metaphase heritable chromatin comparison · chromosomes,andwediscuss therole ofmetaphasechroma- ......

5
Proc. Natl. Acad. Sci. USA Vol. 92, pp. 2379-2383, March 1995 Genetics Metaphase chromosome analysis by ligation-mediated PCR: Heritable chromatin structure and a comparison of active and inactive X chromosomes (cell memory/epigenetic mechanisms/heterochromatin/in vivo footprinting/mitosis) MATY HERSHKOVITZ AND ARTHUR D. RIGGS* Biology Department, Beckman Research Institute of the City of the Hope, Duarte, CA 91010 Communicated by Stanley M. Gartler, University of Washington, Seattle, WA, December 5, 1994 ABSTRACT We report that ligation-mediated PCR (LMPCR) can be used for high-resolution study of metaphase chromosomes, and we discuss the role of metaphase chroma- tin structure in the preservation of differentiated cell states. The X chromosome-linked human PGKI (phosphoglycerate kinase 1) promoter region was investigated, and euchromatic active X chromosome (Xa) metaphase chromatin was com- pared with interphase Xa chromatin and to heterochromatic inactive X chromosome (Xi) metaphase and interphase chro- matin. We find that (i) good-quality data at single-nucleotide resolution can be obtained by LMPCR analysis of dimethyl sulfate-treated intact metaphase cells; (ii) transcription fac- tors present on the Xa promoter of interphase chromatin are not present on metaphase chromatin, establishing that the transcription complex on the PGKI promoter must form de novo each cell generation; and (iii) the dimethyl sulfate reactivity pattern of Xa and Xi chromatin at metaphase is very similar to that of naked DNA. These results are discussed in the context of models for heritable chromatin structure and epigenetic mech- anisms for cell memory, and they are also relevant to more general aspects of chromatin structure and differences between euchromatin and heterochromatin. The preservation of cell phenotype and determined cell state from one cell generation to the next, often called cell memory (1-4), requires either the complete de novo reassembly each generation of very complex nucleoprotein structures or, more likely, the transfer between cell generations of epigenetic information that somehow guides correct reestablishment of chromatin structure and regulatory circuits. The error-free maintenance of proper cell identity is especially crucial for mammals, which have continued cell division of highly differ- entiated cells throughout their life-span. Despite its impor- tance, little is known about the epigenetic mechanisms in- volved in high-fidelity cell memory. Two likely mechanisms are DNA methylation and heritable chromatin structure. It has been established that a DNA methylation system involving clonally heritable patterns of 5-methylcytosine is necessary for normal mammalian development, parental imprinting, and stable maintenance of phenotype in cell culture (5, 6). Thus, cytosine DNA methylation seems clearly to be a component of mammalian cell memory mechanisms. However, mating type in yeast and position effect variegation in Drosophila have clonally heritable aspects, and in neither organism has 5-meth- ylcytosine or other modified bases yet been detected (7). For this reason, methylation-independent heritable chromatin structure could also be one component mechanism of mam- malian cell memory. Perhaps only several mechanisms acting together can provide the required fidelity of epigenetic infor- mation transfer. Parent Cell Interphase Metaphase _ E Alternate Models Decondensatio \ Decondensation and Reestablishment USiing and Reestablishment by Chromatin-Encoded Information De Novo Assembly Daughter Cell Interphase FIG. 1. Metaphase chromosome structure and cell memory. Infor- mation about the alternate activity states (active, open circles; inactive, filled circles) of a gene, chromosomal domain, or an entire chromosome (in the case of the X chromosome) can be either preserved (left pathway) or erased (right pathway) from the chromatin of a metaphase chromo- some when a cell undergoes mitosis; If no information is retained on the metaphase chromosome, then a very complex pattern of active and inactive genes and chromosomal domains must be reestablished de novo each generation in the daughter cells. Mammalian X chromosome inactivation provides a good system for the study of cell memory, epigenetic mechanisms, and imprinting because differences in chromatin structure between the genetically active, euchromatic X chromosome (the Xa) and the inactive and heterochromatin-like X chro- mosome (the Xi) are clonally heritable and show parent-of- origin effects (8). The X chromosome contains several genes such as hypoxanthine phosphoribosyltransferase (HPR7T) and phosphoglycerate kinase (PGK1) that are ubiquitously ex- pressed from the Xa but transcriptionally silent on the Xi. The Xi is constantly in a nucleus containing all transcription factors necessary for efficient transcription of constitutive genes such as PGK1, but some aspect(s) of chromatin structure renders most regions of the Xi resistant and presumably inaccessible to transcription factors. It is especially difficult to imagine how resistance is stably preserved if the complex pattern of active and inactive regulatory complexes and/or chromosomal do- mains is created completely de novo after transcription, DNA replication, or, of most relevance for this paper, each cell generation. More likely, some kind of chromatin structural information survives S phase and metaphase and guides decondensation and activation of chromatin in the next cell generation (see Fig. 1). DNA replication has long been rec- ognized as a critical period for cell memory, and models Abbreviations: LMPCR, ligation-mediated PCR; Xa, active X chro- mosome; Xi, inactive X chromosome; DMS, dimethyl sulfate. *To whom reprint requests should be addressed. 2379 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Upload: dinhthuan

