Methylation of the FSHD Syndrome-Linked Subtelomeric Repeat in Normal and FSHD Cell Cultures and Tissues

Page 1

Siiqolff

Methylation of the FSHD Syndrome-Linked Subtelomeric Repeatin Normal and FSHD Cell Cultures and Tissues1

Fern Tsien,* Baodong Sun,* Nancy Eddy Hopkins,*,† Vettaikorumakankav Vedanarayanan,‡Denise Figlewicz,§ Sara Winokur,\ and Melanie Ehrlich*,¶,2

*Human Genetics Program, †Department of Pediatrics, and ¶Department of Biochemistry, Tulane Medical School, New Orleans,Louisiana 70112; ‡Department of Neurology, University of Mississippi Medical School, Jackson, Mississippi 39216;

Molecular Genetics and Metabolism 74, 322–331 (2001)doi:10.1006/mgme.2001.3219, available online at http://www.idealibrary.com on

§Department of Neurology, University of Rochester Medical Center, Rochester, New York 14642;and \Department of Biological Chemistry, University of California, Irvine, California 92697

July 1

DoeqFncpl

Received May 8, 2001, and in revised form

Facioscapulohumeral muscular dystrophy (FSHD)has an unusual molecular etiology. In a putativelyheterochromatic subtelomeric region of each chro-mosome 4 homologue (4q35), unaffected individualshave 11 to about 95 tandem copies of a complex3.3-kb repeat (D4Z4). Most FSHD patients have lessthan 10 copies at one allelic 4q35. This has beenproposed to lead to the loss of heterochromatiniza-tion and, thereby, inappropriate gene expression byposition effects, explaining the dominant nature ofFSHD and the role of a decreased number of copiesof D4Z4 at 4q35 but not at 10q26. Consistent with theproposed heterochromatinization of this repeat, bySouthern blot analysis, we found that SmaI, MluI,

acII, and EagI sites in D4Z4 are highly methylatedn normal and FSHD cell lines and somatic tissues,ncluding skeletal muscle. Like repeated DNA se-uences in the juxtacentromeric heterochromatinf chromosomes 1, 9, and 16, D4Z4 was hypomethy-ated at numerous CpGs in sperm and in cell linesrom patients with an unrelated DNA methyltrans-erase deficiency syndrome (ICF; immunodefi-

ciency, centromeric region instability, facial anom-alies) in contrast to its hypermethylation in non-ICF postnatal somatic tissues. Our data on FSHDsamples suggest that the disease-associated 4q35

1 This paper is dedicated to the memory of Dr. Kiichii Arahata,an excellent and enthusiastic gentleman-scientist and leader inthe field of FSHD research.

2

To whom correspondence should be addressed at Tulane Uni-versity School of Medicine, Human Genetics Program, SL31, 1430Tulane Avenue, New Orleans, LA 70112. Fax: (504) 584-1763.E-mail: [email protected].

3221096-7192/01 $35.00Copyright © 2001 by Academic PressAll rights of reproduction in any form reserved.

3, 2001; published online October 25, 2001

4Z4 repeats, which constitute a small percentagef the total D4Z4 repeats, are not generally hypom-thylated relative to the other repeats of this se-uence. However, in individuals not affected withSHD, the hypermethylation of tandem, high-copy-umber D4Z4 repeats might help stabilize hetero-hromatinization at allelic 4q35 regions just as hy-ermethylation elsewhere in the genome has been

inked to chromatin compaction. © 2001 Academic Press

Key Words: D4Z4; muscular dystrophy; DNA meth-ylation; CpG; 5-methylcytosine; FSHD; subtelomericrepeats; ICF syndrome.

Deletions of a 3.3-kb subtelomeric sequence(D4Z4) that is tandemly repeated at 4q35 are re-sponsible for the vast majority of cases of FSHD(facioscapulohumeral muscular dystrophy), an auto-somal dominant disease (1). D4Z4 repeats at 4q35are polymorphic in copy number. More than 90% ofFSHD patients have 1–9 copies of this repeat on oneof their chromosome 4 (Chr4) homologues instead of11 to about 95 copies on both homologues (2,3). Ithas been hypothesized that the dominant nature ofFSHD is due to the loss of a postulated heterochro-matic structure in the vicinity of the Chr4 D4Z4repeat region. The partial homology of subfragmentsof D4Z4 repeats to other DNA repeats found inknown heterochromatic portions of the genome (1qhand the heterochromatic regions on the short arms

of the acrocentric chromosomes (4–6)) and theirhigh number of tandem copies are consistent withthe proposed heterochromatic nature of the D4Z4

