Compositional and structural analysis of selected chromosomal domains from Saccharomyces cerevisiae

Chromatin is the template for replication and transcription in the eukaryotic nucleus, which needs to be defined in composition and structure before these processes can be fully understood. We report an isolation protocol for the targeted purification of specific genomic regions in their native chromatin context from Saccharomyces cerevisiae. Subdomains of the multicopy ribosomal DNA locus containing transcription units of RNA polymerases I, II or III or an autonomous replication sequence were independently purified in sufficient amounts and purity to analyze protein composition and histone modifications by mass spectrometry. We present and discuss the proteomic data sets obtained for chromatin in different functional states. The native chromatin was further amenable to electron microscopy analysis yielding information about nucleosome occupancy and positioning at the single-molecule level. We also provide evidence that chromatin from virtually every single copy genomic locus of interest can be purified and analyzed by this technique.


INTRODUCTION
Chromatin is dynamic and changes in its composition and structure depending on the functional state of the respective genomic region. The basic repeating unit of chromatin, the nucleosome, is composed of 147 basepairs (bp) of DNA wrapped around a core of eight histone proteins [reviewed in (1)]. Nucleosomes are not uniform because histone variants and different posttranslational covalent modifications establish distinct signatures in the chromatin landscape [reviewed in (2)], which is further characterized by the specific association of other protein factors with genomic regions. Most of our current knowledge regarding chromatin composition at defined genomic loci is based on chromatin immunoprecipitation (ChIP) [reviewed in (3)]. ChIP is helpful to map DNA interactions of known chromatin components but requires a priori knowledge for the selection of candidate proteins.
Various strategies have been used to isolate and analyze the composition of defined chromosomal domains to derive a description of chromatin at selected genomic loci (4)(5)(6)(7)(8)(9)(10)(11). Most of these attempts suffered from low recovery or insufficient purity of the isolated material, making downstream analyses difficult. Recently, detailed information about protein composition at telomeres in human cell lines and in Drosophila was obtained by a method called PICh for Proteomics of Isolated Chromatin segments (12,13). Another report introduced chromatin affinity purification to investigate the proteome and posttranslational histone modifications at a single-copy locus in S. cerevisiae (hereafter called yeast) (14). Like PICh, chromatin affinity purification is carried out with formaldehyde crosslinked material under denaturing conditions and does not provide a source to native chromatin for functional and structural analysis. Two other studies in yeast analyzed the protein composition of centromere bound kinetochores or replication-dependent changes in post-translational histone modifications on multicopy plasmids (15,16). It remains unclear, however, whether chromatin on multi-copy plasmids fully reflects chromatin at chromosomal loci.
Here, we describe a protocol for the isolation of native chromosomal domains from yeast. We built on a previously established technique based on site specific recombination at chromosomal loci, which had been tagged with a cluster of LexA DNA binding sites ( Figure 1A) (7). After recombination, chromosomal domains were released in the form of a chromatin ring and could be isolated via a co-expressed recombinant LexA protein fused to a tandem affinity purification (TAP) tag (17). While this previously established technique allowed isolation of chromosomal domains (7), the chromatin preparations were of insufficient purity to perform some of the intended analyses. We therefore developed a new purification strategy now enabling compositional and structural analyses. First, the approach was applied to the multicopy ribosomal DNA (rDNA) gene cluster, which consists of 150-200 tandem repeats on chromosome XII ( Figure 1B) [reviewed in (18)]. The rDNA locus contains the 35S ribosomal RNA (rRNA) gene, a bi-directional promoter (Epro) and the 5S rRNA gene transcribed by RNA polymerases (Pols) I, II and III, respectively, as well as an autonomous replication sequence (ARS). These functional genomic elements could be individually purified and analyzed. Finally, isolation of a chromatin domain containing the single copy PHO5 gene and identification of co-purifying proteins by mass spectrometry (MS) suggested that this purification strategy will be applicable to any genomic locus of interest.