Post on 27-Apr-2018

258 views

Category:

Documents


7 download

TRANSCRIPT

Page 1: Metaphase Heritable chromatin comparison · chromosomes,andwediscuss therole ofmetaphasechroma- ... chromatin maturation following DNAreplication. ... reasoned that if heritable chromatin

Proc. Natl. Acad. Sci. USAVol. 92, pp. 2379-2383, March 1995Genetics

Metaphase chromosome analysis by ligation-mediated PCR:Heritable chromatin structure and a comparison of activeand inactive X chromosomes

(cell memory/epigenetic mechanisms/heterochromatin/in vivo footprinting/mitosis)

MATY HERSHKOVITZ AND ARTHUR D. RIGGS*Biology Department, Beckman Research Institute of the City of the Hope, Duarte, CA 91010

Communicated by Stanley M. Gartler, University of Washington, Seattle, WA, December 5, 1994

ABSTRACT We report that ligation-mediated PCR(LMPCR) can be used for high-resolution study of metaphasechromosomes, and we discuss the role of metaphase chroma-tin structure in the preservation of differentiated cell states.The X chromosome-linked human PGKI (phosphoglyceratekinase 1) promoter region was investigated, and euchromaticactive X chromosome (Xa) metaphase chromatin was com-pared with interphase Xa chromatin and to heterochromaticinactive X chromosome (Xi) metaphase and interphase chro-matin. We find that (i) good-quality data at single-nucleotideresolution can be obtained by LMPCR analysis of dimethylsulfate-treated intact metaphase cells; (ii) transcription fac-tors present on the Xa promoter of interphase chromatin arenot present on metaphase chromatin, establishing that thetranscription complex on the PGKI promoter must form de novoeach cell generation; and (iii) the dimethyl sulfate reactivitypattern of Xa and Xi chromatin at metaphase is very similar tothat of naked DNA. These results are discussed in the context ofmodels for heritable chromatin structure and epigenetic mech-anisms for cell memory, and they are also relevant to moregeneral aspects of chromatin structure and differences betweeneuchromatin and heterochromatin.

The preservation of cell phenotype and determined cell statefrom one cell generation to the next, often called cell memory(1-4), requires either the complete de novo reassembly eachgeneration of very complex nucleoprotein structures or, morelikely, the transfer between cell generations of epigeneticinformation that somehow guides correct reestablishment ofchromatin structure and regulatory circuits. The error-freemaintenance of proper cell identity is especially crucial formammals, which have continued cell division of highly differ-entiated cells throughout their life-span. Despite its impor-tance, little is known about the epigenetic mechanisms in-volved in high-fidelity cell memory. Two likely mechanisms areDNA methylation and heritable chromatin structure. It hasbeen established that a DNA methylation system involvingclonally heritable patterns of 5-methylcytosine is necessary fornormal mammalian development, parental imprinting, andstable maintenance of phenotype in cell culture (5, 6). Thus,cytosine DNA methylation seems clearly to be a component ofmammalian cell memory mechanisms. However, mating typein yeast and position effect variegation in Drosophila haveclonally heritable aspects, and in neither organism has 5-meth-ylcytosine or other modified bases yet been detected (7). Forthis reason, methylation-independent heritable chromatinstructure could also be one component mechanism of mam-malian cell memory. Perhaps only several mechanisms actingtogether can provide the required fidelity of epigenetic infor-mation transfer.