Page 2

n4

sgtTsqrstailcoetgbBb(

wstf7rcd

attmb

F TH

repeat region at 4q35. The loss of this heterochro-matin structure has been proposed to result in del-eterious activation of genes in cis when the copy

umber of D4Z4 is below a threshold level in oneq35 allelic region (4,5).It is apparently not the 4q35 D4Z4 repeats them-

elves in a low copy number that abnormally alterene expression and muscle function, but rather,heir effect on as-yet-unelucidated genes on Chr4.his is seen in analyses of D4Z4 repeats in theubtelomeric region of Chr10 (10q26). Of the se-uences that cross-hybridize with the 4q35 D4Z4epeats, only those in 10q26 appear to have theame sequence arrangement and tandem configura-ion (6,7). Furthermore, 10q26 has D4Z4 repeatslmost identical to those on 4q35 and they are sim-larly polymorphic in copy number (1,8). However,ow numbers of tandem copies of D4Z4 at 10q26onfer no phenotype. The disease-producing effectsf a low copy number of D4Z4 only at 4q35 are mostasily explained by a position effect similar to posi-ion effect variegation (PEV). In PEV, silencing ofenes occurs as a result of rearrangements thatring them in the vicinity of heterochromatin (9).ecause constitutive heterochromatin in verte-rates is so frequently rich in 5-methylcytosinem5C) residues (10) and because vertebrate DNA

methylation has been causally linked to chromatincompaction resulting in repression of transcription(11), we were interested in the methylation status ofthe D4Z4 repeat.

We recently found that the D4Z4 repeats arehighly methylated at EagI sites in normal lympho-blastoid cell lines (LCLs) but are hypomethylated inLCLs from patients with the ICF syndrome (immu-nodeficiency, centromeric region instability, facialanomalies (12)). ICF is a unique, recessively inher-ited DNA methyltransferase 3B deficiency diseasethat results in a small percentage reduction in thegenomic m5C content (13,14). The methylation ofD4Z4 in tissue samples, normal or FSHD, had notpreviously been reported and studies of unculturedtissue are important because of tissue-specific differ-ences in DNA methylation and changes in genomicmethylation that can occur during cell culture (15).In the present study, we examined D4Z4 methyl-ation at SmaI, MluI, SacII, and EagI sites in FSHD

METHYLATION O

and normal blood and skeletal muscle samples;FSHD, ICF, and normal LCLs; and a variety ofnormal tissues.

t

MATERIALS AND METHODS

Cell and tissue populations and DNA isolation.FSHD LCLs (s10 and s5) were obtained from K.Arahata and P. Lunt (16, Lunt, unpublished re-sults). LCLs from ICF patients (ICF C and ICF B)and control LCLs (GM02184D and AHH-1) werepreviously described (14,17). Another normal LCL(GM07057) was used. Muscle biopsy samples camefrom mildly affected biceps muscle of FSHD patientsand a patient with an unrelated neuromuscular dis-ease. Preexisting muscle biopsy samples had beenobtained for diagnostic purposes with informed con-sent. Term placenta (with the embryonic mem-branes removed first) and pooled normal semenwere used for DNA isolation. Other tissues werefrom autopsy samples of adult trauma victims; eachtissue was from a different individual. DNA waspurified by standard methods (15). For FSHD andnormal blood samples, the red blood cells were lysedand DNA was isolated from the remaining whiteblood cell fraction. Semen was treated with 0.04 Mdithiothreitol during the proteinase K incubation.