Plasmids and yeast strains
Unless noted otherwise, standard techniques were used for cloning of plasmids and transformation of yeast cells (19,20). Complete lists of oligonucleotides, plasmids and yeast strains can be found in Supplementary Tables S1-S3.
For construction of plasmids containing a modified intergenic spacer (IGS) region of the rDNA locus, a BamHI/NotI fragment of pNOY373 (23) was cloned into BamHI and NotI digested pBluescript KS (Stratagene) resulting in plasmid K1560 (pBluescript BamHI NotI 5S). A NotI/BsrGI fragment containing the E-pro region, a BseRI/SphI fragment containing the 5S rDNA gene and a SphI/BamHI fragment containing the ARS element from K1560 were blunted and cloned into HpaI and XhoI digested, blunted plasmid pM49.2 (24) yielding K1578 (pUS6), pMW5a and pUS1, respectively. The resulting plasmids were digested with BamHI and PstI for pUS6 and pMW5a or with BamHI and SbfI for pUS1, blunted and reinserted into NotI and BsrGI digested, blunted, BsrGI and SphI digested, blunted or SphI and BamHI digested, blunted K1560 yielding pUS9b, pBluescript_5S-RS and K1577 (pUS3), respectively. The self-complementary oligonucleotide 2629 was annealed and ligated with SacII restricted pUS9b to introduce a NotI restriction site yielding K2024 (pUS9b_NotI). pBluescript_5S-RS was digested with NotI and SmaI, and the resulting 2.158 bp fragment was cloned into NotI/ SmaI digested pNOY373 yielding pNOY373_5S-RS. pUS9b_NotI was digested with NotI and BamHI and cloned into NotI and BamHI digested pNOY373 yielding pUS11. pUS3 was digested with NotI and EcoRI, and the blunted 2.197 bp fragment was ligated with NotI and BamHI digested blunted pNOY373 yielding pUS5. pNOY373_5S-RS was digested with SmaI and PstI, and the 9.457 bp fragment was cloned into PstI and BamHI-digested pT4 (21) yielding K1185 (pAG43). pUS5 was linearized with PstI and ligated with PstI linearized pT4 resulting in K2026 (pUS7). pUS11 was digested with BamHI and PstI and cloned into BamHI and PstI digested pT4 resulting in K2027 (pUS12).
A previously described method was used to genetically modify the endogenous rDNA locus (23) such that each rDNA repeat contained the specific insertion of RS elements and LexA-binding sites. To this end, the plasmids K375 (pT11), K389 (pT25), K674 (pT36), K1185 (pAG43), K1785 (pT11-WT), K2026 (pUS7), K2027 (pUS12) were digested with SpeI. The resulting DNA fragments contained a wild-type rDNA repeat (K1785) or a modified rDNA repeat, in which either the ARS (K2026), the 5S rRNA gene (K1185), the E-pro region (K2027), the 18S rRNA coding sequence (K674), the 35S rRNA gene (K389) or a complete rDNA repeat (K375) are flanked by RS elements and include LexA-binding sites. These fragments also contained sequences for integration and expansion of the modified rDNA repeats as described earlier (23). After transformation into yeast strain NOY989, mutant clones containing a re-expanded modified rDNA locus were selected as described (23).
None of these yeast strains had detectable growth defects suggesting that the RS elements and LexAbinding sites integrated in the rDNA locus did not significantly affect rDNA transcription and ribosome production in the cell. We note that some of the strains containing insertions at or in the proximity of the NotI restriction site within the region 3 0 of the 35S rRNA gene [present in pNOY373 (23)] had a lower rDNA copy number. A reduced copy number had also been reported in a strain where the NotI restriction site had been used to delete part of the rDNA enhancer region, before re-integration and re-expansion of the modified rDNA repeat (25).
Generation of plasmids for expression of R recombinase and LexA-TAP A set of plasmids for inducible expression of R recombinase and constitutive expression of LexA-TAP was constructed. The CYC1 or TEF2 promoter regions, Figure 1. Distinct domains of the rDNA locus can be purified from yeast. (A) Purification of chromosomal domains. LEXA, cluster of LexA DNA binding sites; RS, sequences for site specific recombination; LexA-TAP, recombinant LexA fusion protein; filled ovals, chromatin components; filled circle, IgG coated magnetic beads. (B) Genetic manipulation of the rDNA locus. The rDNA locus on chromosome XII is depicted on the top. CEN, centromere; TEL, telomere; 35S, 25S, 18S, 5S rRNA coding regions; IGS, intergenic spacer region; ARS, autonomous replication sequence (gray circles); E-pro, expansion promoter; black arrows, transcription start sites used by Pol I, II, III; gray arrows; insertion sites of RS elements and LexA DNA binding sites; numbers in parentheses represent sizes of the different domains. (C) DNA analysis of samples of a chromatin domain purification and of a control purification. Yeast strains y2381 (35S) and y2378 (control), carrying the 35S rRNA gene domain flanked by RS sites, or lacking recombination sites, respectively, were subjected to the purification procedure. DNA was isolated from samples CE, SUP, P, FT, E and B ( Figure 1A), SacII digested and analyzed in a 1% agarose gel stained with SYBR Õ Safe (Life technologies). The LexA bait protein was purified from 10 11 cells and 0.7% (CE, P), 2.5% (SUP, FT) and 8.3% (E, B) from the respective samples were analyzed. Positions of DNA size markers, and of the linearized 35S rRNA gene ring DNA (35S) are indicated. (D) DNA analysis of purified rDNA chromatin domains. Chromatin domains were purified from yeast strains y2384 (E-Pro), y2379 (5S), y2383 (ARS), y2380 (18S), y2381 (35S) and y2382 (rDNA). DNA was isolated from TEV eluates, digested with NcoI (E-pro, 5S, ARS) or SacII (18S, 35S, rDNA) and analyzed as described in the legend to Figure 1A. The LexA bait protein was purified from 10 11 cells, and 5% of the respective samples were analyzed. Positions of DNA size markers and of restriction fragments of the individual rDNA segments are indicated. Asterisks mark restriction fragments from higher-order recombination products of the respective domain.
supporting either minimal or robust expression of the fusion protein, were amplified with primer pairs 2507, 2511 or 2508, 2512 from yeast genomic DNA. The PCR products were digested with XbaI and XhoI and cloned into XbaI and XhoI digested pBluescript KS yielding plasmids pBluescript_CYC1 and pBluescript_TEF2. The KpnI/XbaI fragment of plasmid pJSS3 (24) containing the coding sequence of LexA-TAP was cloned into KpnI and XbaI digested pBluescript_CYC1 and pBluescript_TEF2 resulting in pBluescript_CYC1 LexA-TAP and pBluescript_TEF2 LexA-TAP. The plasmids were digested with BamHI and KpnI and the resulting 2.734 or 2.848 bp fragments were blunted and cloned into SmaI linearized pB3 (24) yielding K2048 (pSH15) and K2049 (pSH17).
Another plasmid K929 (pKG7) allows for inducible expression of R recombinase and LexA-TAP both under control of the bidirectional GAL1-10 promoter. To this end, plasmid K356 (pJSS3) (24) was linearized with BssHII, and the resulting 7.891 bp fragment was blunted and ligated with SmaI linearized plasmid K355 (pB3) yielding K363 (pR2). K363 was digested with SpeI, and the resulting 11.239 bp fragment was ligated yielding K929 (pKG7). The plasmids K2048 (pSH15), K2049 (pSH17) and K929 (pKG7) allow ectopic expression of R recombinase under control of the inducible GAL1-10 promoter and expression of the LexA-TAP protein under control of the constitutive CYC1, TEF2 or inducible GAL1-10 promoter, respectively. The transformed plasmids have to be maintained in the yeast cell, which is accomplished by a selectable LEU2 marker gene present on all the expression plasmids. However, cells must be grown in minimal medium lacking leucine to obtain uniform cell populations including the plasmid.
To conduct the chromatin preparations from yeast cells grown in full medium, the expression cassettes for R recombinase and LexA-TAP were flanked with homologous sequences of the yeast URA3 gene. To this end, the URA3coding sequence was amplified with primer pair 2686, 2687 from yeast genomic DNA. The PCR product was digested with KpnI and SacII and cloned into KpnI and SacII digested pBluescript KS yielding K2051 (pBluescript_ URA3). Plasmids pSH15 and pSH17 were digested with BssSI, and the resulting 9.314 or 9428 bp fragments were blunted and cloned into StuI linearized plasmid K2051 yielding K2054 (pSH23) and K2052 (pSH21). K2052 (pSH21) was digested with HindIII and PflFI, blunted and religated yielding K2053 (pSH22). The plasmids K2052 (pSH21) and K2054 (pSH23) allow genomic integration of the expression cassettes for R recombinase under control of the inducible GAL1-10 promoter and the LexA-TAP protein under control of the constitutive TEF2 or CYC1 promoters, respectively. Plasmid K2053 (pSH22) allows genomic integration of the expression cassette for R recombinase under control of the GAL1-10 promoter without LexA-TAP expression. The plasmids were digested with SbfI, and the resulting DNA fragments were transformed in the yeast cells allowing the stable chromosomal integration of the expression cassette at the endogenous URA3 locus by homologous recombination. Mutant clones for positive integration were selected on SCD-LEU plates for the LEU2 marker present on the expression cassettes.
Generation of yeast strains expressing MNase fusion proteins Strains expressing yeast proteins with a C-terminal MNase carrying a triple hemagglutinin (3 Â HA) epitope from their chromosomal location were generated as described previously (26). Expression of the MNase fusion protein was verified by western blot analysis with antibody 3F10 (Roche) recognizing the C-terminal 3 Â HA-tag of the fusion proteins [data not shown, see also (26)]. None of the strains expressing MNase fusion proteins showed an obvious growth phenotype (data not shown).