Parent CellInterphase

Metaphase _ E AlternateModels

Decondensatio \ Decondensationand Reestablishment USiing and Reestablishment by

Chromatin-Encoded Information De Novo Assembly

Daughter CellInterphase

FIG. 1. Metaphase chromosome structure and cell memory. Infor-mation about the alternate activity states (active, open circles; inactive,filled circles) of a gene, chromosomal domain, or an entire chromosome(in the case of the X chromosome) can be either preserved (left pathway)or erased (right pathway) from the chromatin of a metaphase chromo-some when a cell undergoes mitosis; If no information is retained on themetaphase chromosome, then a very complex pattern of active andinactive genes and chromosomal domains must be reestablished de novoeach generation in the daughter cells.

Mammalian X chromosome inactivation provides a goodsystem for the study of cell memory, epigenetic mechanisms,and imprinting because differences in chromatin structurebetween the genetically active, euchromatic X chromosome(the Xa) and the inactive and heterochromatin-like X chro-mosome (the Xi) are clonally heritable and show parent-of-origin effects (8). The X chromosome contains several genessuch as hypoxanthine phosphoribosyltransferase (HPR7T) andphosphoglycerate kinase (PGK1) that are ubiquitously ex-pressed from the Xa but transcriptionally silent on the Xi. TheXi is constantly in a nucleus containing all transcription factorsnecessary for efficient transcription of constitutive genes suchas PGK1, but some aspect(s) of chromatin structure rendersmost regions of the Xi resistant and presumably inaccessible totranscription factors. It is especially difficult to imagine howresistance is stably preserved if the complex pattern of activeand inactive regulatory complexes and/or chromosomal do-mains is created completely de novo after transcription, DNAreplication, or, of most relevance for this paper, each cellgeneration. More likely, some kind of chromatin structuralinformation survives S phase and metaphase and guidesdecondensation and activation of chromatin in the next cellgeneration (see Fig. 1). DNA replication has long been rec-ognized as a critical period for cell memory, and models

Abbreviations: LMPCR, ligation-mediated PCR; Xa, active X chro-mosome; Xi, inactive X chromosome; DMS, dimethyl sulfate.*To whom reprint requests should be addressed.

2379

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

Page 2: Metaphase Heritable chromatin comparison · chromosomes,andwediscuss therole ofmetaphasechroma- ... chromatin maturation following DNAreplication. ... reasoned that if heritable chromatin

2380 Genetics: Hershkovitz and Riggs

requiring stable transcription complexes to persist on DNAthrough DNA replication and from one replication cycle to thenext have been one of the leading candidate mechanisms formethylation-independent cell memory (9, 10). However, it hasproven difficult to study the chromatin structure of specificsingle-copy genes during, for example, chromatin maturationfollowing DNA replication. For this reason, we have soughtother approaches to the study of heritable chromatin structureand cell memory. One approach we have investigated is the useof ligation-mediated PCR (LMPCR), a technique which re-cently has enabled high-resolution analysis of mammalianchromosomes and revealed differences in nucleoprotein struc-ture between the Xa and the Xi (11, 12). This paper reports onour first efforts to use LMPCR analysis of metaphase chro-mosomes as an approach to the study of epigenetic mecha-nisms.Chromatin becomes extremely condensed by coiling and

folding as cells approach division, reaching the maximallycompacted state in metaphase chromosomes (10, 13, 14). Wereasoned that if heritable chromatin structure passes epige-netic information to progeny cells, then this information mustbe retained in metaphase chromosomes. Thus some differ-ences between active and inactive chromatin are likely to bepreserved in metaphase chromosomes. Conversely, this logicsuggests that differences observed between, for example, theXa and the Xi may aid our understanding of heritable chro-matin structure and cell memory. We report here that LMPCRdoes enable the analysis of metaphase chromosomes at nucle-otide-level resolution. These initial studies also have shownthat transcription factors at the active PGK1 promoter areabsent from metaphase chromosomes. Thus the PGK1 tran-scription complex, and presumably, a myriad of other tran-scription complexes, must form de novo each cell generation.

MATERIALS AND METHODSCell Lines and Metaphase Cell Isolation. Chinese hamster

hybrid cells containing either the active human X chromosome(Y162-1lC) or the inactive human X chromosome (X8-6T6)were grown in RPMI 1640 medium with 10% bovine calfserum as previously described (15, 16). For metaphase cellpreparation, 3 x 107 cells were plated in 850-cm2 roller bottlesand grown with constant slow rolling at 2.5 rpm on a variable-speed roller apparatus (17). When the cells were 60-80%confluent, they were subjected to a 4-min rapid roller spin (200rpm) to remove all floating cells. The remaining attached cellswere then incubated with conditioned media and Colcemid at75 ng/ml for 2-3 hr. A second 5-min rapid spin detached mostly(>90%) cells in metaphase. Detached cells were quicklycollected by centrifugation (1650 x g, 5 min) and counted.Interphase cells were collected from the roller bottles aftermetaphase cell isolation.LMPCR and in Vivo Footprinting. Cells were washed and