Southern blot hybridization. Genomic DNA (10mg, unless otherwise indicated) was incubated with4–8 U of enzyme/mg of DNA in 60-ml reaction mix-tures for 16–20 h (Roche Diagnostics, Indianapolis,IN; New England Biolabs, Beverly, MA; and GIBCO/BRL, Gaithersburg, MD). An internal control foreach combination of enzyme and human DNA wasmade to check for completeness of digestion, namely,taking a 10- or 20-ml aliquot of the digest immedi-ately after all the reaction components were mixedtogether and 1 mg (5 ml) of pUC18 or phage l DNA

as added for incubation in parallel with the testample. In all cases, complete digestion was ob-ained. The 700-bp hybridization probe was a sub-ragment of D4Z4 (GenBank AF039145) that starts8 bp proximal to the KpnI site of the 3.3-kb D4Z4epeat of Chr4. It was excised from its pBC KS 1/2loning vector (Stratagene, La Jolla, CA (12)) byigestion with EagI and EcoRV and the purified

insert was labeled with [a-32P]dCTP by random hex-mer priming. Electrophoresis was followed by par-ial depurination with 0.25 N HCl for 15 min, dena-uration with 0.4 N NaOH, blotting onto a nylonembrane, neutralization, UV crosslinking, and hy-

ridization with the probe at 18°C below the melting

323E FSHD REPEAT

emperature (Tm), unless otherwise noted. The rela-tive radioactivity in different bands was quantitatedby phosphorimaging.

Page 3

fpft

dD

N ET

RESULTS

Analysis of methylation of D4Z4 repeats in normalhuman tissues at SmaI and EagI sites. We ana-lyzed a variety of normal human tissues for methyl-ation at SmaI sites (59-CCCGGG-39), of which thereare 11 in the D4Z4 repeat of 4q35 and a similarnumber in the D4Z4 repeat of 10q26 (GenBankAF117653, AF039149, AF039158, and AF039163).For all the hybridizations in this study, we used a700-bp probe from the 3.3-kb D4Z4 repeat (Fig. 1B).From brain, lung, and liver, most of the DNA frag-ments hybridizing with the probe (at 18°C below theTm) were in a high-molecular-weight smear, indica-tive of hypermethylation, with only 2–3 and 1–2% ofthe signal in minor, hypomethylated 1.6- and 0.2-kbbands (Fig. 1 and data not shown). Also, ;3-kbragments with 3–6% of the signal were seen thatrobably result from partial methylation of a smallraction of D4Z4 repeats allowing cleavage at one ofhe five clustered SmaI sites near the middle of the

FIG. 1. SmaI sites in D4Z4 repeats from normal somatictissues are highly methylated but are hypomethylated in normalsperm. (A) Digests with SmaI were made from normal tissueDNAs (10 mg) and subjected to blot hybridization with a subfrag-ment of D4Z4 as a probe. Electrophoresis was on a 1.0% agarosegel. The sizes of the low-molecular-weight hybridizing fragmentsare indicated. The 0.2-kb band from the SmaI digest of spermDNA was visible in an overexposure of this blot and in other blots,especially when the hybridization temperature was increasedfrom the Tm minus 18°C to the Tm minus 14°C (Fig. 5A). (B) Amap of SmaI sites in the D4Z4 repeat and hybridizing DNAfragment sizes expected from unmethylated SmaI sites. Theprobe (thick line above the map) hybridizes mostly with thebeginning of the repeat but 18 bp are complementary to the endof the D4Z4 repeat (short line on the right above the map). TheKpnI site is indicated in this and subsequent restriction maps toshow the beginning of the D4Z4 repeat. The left side of the mapis the centromeric direction.

324 TSIE

repeat and then about 3 kb away in an adjacent copyof the repeat (Fig. 1).

In contrast to SmaI digests of somatic DNAs, in

the SmaI digest of sperm DNA, the only bands seenwere the 1.6- and 0.2-kb bands. These containedabout 37 and 17% of the signal, respectively (Fig. 1and data not shown), when the annealing tempera-ture was 18°C below the Tm. When the annealingtemperature was increased to Tm minus 14°C, moreof the signal was specific for D4Z4 repeats because atotal of about 70% of the signal from sperm DNAwas then in these two bands (data not shown). Evenunder these stringent annealing conditions, morethan 70% of the signal from SmaI digests of DNAfrom postnatal somatic tissues was in fragments of.12 kb (Fig. 5A and data not shown). The stringencyof these hybridization conditions was confirmed byusing them to look for the 3.3-kb D4Z4 band in KpnI

igests. Consistent with the results from spermNA digested with SmaI, about 65–75% of the sig-