Affinity purification of chromatin domains
Yeast cells competent for excision of chromatin domains by R recombinase were cultivated overnight at 30 C in Yeast extract-peptone-raffinose or selective complete raffinose medium lacking leucine (SCR-LEU). For rDNA ring purifications, we used yeast strains in which the LexA-TAP protein was constitutively expressed at moderate levels controlled by the TEF2 promoter as opposed to the strong GPD promoter used in earlier studies (7,24,27). For purification of PHO5 gene rings LexA-TAP levels were further reduced by using the weak CYC1 promoter for expression of the protein (27). Reducing the cellular levels of the fusion protein led to a lower background with chromatin from sheared bulk genomic DNA in the preparations. Recombination was induced at a cell density of 5-7 Â 10 7 cells/ml (OD 600 of 0.8-1.0) by adding galactose to a final concentration of 2% (w/v). Cells were grown for an additional 1.5 h at 30 C before harvesting to allow proper expression of R recombinase and formation of chromatin rings. After induction, cells were harvested by centrifugation (10 min, 9000 Â g at 4 C), yielding a wet weight between 3 and 4 g per liter of cell culture. After washing twice with water, cells were pelleted in sealed 20 ml syringes by centrifugation (5 min, 3000 Â g at 4 C). Supernatants were decanted, syringes were unsealed and cells were extruded into liquid nitrogen. The resulting 'cell spaghetti' were stored at À80 C until use. A commercial coffee grinder (TEFAL, Prep'line) was pre-cooled by grinding 30-50 g of dry ice twice. The resulting powder of dry ice was discarded. Appropriate amount of frozen cells (3-4 g for rDNA ring purification or 18-40 g for PHO5 gene ring purification) were mixed with $60 g of dry ice in the coffee mill. Grinding was repeated three times for 1 min with short intervals in between grinding to prevent heating of the coffee mill motor. Shaking of the coffee mill while grinding prevented the dry ice-cell powder from sticking to the inside wall of the grinding chamber. The fine powder of ground yeast can be stored at À80 C.
Protein A-immunoglobulin G (IgG) affinity purification was performed as described previously (28,29) with minor modifications. After evaporation of dry ice, the powder was dissolved in 0.75 ml of cold buffer MB [20 mM Tris-HCl (pH 8), 200 mM KCl, 5 mM MgAc, 0.5% Triton X-100, 0.1% Tween-20, 1 mM DTT] with 1Â Protease inhibitors (0.17 mg/ml PMSF and 0.33 mg/ml benzamidine) per 1 g of ground yeast cells. The cell lysate was cleared from cell debris by centrifugation for 30 min at 16.000 g and at 4 C. From 1 g of yeast cells, $1 ml of cell lysate with a protein concentration of 15-20 mg/ml was obtained. To generate the affinity resin, rabbit IgGs (Sigma) were added to epoxy-activated magnetic beads (BcMag TM , Bioclone Inc.) in a ratio of 0.17 mg IgGs/mg of beads according to the manufacturers instruction. The IgGs coupled to magnetic beads were equilibrated with buffer MB before use. For the purification of rDNA rings, 7.5 mg of magnetic beads with coupled IgGs were added to a total amount of 3 ml of cell lysate. For the purification of PHO5 gene rings 3.75 mg of magnetic beads with coupled IgGs were added to a total amount of 15 ml of cell lysate. The cell lysate-bead suspension was incubated on a rotating wheel for 1 h at 4 C. Beads were washed five times with 750 ml of cold buffer MB with 1Â Protease inhibitors per 3.75 mg of magnetic beads with coupled IgGs. Between each washing step, the beads were gently rotated for 10 min. Finally, the beads were washed with 750 ml of cold buffer MB without Protease inhibitors per 3.75 mg of magnetic beads with coupled IgGs. Chromatin rings were eluted by proteolytic cleavage for 2 h (or overnight) at 4 C with 10 mg 6ÂHistagged recombinant Tobacco Etch Virus (TEV) protease in a total volume of 100 ml per 3.75 mg of magnetic beads with coupled IgGs. The supernatant was transferred to a new microtube, and residual chromatin circles were washed from the beads with another 100 ml of buffer MB and combined with the eluate.
To deplete the 6xHis tagged TEV protease from the purified rDNA chromatin rings, 20 ml of Ni-NTA beads (Qiagen) equilibrated with buffer MB were incubated with the eluate in a sealed 1.5 ml column (Bio-Spin, BioRad) for 30 min at 4 C on a rotating wheel. The column was unsealed, and the sample was transferred to a new microtube. This step removed virtually all of the recombinant TEV protease, which was important for subsequent MS analysis.
For PHO5 chromatin rings eluting from the IgG beads after TEV protease cleavage, the volume was adjusted to a total volume of 400 ml with buffer MB and supplemented to a final concentration of 2 mM CaCl 2 . The elution sample was applied to 100 ml of calmodulin sepharose beads (Stratagene) equilibrated with buffer MB containing 2 mM CaCl 2 for at least 2 h. Beads were incubated for 1 h at 4 C and washed 4Â with 1 ml of buffer CWB [20 mM Tris-HCl (pH 8), 300 mM KCl, 5 mM MgAc, 0.5% Triton X-100, 0.1% Tween-20, 2 mM CaCl2, 1 mM DTT]. Chromatin rings were eluted from calmodulin beads by gravity flow with 100 ml of buffer CEB [20 mM Tris-HCl (pH 8), 200 mM KCl, 5 mM MgAc, 1 mM EDTA, 10 mM EGTA, 0.5% Triton X-100, 0.1% Tween-20, 1 mM DTT]. After sealing the column, 200 ml of buffer CEB were added to the beads, and the slurry was incubated for 20 min at 4 C under shaking at 1000 rpm in a IKA-Vibrax shaker. Buffer CEB was collected by gravity flow before another 100 ml of buffer CEB were applied to wash residual chromatin rings from the beads. All elution steps were combined yielding a total volume of 400 ml.

Blot analysis
Western blot analysis was performed according to standard procedures (30). Supplementary Table S4 contains a complete list of antibodies used for detection.
Southern blot analysis was performed as described (26). Supplementary Table S5 contains a complete list of hybridization probes.