resuspended at 2 x 106 cells per ml in serum-free medium,then treated with 0.1% dimethyl sulfate (DMS; Aldrich) inmedium without serum for 10 min at room temperature. Thereaction was stopped by adding 140 mM 2-mercaptoethanoland quickly diluting the mixture 10-fold with cold calcium- andmagnesium-free PBS solution (Irvine Scientific). DNA wasisolated from DMS-treated metaphase cells essentially asdescribed (11) except that the nuclei isolation step was omitted.The isolated genomic DNA was analyzed by alkaline agarosegel electrophoresis to determine by fragment size. ForLMPCR we used preparations of average fragment range of400-1000 bases from either metaphase or interphase cells. Toobtain the base-specific DNA cleavage we followed the stan-dard Maxam-Gilbert protocols for G, G+A, C+T, and Cbase-specific modifications (18), using genomic DNA isolatedfrom untreated Y162-11C or HeLa cells (for Xa experiments)and from X8-6T6 cells (for Xi experiments). Adenosine-

specific cleavage was generated by potassium tetrachloropal-ladate at pH 2.0 as described (19). Chemically modified DNAwas dissolved in 1 M piperidine and incubated for 30 min at88°C with precipitated ethanol, and dried extensively beforeresuspension in water. The human PGKI primer sets used forLMPCR analysis are illustrated in Fig. 3 and are the same aspreviously described (11). Primers Al, A2, and A3 (primer setA) and Gi, G2, and G3 (primer set G) bind to the "lower"strand of the human PGK1 promoter and reveal informationdownstream from their binding sites. Primers Hi, H2, and H3(primer set H) bind to the "upper" strand and reveal infor-mation upstream of position -38. First primers (Al, Gl, Hl)were used for primer extension, second primers (A2, G2, H2)were used for the PCR amplification step, and third primers(A3, G3, H3) were used to generate radiolabeled probes bylinear PCR. For the experiment shown in Fig. 4, LMPCRanalysis of the DMS-treated DNA samples was performed aspreviously described by Pfeifer et al. (11). Briefly, primer Hiwas used for primer extension by Sequenase (Version 2,United States Biochemical) followed by ligation of the gener-ated blunt ends to a common linker. Primer H2 and the linkerprimer were used for PCR amplification by AmpliTaq (Per-kin-Elmer/Cetus) for 19 cycles. Amplified DNA sequenceswere separated on sequencing gels, electroblotted, hybridizedto a single-strand probe, and visualized by autoradiography asdescribed (11, 20). For the experiment shown in Fig. 5, aLMPCR method using Vent (exo-) DNA polymerase (NewEngland Biolabs) for first primer extension and Vent poly-merase (New England Biolabs) for PCR was performed es-sentially as described (21, 22). The PCR amplification step wasstarted with 6-min denaturation at 96°C followed by a "stepdown" procedure, the first cycle of which was at 65°C for 3 minand 75°C for 3 min. After 1 min at 95°C, the annealing step ofthe second cycle was decreased to 64°C, and then 18 cycleswere done at 95°C for 1 min, 63°C for 2 min, and 75°C for 3min.

Synchronization Analysis. Mitotic cells were resuspended inhigh-citrate fixative (0.5 M sodium citrate, pH 2.35), incubatedfor 48 hr at 4°C in the dark, treated with RNase (BoehringerMannheim DNase-free RNase; 0.2 ,tg/ml) for 30 min at 37°Cin 0.1% bovine serum albumin and 10 mM Hepes solution (pH7.0), resuspended in 1 ml of propidium iodide at 100 mg/ml,incubated 16 hr at room temperature in the dark, and analyzedby flow cytometry as described (23).