nal was in a 3.3-kb band in these digests of somaticDNAs (data not shown). Under the less stringentconditions (Tm 2 18°C) about half of the signal wasin the 3.3-kb band. This 3.3-kb KpnI band should bederived only from tandem copies of D4Z4, which arepresent at 4q35 and 10q26. The partially homolo-gous repeats that are located on the acrocentricchromosomes contain other interspersed sequences(6, J. Hewitt, personal communication). We concludethat about half to three-fourths of the hybridizingfragments in our Southern blots were from the D4Z4repeats at 4q35 or 10q26. Consistent with this con-clusion are results from other types of enzymaticdigests in which the only discrete bands obtainedhad the sizes predicted for the tandem, highly ho-mologous 4q35 and 10q26 D4Z4 repeats (e.g., 0.8-,1.2-, 1.6-, 2.1-, 2.2-, 2.8-, and 3.3-kb fragments fromsperm or ICF DNA from sequences spanning thecentromeric and/or distal ends of the probe; Figs. 1,2, 4, and 5).

We next examined normal tissue DNAs digestedwith the CpG methylation-sensitive enzyme EagI(59-CGGCCG-39). In DNA digests from various post-natal somatic tissues, the vast majority of the hy-bridizing material was in high-molecular-weightfragments and only a faint hypomethylation-specific2.2-kb band with about 1% of the signal was ob-served upon overexposure of the blot (Fig. 2 and datanot shown). Also, about 1–2% of the signal fromthese digests was in monomer-sized, 3.3-kb hybrid-izing fragments that should have originated frompartial methylation at EagI sites. In contrast to

AL

somatic DNA digests, EagI digests of sperm DNAhad a major 2.2-kb band and only a minor 3.3-kbband.

Page 4

iydwEb

wh

F TH

Double digests with EagI and BlnI, which is in-sensitive to CpG methylation, allowed us to comparemethylation at the highly homologous 4q and 10qD4Z4 repeats in these blots (Fig. 2). There is a singleBlnI site in the D4Z4 repeat unit on 10q26 but nonen the typical 4q35 repeats (8). Therefore, unmeth-lated copies of the 10q D4Z4 repeat in these doubleigests should give only a 0.8-kb hybridizing bandhile copies of 4q D4Z4 repeats unmethylated at allagI sites should yield only a 2.2-kb hybridizingand (Fig. 2B). In EagI/BlnI double digests, 0.8-kb

bands containing 7, 6, 8, or 5% of the signal, and3.3-kb bands containing 5, 4, 26, or 21% of the signalwere obtained from spleen, brain, lung, and heartDNAs, respectively, but there were no detectable2.2-kb bands even in overexposed X-rays. However,in EagI/BlnI digests of sperm DNA, the only majorbands contained 2.2-kb (4q35-specific) and 0.8-kb(10q26-specific) fragments (Fig. 2B). Because there

FIG. 2. EagI sites at both 4q35 and 10q26 D4Z4 repeats arehighly methylated in normal somatic tissues but hypomethylatedin sperm. (A) Normal tissue DNAs were digested with EagI orEagI plus BlnI or BlnI alone and subjected to blot hybridizationas described in Fig. 1. In the double digests, the 0.8- and 2.2-kbbands indicate hypomethylation at EagI sites in 10q26 repeatsand 4q35 3.3-kb repeats, respectively. (B) Map showing how the2.2- and 0.8-kb hybridizing fragments are generated from adja-cent copies of the repeat. The BlnI site is present only in thecanonical 10q D4Z4 repeats.

METHYLATION O

was much more 3.3-kb signal in the EagI/BlnI dou-ble digests of somatic tissue DNAs than in EagIsingle digests, the 3.3-kb EagI/BlnI bands in Fig. 2

EarB

are derived mostly from 10q copies of the D4Z4repeat that are methylated at all their EagI sites butcleaved once per repeat by BlnI rather than fromhypomethylated 4q D4Z4 repeats cleaved at one ofthree EagI sites in successive copies of the tandemrepeat.