Analysis of histone modifications by MALDI TOF/TOF MS
Histones H3 and H4 were extracted from an SDSpolyacrylamide gel and analyzed as previously described (31) with minor modifications. After destaining the gel pieces with 0.1 M ammonium bicarbonate in 30% high performance liquid chromatography-grade acetonitrile, samples were washed twice with 70% high performance liquid chromatography-grade methanol and acylated with 0.71 M propanoic anhydride in 70% methanol. Proteins were then digested with trypsin (Roche Applied Science) according to manufacturer's instructions, and 0.5-1 ml of aliquots were spotted in triplicate onto a stainless steel MALDI target plate, allowed to dry, overlaid with 0.5 ml of freshly prepared, ice-cold 5 mg/ml alpha-cyano-4hydroxy cinnamic acid (for mass spectrometry, Fluka) in 50% acetonitrile/0.3% TFA, dried again and analyzed by MALDI-TOF/TOF in a 4800 instrument from Applied Biosystems, operated according to manufacturer's instructions. Full MS spectra were acquired in the reflector modus, with m/z from 700 to 2000 (H4) or 700 to 2500 (H3); focus on 1500 (H4) or 1600 (H3), and 50 Â 50 spectra were acquired per spot. For the tandem MS/MS spectra, the peptide 4-17 of H4 was fragmented in PSD modus, whereas the peptides of H3 were fragmented with collision-induced dissociation; in all the cases, the isolation window width was 1.0 Da; an inclusion list was used for parent ion selection (tolerance 0.3 Da), and 85 Â 45 shots per spectra were acquired. For all MS and MS/MS spectra, the laser intensity was manually adjusted for optimized S/N. Spectra were processed with mMass 5.2.0 (32,33). After recalibration by using signals of the histones (peptides 46-55 and 79-92 for H4 and peptides 64-69 and 27-40 unmodified for H3), the relative proportion of a given modification for the peptides of interest was calculated by dividing the absolute intensity of the signal corresponding to that modification through the sum of the intensities for all the signals corresponding to any modified species from the same peptide. The results for all the MALDI replicas of each measurement were averaged.

Comparative iTRAQ MALDI TOF/TOF MS analysis
Proteins present in the elution samples of chromatin ring preparations (see earlier in the text) were precipitated by methanol/chloroform (34). Digestion of the proteins with trypsin, iTRAQ labeling and quantitative MALDI MS analysis were essentially performed as described elsewhere (29). The processed data set is presented in Supplementary Data Sets 1 and 2.

Chromatin endogenous cleavage and ChIP
For chromatin endogenous cleavage (ChEC) experiments, yeast strains expressing the MNase fusion proteins from their endogenous genomic location were grown in yeast peptone dextrose at 30 C to a final OD 600 of 0.5. ChEC analyses were performed as previously described (26). ChIP was performed as described elsewhere (35). Primer pairs used for amplification are listed in Supplementary  Table S1. Data were collected with a Rotor-Gene 3000 system (Corbett Research) and analyzed using the comparative quantitation module of the system software. Retention of specific DNA fragments was calculated as a percentage of total input DNA. The mean values and error bars are derived from at least three independent ChIP experiments analyzed in triplicate quantitative PCR reactions.

Restriction endonuclease digestion analysis of chromosomal and purified chromatin domains
For restriction endonuclease digestion of chromosomal and affinity purified chromatin domains, crude nuclei were prepared as described in (21). The 5S rRNA gene chromatin rings were purified as detailed earlier in the text. An aliquot of the crude nuclei corresponding to 2 Â 10 8 cells or 1.25% of the TEV elution were adjusted to a final volume of 100 ml with the buffer recommended for the respective restriction enzyme (NEB). Digestion was performed for 60 min at the optimal reaction temperature using different amounts of the respective restriction endonuclease (10 or 50 U for nuclei and 2 or 20 U for the purified chromatin ring). The reaction was terminated by adding one volume of IRN buffer [50 mM Tris-HCl (pH 8), 20 mM EDTA, 500 mM NaCl]. Samples were treated with RNase A at a final concentration of 0.33 mg/ml for 1 h at 37 C. Proteinase K and SDS were added to a final concentration of 0.33 mg/ml and 0.5%, and incubation was continued for 1 h at 56 C. After phenol/chloroform extraction, DNA was precipitated with ethanol in the presence of 40 mg of glycogen as a carrier. The DNA was suspended in 20 ml of H 2 O, and DNA molecules were linearized by restriction enzyme digestion overnight at 37 C in a final volume of 30 ml with 20 U of PvuI and SphI (nuclei) or 20 U NcoI (chromatin rings), respectively. The DNA samples were subjected to indirect endlabeling Southern blot analysis.
As a control, each restriction enzymatic digestion was also performed with samples deproteinized by phenol/ chloroform extraction. The DNA was almost completely digested in all cases (not shown).
Trimethylpsoralen crosslinking of rDNA chromatin rings and preparation of DNA for electron microscopy analysis Crosslinking was performed as described (36) with some modifications. Purified chromatin rings in 5 ml of buffer CEB were placed onto a 10-cm petri dish floating on an ice water slurry; after addition of 0.05 volumes of 400 mg/ml trimethylpsoralen in ethanol the sample was incubated in the dark on ice for 5 min. Samples were positioned 5-cm away from five 366-nm ultraviolet bulbs in a Stratalinker 2400 (Stratagene) and then irradiated for 5 min. Addition of psoralen, incubation in the dark, and irradiation were performed a total of seven times for each sample. After treatment with RNase A (at a final concentration of 0.33 mg/ml for 2 h at 37 C), Proteinase K and SDS were added to a final concentration of 0.33 mg/ml and 0.5% and incubation was continued for 4 h at 55 C. DNA was extracted with phenol/chloroform and precipitated. DNA was suspended, digested with NcoI, purified using a DNA Clean and Concentrator Kit (ZymoResearch) and eluted from the column with 8 ml of TEN (30 mM Tetraethylammoniumchloride, 20 mM EDTA, 10 mM NaCl). Denaturing, spreading, staining with uranyl acetate and rotary metal shadowing were performed as previously described (36).

Electron microscopy
Images were taken on a JEOL 1230 electron microscope at 120 keV at 20 000Â magnification. Images were processed and analyzed in ImageJ (37).