RESULTSIsolation of Metaphase Cells. Previous studies of DMS-

treated tissue culture cells or blood lymphocytes demonstratedthat LMPCR enables true in vivo footprinting and high-resolution analysis of chromatin structure at the human PGK1locus (11, 12). Routinely, 1 p,g of genomic DNA per gel laneis used for LMPCR, but with optimization of the procedure, aslittle as 0.1 ,tg can suffice. For this reason we thought thatanalysis of metaphase chromosomes would be feasible if theycould be obtained adequately pure. Initial studies showed thatmitotic cells obtained by manually shaking them off the bottlesurface were indeed adequately pure but difficult to get insufficient quantities. We therefore used an automated rollerbottle apparatus and rapid spinning to dislodge metaphasecells, as described by Klevecz (17). Fig. 2 shows cell sorting dataof the isolated metaphase cells and indicates that about 90%are in the G2/M peak. Microscopic examination confirmedthat the peak labeled G2/M consisted of cells in metaphase.The results shown in Fig. 2 are for X8-6T2 cells, which havehuman Xi. Similar purity was obtained for Y162-11C cells,which have human Xa. To obtain a better yield of metaphasecells, and because preliminary experiments showed that col-lected metaphase cells progressed into G1 during the period ofDMS treatment, the mitotic arresting agent Colcemid was

Proc. NatL Acad ScL USA 92 (1995)

Page 3: Metaphase Heritable chromatin comparison · chromosomes,andwediscuss therole ofmetaphasechroma- ... chromatin maturation following DNAreplication. ... reasoned that if heritable chromatin

Proc. NatL Acad Sci USA 92 (1995) 2381

G Met Int

NFIl-Iike 4-138.<-138

i..i111<.t=_

CCAAT

FIG. 2. DNA histogram from flow-cytometer analysis of isolatedmetaphase X8-6T6 cells. The G1 peak designates cells before DNAreplication, and the G2/M peak designates the position of cells afterDNA replication. Cells were isolated by the mitotic shake off proce-dure, fixed, and stained as described in Materials and Methods.

added 2-3 hr prior to mitotic shake off. Control experimentsestablished that without DMS treatment the collected meta-phase cells remained viable and the Xi of X8-6T6 cells remainedinactive, as evidenced by assay for reactivation of hypoxanthinephosphoribosyltransferase (data not shown).Chromatin Structure of the Active Human PGKI Promoter

in Interphase and Metaphase Y162-11C Cells. Previouslypublished in vivo footprinting studies (11, 12) of the activehuman PGK1 promoter revealed five strong footprints, indic-ative of a transcription complex. The footprints seen were attwo GC-box regions (-32 to -63 and -235 to -255), aCCAAT region (-84 to -115), an NF-1-like region (-123 to-160), and a region from -187 to -223 (Fig. 3). The latterregion was not named previously although a footprint wasdetected. We have named it here as HIF-1-like because theprotected sequence contains a region that is 89% identical tothe mouse hypoxia-induced factor 1 (HIF 1) binding site (24).In all our experiments the same DMS concentrations wereused on metaphase and interphase cells, resulting in similarfragment size distributions. This fact by itself suggests that theoverall accessibility to DMS of chromatin in interphase andmetaphase cells is similar. LMPCR analysis confirmed thisobservation, since lanes of interphase and metaphase DNAhave an overall similar band intensity. Footprints were iden-tified by comparing the intensities ofbands of the DNA treatedin vivo to those of nakedDNA as well as comparing the relativeintensities of bands in the same lane.

In vivo DMS footprinting of interphase (cells remainingattached after removal of mitotic cells) Y162-11C cells hasreproduced the footprints that were previously reported on theactive human PGKI promoter region. Protected guanosineresidues were seen in various regions of the active promoter(Fig. 4). Altered DMS reactivity was observed at the GC-boxregion as well as in the CCAAT region. Protected and

set G

set A

-80GC-box HIF1 like NF1 like CCAAT

GC-boxes rD

set H

+110

FIG. 3. Footprints on the human PGK1 promoter. Boxes indicatethe positions of footprints as previously described (11, 12) andconfirmed by us. The arrows indicate the primer sets that were usedto analyze the promoter region by LMPCR. The transcription start siteis position + 1.

GC-box 4

4

-44

.4

43

4 -38

FIG. 4. In vivo DMS footprinting of the active human PGK1promoter region in interphase (Int) and metaphase (Met) cells. Cellswere treated with DMS, DNA was isolated, and LMPCR was per-formed according to the Sequenase/Taq procedure. Primer set H wasused for LMPCR analysis of the upper strand. A naked DNA controlis in lane G. Open triangles indicate that changes in band intensityappear only in interphase DNA. Footprint regions are indicated, withnames referring to consensus binding sites of transcription factors.

hypersensitive sites were also seen at the upstream GC-boxregion, the NF-1-like region, and the HIF-1-like region (datanot shown).