We also digested normal tissue DNAs with theCpG methylation-insensitive BglII together withEagI and BlnI. In 4q repeats in these triple digests,we could assess methylation of the same single EagIsite (at the distal end of the probe; Fig. 3B) as wasanalyzed for 10q repeats in EagI/BlnI double digests(Fig. 2B). Hybridizing fragments of 1.8 kb (EagI/BglII) and 3.3 kb (BglII/BglII) should be derivedonly from 4q-type D4Z4 because 10q copies will befurther cleaved by BlnI to 0.8- or 2.3-kb fragments(Fig. 3B). The percentages of 4q repeats unmethyl-ated at the single analyzed EagI site (1.8-kb signaldivided by the total signal in the 1.8- and 3.3-kbbands) were 19, 11, and 43% for lung, brain, andheart, respectively. In contrast, there was ,1% ofthe signal in the 2.2-kb band of EagI/BlnI digestsindicative of hypomethylation at both this EagI siteand the proximal EagI site in 4q D4Z4 repeats inthese same DNAs (Fig. 2).

FIG. 3. Analysis of methylation at a single EagI site in 4qD4Z4 repeats in normal somatic tissue DNAs. (A) DNA (20 mg)

as triple-digested with EagI, BglII, and BlnI and then blotybridized, as described in Fig. 1, to examine methylation of the

325E FSHD REPEAT

agI site at the distal end of the probe. ICF LCL DNA served ashypomethylated standard. (B) Map of two copies of the D4Z4

epeat showing the predicted fragments from either 4q D4Z4 (nolnI site) or 10q D4Z4 (BlnI site).

Page 5

pwDC

Detr

N ET

Analysis of methylation of D4Z4 repeats in ICFLCLs, normal LCLs, and normal tissues at MluI andSacII sites. By Southern blot analysis, we havepreviously shown that DNA from LCLs derived fromICF patients is hypomethylated at EagI sites com-

ared with normal LCLs (12). In the present study,e found that ICF LCL DNA and normal spermNA are hypomethylated at sites in D4Z4 for thepG methylation-sensitive enzymes MluI (59-

ACGCGT-39), SacII (59-CCGCGG-39), and SmaI andthat normal LCLs and normal somatic tissue DNAsare highly methylated at these sites (Figs. 4 and 5and data not shown). In SacII digests, the percent-

FIG. 4. SacII sites in D4Z4 are highly methylated in normalsomatic tissue DNAs but not in sperm and ICF LCL DNAs. (A, B)Southern blots containing SacII digests of normal tissue DNAs or

NA from a normal LCL (AHH-1) or an ICF LCL using a mod-rate exposure (A) or an overexposure to X-ray film (B). Unlike forhe previous figures, the hybridization temperature was 16°Cather than 18°C below the Tm. Less placental DNA was present

compared with the other DNAs but the amounts of DNA in theother lanes of this gel and in the lanes of Fig. 1, 2, 3, or 5 wereapproximately the same as seen by the ethidium bromide-inducedfluorescence of total DNA in the gels. (C) Map illustrating DNAfragments produced from completely or partially unmethylatedD4Z4 repeats. The SacII site in each repeat that is indicated byan asterisk is deduced to be present in canonical 4q D4Z4 repeatsbut not 10q D4Z4 repeats.

326 TSIE

ages of the combined signal in the 1.3- and 2.8-kbbands indicative of hypomethylation (see below)were 3–6, 67, 59, and 13% in normal postnatal tis-

dea

sue DNAs, an ICF LCL, normal sperm, and placen-tal DNA digests, respectively (Fig. 4). The largeramount of hypomethylation in the D4Z4 repeat ofplacental DNA than in postnatal somatic tissueDNAs can be explained by the largely trophoblasticorigin of placenta and is consistent with hypomethy-lation of many placental DNA sequences (15,18). InMluI digests, the hypomethylation-specific 1.2- and2.1-kb bands contained a total of 1–3, ,1, 42, and47% of the hybridization signal in DNA from normalpostnatal somatic tissues, a normal LCL, an ICFLCL, and normal sperm, respectively (Fig. 5 anddata not shown).

We found evidence that SacII digests may allowdiscrimination of hypomethylated 4q-derived D4Z4 re-peats from those of 10q. Partial and overlapping sub-clones of the 10q D4Z4 repeat have been sequenced(GenBank AF039149, AF039158, and AF039163) and

FIG. 5. FSHD LCLs are hypermethylated and ICF LCLs arehypomethylated at SmaI and MluI sites in D4Z4. (A) Hybridiza-tion was done on LCL and lung DNAs after digestion with SmaIor MluI. The annealing temperature for this blot hybridization

AL

was 14°C below the Tm. (B) A restriction map of two D4Z4 repeatsiagraming the smaller hybridizing products from DNA withither complete (1.2 or 2.1 kb) or partial (3.3 kb) hypomethylationt the MluI sites.