Distinct chromatin domains of the rDNA can be purified from yeast
Yeast strains were constructed in which either the ARS, the 5S rRNA gene, the E-pro region, the 35S rRNA gene, the 18S rRNA coding sequence or a complete rDNA repeat are flanked by RS elements and include LexAbinding sites ( Figure 1B). The cells also contained a chromosomally integrated cassette for constitutive expression of a LexA-TAP fusion protein. Chromosomal integration of the expression cassette as well as reducing the LexA-TAP expression levels by the choice of an appropriate yeast promoter were major differences to the previous purification protocol (7) (see 'Materials and Methods' section for details). After recombination, cells were lysed, and the whole-cell extract was subjected to a single-step affinity purification protocol using IgG coupled to magnetic beads (28). This change in the purification strategy proved to be important to obtain chromatin preparations, that are suitable for MS and electron microscopy (EM) analyses. Figure 1C shows a representative nucleic acid analysis of different samples taken during the purification of a chromosomal 35S rDNA domain ( Figure 1C, lanes 1-6). The 35S rDNA domain was efficiently eluted under native conditions following TEV protease cleavage between the C-terminal protein A moiety and the calmodulin binding peptide (CBP) of LexA-TAP ( Figure 1C, lane 5). Similarly, the E-pro-, 5S-, ARS-, 18S-regions as well as the entire rDNA repeat could be purified ( Figure 1D; Supplementary Figure S1). No nucleic acids were detected in the control purification from an isogenic yeast strain lacking sites for recombination and LexA binding ( Figure 1C, lane 11). In some preparations, larger DNA fragments were detected in addition to the fragment expected for the specific domain ( Figure 1D, labeled by asterisks). These fragments originated from higher-order recombination products as confirmed in Southern blot analysis (Supplementary Figure S1). Around 50-500 fmoles of purified domains were obtained from one liter of exponentially growing yeast culture ($10 11 cells) corresponding to 2-13% of total cellular rDNA domains (Table 1). In the final elution fraction the specific chromosomal domains were present in a 4000-fold to 30 000-fold excess over another unrelated single copy gene locus as determined by quantitative PCR (Table 1).

rDNA subdomains exhibit distinct histone modification patterns
Proteins co-purifying with individual domains or in the control purification were analyzed by SDS-PAGE and Coomassie staining. The LexA fusion protein (LexA-CBP) was similarly enriched in all preparations ( Figure 2A, lanes 1-6). Small proteins specifically co-purifying with the rDNA domains were identified as the canonical histones after excision from the gel and MS analysis ( Figure 2A, lanes 2-5, Figure 2C). The post-translational modification state of histones was investigated by western blot analyses ( Figure 2B). Similar amounts of unmodified histone H3 were detected in the chromatin preparations, but not in the control purification. H3 lysine 4 tri-methylation (H3K4me3) was enriched in the ARS domain purification, corresponding well with ChIP data in a previous report (38). Histone H3 lysine 36 tri-methylation (H3K36me3) was instead preferentially found in preparations of the E-pro and 35S rDNA domains. The level of acetylated lysine residues 5, 8, 12 and 16 within histone H4 (H4ac) was increased in purified 5S rDNA domains again in good correlation with ChIP experiments (39).
Post-translational covalent modifications of gel purified histone proteins in preparations of ARS and 5S rDNA subdomains were additionally assessed by MS confirming the increased acetylation within the histone H4 N-terminal residues 4-17 in the preparations containing purified 5S rDNA subdomain ( Figure 2C). Monoacetylated H3 peptides including K9/K14 and K18/K23 were slightly enriched in the 5S purification compared with the ARS sample (K9/K14 Ac, K18/K23 Ac) correlating with previous ChIP results (39). H3K56 acetylation, playing a role in the regulation of replication-coupled nucleosome assembly (40,41), was enriched within the ARS preparation ( Figure 2C, K56 Ac). We also quantified the relative amounts of unmodified, mono-, di-and trimethylated peptides including K79 of histone H3 ( Figure 2C, K79 Me). The percentage of unmodified K79 was significantly enriched in the 5S sample compared with the ARS chromatin purification. Histone H3 molecules associated with the ARS and 5S rRNA gene rings appeared to be hypomethylated at K79 when compared with the 18S rRNA gene domain or a preparation of bulk yeast chromatin (data not shown). This observation is also supported by previous ChIP analyses (42). We conclude that distinct histone modification patterns of purified rDNA domains are preserved after native purification and can be determined by MS. The absolute molar amounts of the DNA of the respective chromosomal domain and the single copy were determined by quantitation of the Southern blots shown in Supplementary Figure   Coomassie-stained histone bands were excised, and trypsin-digested peptides were subjected to MALDI TOF/TOF analysis. Spectra were quantified, and the results for a respective peptide are depicted in bar diagrams, as a percentage of the total amount of this peptide in the analysis. Average values and standard deviation are from two biological replicates. Cartoons below the graph show the relevant amino acid sequence of H4 and H3 with the lysine residues carrying the respective modification.