In contrast to interphase cells, metaphase cells have no DMSfootprints on the active humanPGKI promoter region (Fig. 4).Note that the Xa chromatin at the promoter in metaphase cellsexhibits the same DMS reactivity pattern as naked DNA andis thus quite different from interphase chromatin. Theseresults indicate that the transcription factors that activate thehuman PGKI gene during interphase are not detectablypresent on the promoter of metaphase chromosomes.Chromatin Structure of the Inactive Human PGKI Pro-

moter in Interphase and Metaphase X8-6T6 Cells. PreviousLMPCR studies showed that the inactive human PGKI pro-moter region is fully methylated and has no detectable tran-scription factor footprints (11, 12). Our DMS in vivo footprint-ing studies of the PGK1 promoter in interphase or metaphaseX8-6T6 cells, which contain the human Xi, have confirmed, as

expected, the absence of the transcription factors that wereseen on the active promoter (Fig. 5). Moreover, the in vivoDMS reactivity pattern of the inactive human PGKI promoteris very similar in interphase and metaphase (Fig. 5). In initialstudies using Sequenase and Taq polymerases for LMPCR,very subtle differences were reproducibly seen between thepattern for naked DNA and in vivo interphase or metaphasechromatin. We conclude, though, that Xa and Xi chromatinreact very similarly with DMS because, as shown in Fig. 5,significant differences have not been seen when Vent (exo-)and Vent were used as polymerases for LMPCR. Xi interphase

G2/MC-,

a)

ur4-

0

a)CK

DNA Content

Genetics: Hershkovitz and Riggs

Page 4: Metaphase Heritable chromatin comparison · chromosomes,andwediscuss therole ofmetaphasechroma- ... chromatin maturation following DNAreplication. ... reasoned that if heritable chromatin

2382 Genetics: Hershkovitz and Riggs

- -~44

FIG. 5. In vivo DMS footprinting of the inactive human PGKJ

promoter region at interphase (Int) and metaphase (Met). Cells were

treated with DMS, DNA was isolated, and LMPCR was performed

according to the Vent (exo-)/Vent procedure. Primer set A was used

for LMPCR analysis of the lower strand. Lanes CT, A, and G are

chemically cleaved naked DNA controls.

and metaphase chromatin also are very similar. It should be

noted that the DMS pattern seen for Xa naked DNA and

metaphase chromatin is not exactly the same as the pattern for

Xi naked DNA and metaphase chromatin. This is because the

Xi promoter region is heavily methylated at CpG sequences

and methylation does cause slight changes in DMS reactivity

DISCUSSION

The detailed structure of metaphase chromosomes has been

much studied at the cytogenetic level (14, 25) but relativelylittle studied at the biochemical level. Metaphase chromo-

somes are in the most tightly compacted chromosome state,

approximately 10,000-fold condensed relative to extended

DNA. The most widely accepted model for chromosome

structure is that DNA is wrapped around a core histone

octamer to form a 10-nm chain of nucleosomes, which is then

coiled into a 30-nm solenoid or chromatin fiber. This chro-

matin fiber is then folded into 75-kb average size loopstethered to a protein scaffold. The folded chromatin fiber is

about 250 nm wide and is further coiled into the 700-nm

chromatid of a metaphase chromosome (13, 14, 25).With respect to cell memory, the central question illustrated

in Fig. 1 is whether information engraved in chromatin struc-

ture is passed to progeny cells by way of metaphase chromatin.

Since metaphase chromosomes may be "stripped to their

essence," a study of metaphase chromatin thus may revealfeatures relevant to the epigenetics of cell memory. ForX-linked genes, such as the PGK1 gene studied here, it is well