Page 6

F TH

indicate that one of the three SacII sites (denoted byan asterisk in Fig. 4C) that is present in the 4q D4Z4repeat (GenBank AF117653, D38024, and L32607)may generally be missing in the 10q repeat because ofa C-to-A substitution. Unrecombined D4Z4 repeats(19) at 4q35 display extremely high sequence conser-vation (,1% divergence; e.g., GenBank AF117653 andD38024). However, there is a little more divergencebetween 4q (AF117653) and 10q repeats (1.7%). Theprominent 1.3- and 2.8-kb bands in the ICF LCL andsperm DNA digests and the same minor bands insomatic tissue DNA digests may have arisen mostlyfrom unmethylated SacII sites in 4q and 10q D4Z4repeats, respectively (Fig. 4). The 1.9-kb bands fromsomatic tissues (1–2% of the signal) could representpartially methylated 4q repeats, while the 3.3-kbbands (3–6% of the total signal) should arise frompartial methylation at either 4q or 10q repeats. Also,in both the SacII and MluI digests, we observed bandsof .3.3-kb with sizes predicted for tandem D4Z4 re-peats that are partially methylated at the correspond-ing restriction sites, as was the case in SmaI digests(Figs. 4 and 5, and data not shown).

Analysis of methylation of D4Z4 repeats in FSHDsamples. Because of the central role of the D4Z4copy number of FSHD, we analyzed methylation ofthese repeats in LCLs, uncultured blood, and skele-tal muscle biopsy samples from FSHD patients vsnon-FSHD individuals. The two examined FSHDLCLs, s10 and s5, were mostly methylated at EagI,SmaI, and MluI sites (Fig. 5 and data not shown). InEagI/BlnI double digests, no trace of the hypomethy-lation-specific 2.2-kb band from 4q D4Z4 was visiblein an overexposed X-ray from a blot with 20 mg of theFSHD or normal LCL DNA digests (data notshown). However, in each of these digests, there wasa minor 0.8-kb band containing about 1% of thesignal, which should be derived from the BlnI-sen-sitive 10q copies of D4Z4 that were unmethylated atthe assayed EagI site (Fig. 2B). The copy numbers ofD4Z4 repeats at the disease-conferring 4q35, thenormal 4q35, and at each allelic 10q26 must beconsidered in assessing these results. In the s10 ands5 patients, there was only a single copy of thisrepeat at the deletion-containing 4q35 (16; P. Lunt,unpublished data). The s5 patient had a normal butunspecified number of these repeats at 10q26 and atthe other 4q35. The s10 patient was mosaic with 13

METHYLATION O

or 28 copies of the repeat at the normal 4q35 and 15and 19 copies at the allelic 10q26 regions (K. Ara-hata, unpublished data). Therefore, about 2% of the

total D4Z4 repeats in s10 are at the deletion-con-taining 4q35. EagI/BlnI digests do not give informa-tion about the affected 4q35 in s10 and s5 because a4q35 locus containing a single D4Z4 repeat has noEagI site within 25 kb centromeric to the EagI siteat the end of the probe region (Fig. 2B; (20)).

MluI and SmaI digests of s10 and s5 DNA areinformative about methylation of the single, dele-tion-associated D4Z4 (Figs. 1B and 5B). In the SmaIdigests of s10, s5, and two normal LCLs, ,2% of thesignal was in the 1.6- and 0.2-kb bands (Fig. 5A anddata not shown). In contrast, 71% of the signal in theSmaI-digested ICF LCL DNA lane was in hypom-ethylation-specific 1.6- and 0.2-kb SmaI bands (Fig.5). In MluI digests of s10, s5, or normal LCL DNAs(Fig. 5 and data not shown), about 0.5% of the hy-bridizing fragments were in a hypomethylation-spe-cific 1.2-kb band seen in overexposed X-rays, andthis band could have been derived from either 4q or10q repeats. Therefore, no more hypomethylationwas seen in the FSHD LCLs than in normal LCLs,and the amount of hypomethylation observed wasless than predicted if there were hypomethylation ofthe 4q35 deletion-associated copy of D4Z4.