Specific chromatin components are enriched in purifications of rDNA subdomains
The silver stained gel in Figure 2A revealed specific protein bands present in individual chromatin preparations (Figure 2A, bands labeled with asterisks). DNA migrating in the upper part of the gel could be detected by this staining procedure (Figure 2A, lanes 8-12). The copurification of candidate factors with the rDNA domains was investigated by western blot analyses ( Figure 2B). While the LexA-CBP bait protein was present in similar amounts in all samples analyzed, the Pol I specific subunit Rpa135 was specifically detected in the 35S rDNA purification. Enrichment of the RNA polymerase I enhancer binding protein 1 (Reb1) was observed in the purification of the ARS domain containing a Pol I promoter-proximal Reb1-binding site supporting robust Reb1 binding in vivo (35). The TATA binding protein (Spt15/TBP) was present in all purifications containing rDNA consistent with its role in transcription initiation by all three RNA polymerases [reviewed in (43)]. TBP association with the ARS domain could correlate with a role for yeast TBP in DNA replication (44,45). With the exception of LexA-CBP, none of the above proteins was detected in the control purification. Taken together, interactions of nonhistone proteins with the isolated rDNA domains are at least partially preserved during the purification procedure.
Compositional analysis by comparative MS reveals distinct proteomes co-purifying with rDNA subdomains MS and the isobaric tag for relative and absolute quantitation (iTRAQ) technology (46) were used to analyze proteins co-purifying with rDNA domains. Tryptic peptides derived from protein samples from rDNA domain or control purifications were separately labeled using two different iTRAQ reagents ( Figure 3A). The relative amounts of the differentially labeled peptides in the rDNA domain purification and in the control purification were determined by LC-MALDI-TOF/TOF MS and are depicted as iTRAQ ratio in Figure 3C; and Supplementary Data Set 1 (full MS summary). Between 250 and 350 proteins were identified in each of the purifications of the four rDNA subdomains. The proteins were classified according to their biological function ( Figure 3B, pie charts). Very similar amounts of peptides from abundant housekeeping proteins and ribosomal proteins were identified in all proteomes co-purifying with rDNA subdomains and in the respective control purification ( Figure 3B) likely representing background contaminants. Therefore, the iTRAQ ratio of every protein identified in the individual analyses was divided by the average iTRAQ ratio of all housekeeping proteins and ribosomal proteins. The average iTRAQ ratio for each protein was then calculated from three biological replicate experiments and plotted against the sum of peptides of the protein identified in these replicates ( Figure 3C). Owing to the background correction, housekeeping proteins and ribosomal proteins form a cluster in the center of each diagram ( Figure 3C, light gray squares and gray diamonds in all graphs). Only proteins with an average iTRAQ ratio of at least 1.5, identified with at least 2 peptides were considered as enriched in individual rDNA domain purifications and are discussed later in the text (Table 2 and Figure 3C, area framed by dashed lines in the diagrams). If at least 50% of multi-protein complex components fulfilled part of these criteria, the average peptide count and iTRAQ ratio for the respective complex is depicted.
Distinct proteomes co-purified with the individual rDNA domains. Histones, including Htz1, the yeast homolog of H2A.Z, were equally enriched in all preparations correlating with the SDS-PAGE analysis (Figure 2A). The yeast homolog of the linker histone H1, Hho1, instead was preferentially found in E-pro and 35S rDNA domain preparations. This result agrees with a previously reported role for Hho1 in 35S rDNA chromatin structure (47,48). The MCM2-7 complex, the predominant replicative helicase [reviewed in (49)], was selectively identified in the ARS domain preparation, as was Reb1 in good agreement with the western blot analysis ( Figure 2B). The Spt15/ TBP-containing Pol III initiation complex TFIIIB copurified with the 5S rDNA domain in good agreement with the results in Figure 2B (panel Spt15/TBP). In the 35S rDNA chromatin preparation, 11 of 14 subunits of Pol I were identified. Unexpectedly, 8 of 12 subunits of Pol II were also detected in this sample. Pol II has been shown to transcribe rDNA in some conditions (50-54) but might artificially associate with this domain during purification (see Discussion). In the 35S rDNA preparation, a specific group of ribosome biogenesis factors including constituents of the H/ACA box and C/D box snoRNPs were identified. These proteins are thought to bind co-transcriptionally to nascent rRNAs extending from elongating Pol I (55-58). Although C/D box snoRNPs were strongly enriched in the 35S rDNA preparation, they were also detected in all other rDNA purifications, which could point to a more general function of these complexes or be explained by the presence of higher-order recombination products including 35S rDNA (see Figure 1D and Supplementary Figure S1) in these preparations. Different chromatin remodeling complexes co-purified with individual rDNA domains. Whereas 10 of 17 subunits of the RSC complex were detected in the ARS domain purification, many subunits of the INO80 chromatin remodeling complex co-purified with the 35S rDNA. ISW1 complex components were enriched in ARS and 35S rDNA preparations, corroborating the reported interaction of this complex with the rDNA locus in vivo (59,60). The main components of a Pol II elongation factor, the FACT complex, were present in the E-pro domain purification. In summary, for many of the factors identified by MS in vivo interaction with the respective domain have been reported, validating our approach. Finally, many proteins with unknown function in rDNA biology were also identified in these purifications ( Table 2).
Two uncharacterized proteins, Tbs1 and Ylr278c, and the INO80 complex associate with rDNA chromatin in vivo We arbitrarily chose three proteins co-purifying with the ARS or E-Pro domains, Yll054c, Tbs1 or Ylr278c, for which interaction with rDNA had not been previously reported. To verify their interaction with rDNA in vivo we used ChEC analysis (61), which has been used in the past to identify many protein-rDNA interactions (26,35). Yeast cells expressing the endogenous factors fused to micrococcal nuclease (MNase) were crosslinked with formaldehyde. Nuclei were prepared, and calcium was added to induce cleavage by the MNase fusion proteins, which can be mapped to the DNA sequence by indirect endlabeling Southern blot analysis with a probe detecting a fragment containing the IGS of the rDNA locus. While Yll054c-MNase did not cleave within this rDNA region ( Figure 4, lanes 1-5), weak Tbs1-MNase dependent cleavage within the ARS region could be detected ( Figure 4C, lanes 6-10, marked by asterisks). ChEC analysis with Ylr278c-MNase expressing strains also showed cleavage within the ARS region and a relatively strong cleavage within the E-pro region ( Figure 4A, lanes [11][12][13][14][15]. Tbs1 and Ylr278c belong to a class of yeast transcription factors bearing a Zn(II)2Cys6 binuclear cluster DNA binding motif (62). A consensus sequence for Ylr278c (63) and a recognition sequence for Tbs1 (63,64) containing one mismatch are located within the E-Pro and ARS regions, respectively. ChEC was also performed with yeast cells expressing MNase fusion proteins of the Ies1, Ies4 and Arp4 subunits of the INO80 complex, the Isw1 protein and the Fpr4 protein ( Figure 4B). Isw1 and Fpr4, for which rDNA interaction had been reported (59,60,65), were were found to be enriched in 35S rDNA chromatin preparations and included as controls in these analyses. Southern blot analysis was performed with a probe detecting a fragment containing a large part of the 35S rDNA including the Pol I promoter region ( Figure 4B). For all MNase fusion proteins, we observed cleavage within this sequence. Some cleavage mediated by the MNase fusion proteins was also observed outside of the 35S rDNA (Supplementary Figure S3). We note that MNase fusion protein mediated cleavage was weak but specific in all cases, probably indicating that these factors associate with a subpopulation of rDNA repeats. These results support the established rDNA interactions of the Isw1 and Fpr4 proteins and suggest that the INO80 complex and two proteins with yet unknown function interact with rDNA chromatin in vivo, indicating that the native purification approach is suitable to detect novel chromatin components.

Major structural features of 5S rRNA gene chromatin are preserved upon isolation
Structural integrity of the 5S rDNA domain was verified by treating purified chromatin or the chromosomal 5S rRNA gene in yeast nuclei with restriction enzymes. Samples were treated with five different enzymes with recognition sites within the 5S rRNA gene sequence and in the flanking regions, respectively ( Figure 5A). DNA was isolated and analyzed by Southern blot. Overall restriction enzyme DNA accessibilities were similar in the isolated chromatin domain and on the chromosome in isolated yeast nuclei ( Figure 5A, compare upper and lower panel on the left, see graph on the right for quantitation). We conclude that major structural properties of 5S rDNA chromatin are preserved upon purification.