established that cell memory is excellent, and the same alleleis monoallelically expressed in female cells at every cellgeneration. Also it is known that the Xi is methylated at CpGislands (8) and thus follows the left pathway of Fig. 1 withrespect to methylation. The important remaining questionrelevant for this discussion is whether information other thanmethylation is maintained through metaphase. For autosomalgenes, it is not known which pathway in Fig. 1 is followed.Information relevant to these questions has been heretoforevery limited, but some of the leading models for cell memoryinvoke stable transcription complexes that remain bound toDNA from one DNA replication cycle to the next (9, 10).These models implicitly predict that nucleoprotein complexesremain on metaphase chromosomes. A recent version of themodel suggests that enhancer and promoter complexes areoften separated by considerable distances to help preserveinformation through replication (10, 26) by having one nucleo-protein complex guide reformation of the other. For thismodel to apply at metaphase, one of the complexes mustremain intact on the chromosome. Only a few specific proteins,such as topoisomerase II, have been identified so far ascomponents of metaphase chromosomes (14, 25), but detec-tion of specific factors bound only at relatively few sites wouldbe difficult without the use of LMPCR. There is one reportthat RNA polymerase may remain on metaphase chromo-somes (27), but transcription from metaphase chromosomes isknown to be low or absent (28). Absence of transcription,though, is weak evidence with regard to transcription com-plexes because, obviously, the complexes could be present butinactive. Beyond the models suggesting transcription com-plexes that are stable to replication, evidence favoring reten-tion of specific nucleoprotein complexes on metaphase chro-mosomes is twofold. First, Groudine and Conklin (29) ob-served that hypersensitive sites (enhancer or promotercomplexes?) are retained in condensed sperm chromatin.Second, and most directly relevant, cytogenetic studies (30)and studies of isolated chromosomes (31) indicated that themetaphase Xa is more sensitive to DNase I than is themetaphase Xi. This differential nuclease sensitivity could bedue to retention of generalized DNase I sensitivity and/or toretention of hypersensitive sites.The aim of this initial study was to demonstrate the feasi-

bility of using the high-resolution technique of LMPCR toprovide information on the in vivo chromatin structure ofmetaphase chromosomes. We find that LMPCR can indeed beused for DMS footprinting of metaphase chromatin in intactcells. The region analyzed in these pilot experiments is thehuman PGK1 promoter, which has been previously well stud-ied by LMPCR and is known to have several transcriptionfactor footprints on the Xa and a wrapped-DNA, presumablynucleosomal, structure on the Xi (12). We have confirmedhere the presence of transcription factors on interphase chro-matin and have shown that these factors are not on metaphasechromosomes; thus at least the PGK1 transcription complexforms de novo at each cell generation. Our present data giveno information as to the mechanism of removal, but theycertainly are consistent with its being the result of phosphor-ylation of a component of the polymerase or other transcrip-tion complex machinery, as suggested by Gottesfeld et al. (32).The metaphase chromatin of the PGKI promoter on the

active Xa chromosome has a DMS reactivity pattern similar tothe naked DNA control. Thus even the most extremelycondensed form of previously active chromatin is still trans-parent to DMS. This is not inconsistent with nucleosomalstructure, because previous work has shown that nucleosomalstructure does not affect DMS reactivity (33, 34). Numerousfootprinting studies have established, however, that mostDNA-bound proteins cause a specific pattern of decreased or,rather commonly, increased DMS reactivity (35). Interestingly,we find a different picture for the inactive, Xi, chromosome,

Proc. Natt Acad ScL USA 92 (1995)

Page 5: Metaphase Heritable chromatin comparison · chromosomes,andwediscuss therole ofmetaphasechroma- ... chromatin maturation following DNAreplication. ... reasoned that if heritable chromatin

Proc. Natl. Acad. Sci USA 92 (1995) 2383

which in interphase cells is cytologically like heterochromatin.Here we find that the reactivity patterns of interphase andmetaphase chromatin are similar. The inactive X chromosomein female mammals recently has been shown to be distin-guished by a lack of histone H4 acetylation (36), a chromatinfeature long associated with gene expression (34, 37). It iscertainly possible that the lack of histone acetylation mightcause some perturbation of nucleosomal structure. However,recent in vitro hydroxyl radical footprinting studies (34) haveconfirmed earlier work (33) that the location and extent ofnucleosomal protein-DNA contacts is not altered by acetyla-tion, so an effect on DMS reactivity is not likely. It is worthnoting that the metaphase-condensed chromatin at the Xapromoter, as probed by DMS, is the same as naked DNAdespite the heavy phosphorylation of metaphase histones H3and Hi, which is thought to have perhaps triggered conden-sation (38). The LMPCR approach now needs to be extendedin many ways, including the use of other footprinting agentsand analysis of other genes with candidate stable transcriptioncomplexes, and other elements such as enhancers, matrixattachment sites, locus control regions, and replication origins.

We thank Dr. R. R. Klevecz for his advice in isolation andcharacterization of mitotic cells and Dr. Z. Gebreyes for help with theVent experiments. This work was supported by National Institutes ofHealth Grant GM50575 to A.D.R.

1. Alberts, B., Bray, D., Lewis, J., Roberts, K. & Watson, J. D.(1983) Molecular Biology of the Cell (Garland, New York).

