, A. Wang, N. Suka and M. Grunstein, Cell 109, 437 (2002)).or the expression of the RP genes (B.E. Bernstein, J.K. Tong and S.L. Schreiber, Proc. Natl. Acad. Sci. USA 97, 13708 (2000)). Thus, the ribosomal protein gene promoters as targets of Rpd3 are only detectable by binding arrays. This clearly illustrates the importance of combining different types of arrays (binding, acetylation and expression) to fully comprehend HDAC function.