EM analysis determines nucleosome positions at 5S rDNA
Restriction enzyme accessibility allows evaluating the average nucleosomal protection against nuclease attack in a population of chromatin rings. Additionally, nucleosome occupancies and positions on single chromatin molecules could be visualized by EM (66,67). Isolated 5S rDNA rings were crosslinked with psoralen. Psoralen intercalates into double-stranded nucleic acids and establishes covalent crosslinks between DNA strands upon irradiation with long wave ultraviolet light. DNA assembled into a nucleosome is protected from psoralen incorporation leaving a footprint of the nucleosome of $150 bp of uncrosslinked DNA ( Figure 5B). DNA isolated from psoralen treated purified 5S rDNA was analyzed by EM under denaturing conditions. In these analyses, 5S rDNA molecules appeared as a circle with defined crosslinked double-stranded regions and singlestranded DNA bubbles (see Figure 5B electron micrograph on the right; 150 bp scale to $50 nm).
DNA was linearized such that the 5S rRNA gene was in the center of the molecule to map the positions of the observed single-stranded DNA bubbles ( Figure 5C). A total of 334 linear DNA molecules were evaluated and classified into 12 different groups according to number, position and size of single-stranded DNA bubbles. None of these classes was strongly overrepresented indicating marked heterogeneity in chromatin structure within the region flanking and encompassing the 5S rRNA gene. Analysis of bubble size showed that 57% of the bubbles had either a size expected for the protection by one nucleosome (130-180 bp) or two nucleosomes (280-360 bp) ( Figure 5D). Around 26% of the DNA bubbles had an intermediate size between 180 and 260 bp, which might be the consequence of incomplete crosslinking of naked DNA due to sequence preferences for psoralencrosslinking, or protection by chromatin components other than nucleosomes. The 5S rRNA coding sequence was psoralen crosslinked in three different molecule classes representing 31% of the analyzed molecules ( Figure 5C, left panels, position of the 5S rRNA gene marked by a black bar), which likely represented the nucleosome-free transcriptional active state of the gene (68). These results are in good agreement with the restriction enzyme accessibility in the purified chromatin domain, where 33-36% of the molecules are cut by SfcI in the center of the 5S rRNA coding sequence ( Figure 5A) and  Only proteins with an average iTRAQ ratio of at least 1.5, identified with at least 2 peptides are depicted. If 50% of multi-protein complex components were identified, identified subunits are highlighted in bold and additional proteins are depicted.
a Identified with an average iTRAQ ratio of at least 1.5 and 1 peptide. b Identified with an average iTRAQ ratio greater than one and at least two peptides.
with the observation that up to one-third of the 5S rRNA genes on the chromosome are actively transcribed (69).

MS identifies proteins co-purifying with the single copy PHO5 gene
To test whether the approach is suitable to purify and analyze chromatin from a single copy locus, PHO5 gene chromatin was isolated from a strain in which the gene was tagged with a cluster of LexA-binding sites and flanked by RS elements. To limit background contaminations, a second affinity purification step using calmodulin sepharose was introduced. DNA analysis of different samples of the TAP revealed a single DNA band at the expected size for the linearized PHO5 gene ring in the final elution ( Figure 6A, lane 9). Around 50 fmoles of purified ring DNA could be obtained from ten liters of exponentially growing yeast culture ($10 12 cells) corresponding to a recovery of $3% of total PHO5 loci (Table 1). In the final elution fraction of the TAP, the specific chromosomal domain was >100 000-fold enriched over another single copy locus (Table 1). Equal enrichment of the LexA-CBP bait proteins in the final elution of the control and PHO5 chromatin preparation was verified by SDS-PAGE and silver staining ( Figure 6B). Proteins migrating with the mobility of histone proteins were detected in the sample containing the purified PHO5 domain but not in the control purification ( Figure 6B, lanes 1 and 2). The amount of histone proteins co-purifying with the PHO5 ring correlated well with the expectation from the amount of ring DNA analyzed in the gel. The protein composition of purified PHO5 gene chromatin was further analyzed by semiquantitative MS comparing ring purifications with control purifications. Chromatin rings were isolated from two different strains in which transcription of the PHO5 gene was either repressed or constitutively activated by the deletion of the PHO80 gene (70,71). Transcriptional activation leads to drastic structural changes in PHO5 promoter chromatin (72,73). In good accordance with the results of the SDS-PAGE analysis, canonical histone proteins were specifically enriched in all PHO5 chromatin purifications in comparison with purifications from the respective control strains lacking RS elements and LexA-binding sites ( Figure 6C). Probably owing to the more stringent purification conditions, only a few proteins besides the canonical histones met our criteria for enrichment in these purifications. One of these proteins Snf6, a subunit of the SWI/SNF chromatin remodeling complex, co-purified specifically with the activated PHO5 chromatin ( Figure 6C and Supplementary Data Set 2 for full MS summary).
Snf6 associates with the activated PHO5 promoter region in vivo SWI/SNF has been reported to play a role in the activation of PHO5 expression on phosphate starvation (74)(75)(76), as well as mitotic induction of the gene (77). ChEC was used to investigate Snf6-MNase-mediated cleavage in the parental yeast strains used for the purification of the PHO5 chromatin domains being repressed or constitutively activated in transcription. We also created strains expressing TBP/Spt15 and a subunit of RNA Pol II, Rpb3, as MNase fusion proteins. After ChEC, isolated DNA was analyzed by Southern blot analysis using a probe detecting a fragment containing the PHO5 locus ( Figure 7A). Significant TBP/Spt15-MNase-mediated cleavage events at the PHO5 promoter were observed in the strain carrying the constitutive active gene, whereas only little cleavage was observed in the strain in which PHO5 expression was repressed ( Figure 7A, compare  lanes 1-4 with 5-8, asterisks mark cleavage events at the transcriptionally repressed and active PHO5 locus, whereas triangles label cleavage events specifically detected at the transcriptionally active PHO5 locus). This was in good agreement with ChEC results in TBP/Spt15-MNase-expressing strains when PHO5 gene expression was induced upon phosphate starvation (78).
A very similar picture was obtained for Rpb3-MNase with cleavage events extending into the coding sequence (CDS) of the constitutively active PHO5 gene ( Figure 7A, compare lanes 9-12 with 13-16, a bar marks the cleavage events within the CDS). Enhanced interaction of TBP/Spt15 and Rbp3 with the PHO5 locus correlates well with its activation in the PHO80 deletion strain. Finally, weak cleavage events mediated by Snf6-MN were detected mainly within the PHO5 promoter region in conditions when the gene was actively transcribed, whereas very limited cleavage occurred when PHO5 transcription was repressed ( Figure 7A, compare lanes 17-20 with lanes 21-24). As a control, the same membrane was subsequently hybridized with a probe detecting a fragment containing the IGS of the rDNA ( Figure 7B). Cleavage events were observed for all of the MNase fusion proteins within this region. In contrast to the results obtained at the PHO5 locus, the MNase fusion proteins cleaved to a similar extent within the IGS region, regardless if PHO5 expression had been activated or repressed ( Figure 7B compare lanes 1-4, 9-12, and 17-20 with lanes 5-8, 13-16, and 21-24, respectively). As observed earlier (35), TBP/Spt15-MNase dependent cleavage events could be observed at the Pol I promoter, the Pol III promoter, E-Pro and the ARS region ( Figure 7B, lanes 1-8), consistent with the results of the western blot analysis ( Figure 2B). Rpb3-MNase and Snf6-MNase-mediated cleavage was mainly restricted to the E-pro region, with additional weak cleavage events of Rbp3-MN at the ARS and at a site directly upstream of the Pol I promoter ( Figure 7B, lanes 9-16 and 17-24). Taken together, we conclude that the analysis of histone molecules and the identification of chromatin components at many single copy genomic loci in yeast might be feasible building on the herein described purification protocol.