2. Riggs, A. D. (1989) Cell Biophys. 15, 1-13.3. Riggs, A. D. (1990) Philos. Trans. R. Soc. London B 326, 285-297.4. Paro, R. (1993) Curr. Opin. Cell Biol. 5, 999-1005.5. Li, E., Beard, C. & Jaenisch, R. (1993) Nature (London) 366,

362-365.6. Jost, J. P. & Saluz, H. P., eds. (1993) DNA Methylation: Molecular

Biology and Biological Significance (Birkhauser, Boston).7. Rivier, D. H. & Rine, J. (1992) Curr. Opin. Genet. Dev. 2,

286-292.8. Riggs, A. D. & Pfeifer, G. P. (1992) Trends Genet. 8, 169-174.9. Brown, D. D. (1984) Cell 37, 359-365.

10. Wolffe, A. (1992) Chromatin (Academic, New York), pp. 100-166.

11. Pfeifer, G. P., Tanguay, R. L., Steigerwald, S. D. & Riggs, A. D.(1990) Genes Dev. 4, 1277-1287.

12. Pfeifer, G. P. & Riggs, A. D. (1991) Genes Dev. 5, 1102-1113.13. Van Holde, K. E. (1988) Chromatin (Springer, New York).14. Manuelidis, L. (1990) Science 250, 1533-1540.15. Hansen, R. S., Ellis, N. A. & Gartler, S. M. (1988) Mol. Cell. Biol.

8, 4692-4699.16. Pfeifer, G. P., Steigerwald, S. D., Hansen, R. S., Gartler, S. M. &

Riggs, A. D. (1990) Proc. Natl. Acad. Sci. USA 87, 8252-8256.17. Klevecz, R. R. (1975) Methods Cell Biol. 10, 157-172.18. Maxam, A. M. & Gilbert, W. (1980) Methods Enzymol. 65,

499-560.19. Iverson, B. L. & Dervan, P. B. (1987) Nucleic Acids Res. 15,

7823-7830.20. Pfeifer, G. P., Singer-Sam, J. & Riggs, A. D. (1993) Methods

Enzymol. 225, 567-583.21. Hornstra, I. K., Nelson, D. L., Warren, S. T. & Yang, T. P. (1993)

Hum. Mol. Genet. 2, 1659-1665.22. Tommasi, S., LeBon, J. M., Riggs, A. D. & Singer-Sam, J. (1993)

Somat. Cell Mol. Genet. 19, 529-541.23. Klevecz, R. R. & Braly, P. S. (1991) Ann. N.Y Acad. Sci. 618,

257-276.24. Firth, J. D., Ebert, B. L., Pugh, C. W. & Ratcliffe, P. J. (1994)

Proc. Natl. Acad. Sci. USA 91, 6496-6500.25. Saitoh, Y. & Laemmli, U. K. (1994) Cell 76, 609-622.26. Wolffe, A. P. (1990) Prog. Clin. Biol. Res. 322, 171-186.27. Matsui, S.-I., Weinfeld, H. & Sandberg, A. A. (1979) J. Cell Biol.

80, 451-464.28. Prescott, D. M. & Bender, M. A. (1962) Exp. Cell Res. 26,

260-268.29. Groudine, M. & Conklin, K.-F. (1985) Science 228, 1061-1068.30. Kerem, B. S., Goitein, R., Richler, C., Marcus, M. & Cedar, H.

(1983) Nature (London) 304, 88-90.31. Gazit, B. & Cedar, H. (1982) Science 217, 648-650.32. Gottesfeld, J. M., Wolf, V. J., Dang, T., Forbes, D. J. & Hart, P.

(1994) Science 263, 81-84.33. McGhee, J. D. & Felsenfeld, G. (1979) Proc. Natl. Acad. Sci. USA

76, 2133-2137.34. Bauer, W. R., Hayes, J. J., White, J. H. & Wolffe, A. P. (1994) J.

Mol. Biol. 236, 685-690.35. Riggs, A. D. & Pfeifer, G. P. (1994) Advances in Molecular and

Cell Biology (JAI, Greenwich, CT), in press.36. Jeppesen, P. & Turner, B. M. (1993) Cell 74, 281-289.37. Kornberg, R. D. & Lorch, Y. (1992) Annu. Rev. Cell Biol. 8,

563-587.38. Th'ng, J. P. H., Guo, X.-W., Swank, R. A., Crissman, H. A. &

Bradbury, E. M. (1994) J. Biol. Chem. 269, 9568-9573.

Genetics: Hershkovitz and Riggs