DISCUSSION
The possibility to purify distinct chromosomal domains under native conditions from yeast provides a solid basis to investigate chromatin biology. Similar approaches reported so far are restricted to chromatin on plasmids (15,16), which might not fully reflect the situation on a chromosome. Other attempts to derive information about chromatin composition at defined genomic locations made use of chemical crosslinking (12)(13)(14), which prevents subsequent structural or functional analyses of the isolated chromatin. Thus, the work presented here represents an important step towards the detailed characterization of native chromatin on a molecular level.
As described in the introduction, the current technique is partly based on previous work (7). Although the earlier study provided evidence that major features of chromatin isolated from a single copy locus were preserved on purification, the purified material did not allow for the determination of the protein composition because of high background contaminations (unpublished results). Furthermore, the identification of histone modifications and structural analysis of the purified material remained unachieved. Eventually, these analyses were only possible by rigorous modification of the earlier protocol, which involved creation of a large set of yeast strains each carrying genomic modifications allowing site-specific recombination at a distinct target locus. In addition, various yeast integration vectors were constructed to optimize endogenous expression levels of the recombinant LexA-TAP protein. This led to a substantial reduction of chromatin fragments, that were non-specifically enriched in the presence of excess LexA-TAP [data not shown and (16)]. Other important improvements were to partially adapt a protocol, which has been used for the isolation of ribonucleoprotein complexes in the past, and to turn to comparative MS to define specifically co-purifying proteomes. Finally, the successful purification of chromatin domains from the multi-copy rDNA locus helped defining the amounts of material needed for the isolation of chromatin from a single-copy gene locus that would be amenable to MS and EM analyses.
Chemical crosslinking has been used to stabilize chromatin upon purification (12)(13)(14). In a previous study, Pol I or Pol II were isolated from cells after crosslinking the proteins to their chromatin template, and co-purifying proteomes were compared by semiquantitative MS (58). As the only essential function of yeast Pol I is the transcription of the 35S rRNA gene (79), the Pol I proteome likely represents actively transcribed 35S rRNA gene chromatin, which is largely depleted of nucleosomes (26).
Strikingly, there was a strong overlap between the Pol I enriched proteome under crosslinking conditions and the proteome co-purifying with the native 35S rDNA domain [compare Table 2 with Figure 6B and C in (58)]. Amongst others, a large set of early acting ribosome biogenesis factors was identified in both studies, indicating that even certain co-transcriptional assembling ribonucleoprotein complexes (55)(56)(57)(58) partly persist throughout the native purification procedure. Some of these proteins had been previously identified as part of a nucleolar complex co-purifying with Pol I (80). In summary, only a few proteins enriched in the purification of chromatin fragments crosslinked to Pol I were not present in native 35S rDNA domain preparations (compare Table 2 with Figure 6B and C in (58)). In contrast, purified 35S rDNA chromatin additionally contained non-transcribed nucleosomal, rRNA gene chromatin evident from enrichment of histone proteins and chromatin remodeling complexes, which were either depleted or not present in the chromatin preparation crosslinking to Pol I (58). Thus, chemical crosslink might not be a necessity to preserve many of the protein DNA interactions in chromatin.
One of the few unexpected findings was that a significant subset of specific Pol II subunits co-purified with the 35S rDNA domain. It has been previously reported that the formation of extrachromosomal ribosomal circles in respiratory deficient yeast strains leads to synthesis of rRNA by Pol II instead of Pol I (50). The formation of extrachromosomal ribosomal circle resembles the release of rRNA genes by site-specific recombination used for isolation of the chromatin domains in this study. Thus, it is possible that rDNA recombination events are linked to Pol II recruitment to this locus. However, ChIP experiments did not support a direct correlation between rDNA circle generation by R recombinase and association of Pol II (Supplementary Figure S3). Another explanation for the aforementioned observation could be that Pol II can access the rDNA locus on the chromosome to some extent in normal conditions. Pol II transcription was suggested to occur at different locations of rDNA, originating from the E-Pro region (81), a cryptic promoter within IGS2 (39), or a Pol II dependent gene located antisense within the 25S rDNA region [reviewed in (82)]. Additionally, there is evidence that Pol II can transcribe nucleosomal rRNA genes, which are not transcribed by Pol I (83). Thus, mass-spectrometric analysis might be sensitive enough to detect even a limited number of Pol II molecules associated with rDNA, and it cannot be excluded that the purification protocol partly selects for the subpopulation of genes transcribed by Pol II. Alternatively, Pol II association with the 35S rDNA domain may occur as an artifact of the purification at a step downstream of the recombination event.
EM analysis of DNA derived from psoralen treated rDNA chromatin purified from exponentially growing cells provided detailed information about nucleosome occupancy and positioning on the 5S rRNA gene sequence. Future research will aim to correlate different nucleosomal configurations with the transcriptional state of this locus. Comparative EM analyses of the single copy PHO5 gene in a transcriptionally activated and repressed state have thus provided further clues to elucidate the mechanism of transcriptional activation (see ''note added in proof'' for C.B. and H.B., accepted manuscript). Furthermore, structural analysis of purified chromatin under native conditions is likely to enhance our understanding of the structure-function relationship of genes in defined functional states.
Besides the benefit to obtain compositional and structural information, the purified native chromatin provides a highly defined template for in vitro experiments likely reflecting the in vivo situation with regard to nucleosome positioning and histone modification. This can be an advantage over the use of artificial, in vitro reconstituted nucleosomal arrays. Along these lines, purified chromatin of transcriptionally activated and repressed PHO5 genes has already been subjected to in vitro manipulation (7,84,85).