We also assayed DNA from four FSHD and fournormal blood samples. The D4Z4 copy numbers onthe affected 4q35 region in the FSHD blood sampleswere 4, 5, 5, and 6 for samples FBld1, FBld45,FBld6, and FBld19, respectively, as determined bySouthern blot analysis of EcoRI digests (Figlewicz,unpublished data). The D4Z4 copy number at theunaffected allelic regions is not known for thesesamples but the mean number of D4Z4 repeats pernormal 4q or 10q allele is about 29 (21). Therefore,about 4–7% of the D4Z4 sequences in these samplesis estimated to be in the allele-associated D4Z4 re-peats. EagI/BlnI and MluI digests revealed thatmost of the assayed CpG sites were methylated (Fig.6A and data not shown). In the EagI/BlnI digests,there was generally a major 3.3-kb band, a minor10q-type 0.8-kb band, and a barely detectable 4q-type 2.2-kb band (Fig. 6A). There was no consistentdifference between FSHD and normal blood samplesin the percentage of the signal in 2.2-kb bandsspecific for hypomethylated 4q-type sequences (0.3–2.2% for normal blood and 0.5–1.0% for analogousFSHD samples; Fig. 6A). As described above, gener-ation of the 2.2-kb band in EagI/BlnI digests re-quires cleavage at two EagI sites in a 4q D4Z4 re-

327E FSHD REPEAT

peat. The approximate percentages of 10q D4Z4repeats unmethylated at the single assayed EagIsite (the signal in 0.8-kb band divided by that in the

Page 7

t(tottry(tsr(

vn3

nsSbwmh

N ET

0.8- plus 3.3-kb bands) were 9, 11, 23, and 54 forFSHD samples FBld45, FBld6, FBld1, and FBld19,and 7, 13, 27, and 51 for normal samples NBld44,NBld7, NBld23, and NBld13, respectively.

Lastly, we examined methylation in the D4Z4 re-peats of small DNA samples that we obtained fromskeletal muscle biopsies of three FSHD patients anda patient with a mild neurogenic disease unrelatedto FSHD. The number of copies of the D4Z4 repeaton chromosome 4 homologues is unknown for theseFSHD muscle samples. The distribution of copynumbers in the affected 4q35 of 130 unrelatedFSHD patients had previously been found to beabout 9% of patients with only 1 copy and 72% with3–9 copies (2). The patients in our study displayedthe classical symptoms of FSHD. Two of these cases

FIG. 6. Blood and muscle biopsy DNA from FSHD patientsand unaffected individuals is highly methylated in D4Z4. (A)Blood DNA (about 2 mg) from FSHD patients (FBld) and from

ormal individuals (NBld) was analyzed for methylation at EagIites in EagI/BlnI digests. (B) Hybridization was done on MluI oracII digests of DNA (0.3–1 mg) from FSHD skeletal muscleiopsy samples (FMus) or a skeletal muscle biopsy from a patientith a neurological disorder with secondary, rather than pri-ary, muscle involvement (NMus). ICF LCL DNA served as aypomethylated standard.

328 TSIE

were familial while the third was suspected to befamilial. MluI digests of the FSHD muscle samples(FMus1–3) and the non-FSHD sample (NMus1) re-

smr

vealed that they all had the 1.2-kb band indicative ofhypomethylation of some copies of the FSHD repeatat neighboring MluI sites with 8, 8, 11, and 6% of thesignal in this band for FMus1, FMus2, FMus3, andNMus1, respectively, and 1–2% in the 2.1-kb band(Fig. 6B). Upon digestion of one of the FSHD sam-ples (FMus1) with SacII, no low-molecular-weighthybridizing bands were seen (Fig. 6B). In summary,consistent FSHD-associated hypomethylation wasnot observed in the D4Z4 repeats in these musclesamples.

DISCUSSION

Gene assignment(s) for FSHD have not yet beenmade but about 95% of FSHD patients have a dele-tion of an integral multiple number of copies of thecomplex 3.3-kb repeat unit at 4q35 of one of thechromosome 4 homologues (22). The disease is dom-inant with increasing severity associated with thelarger deletions (1). Monosomy for a large part of thesubtelomeric region does not cause the syndrome(23). There is evidence for an unusual double ho-meobox-containing gene lacking a polyadenylationsignal (24) within the D4Z4 repeat at 4q35 and atthe highly homologous repeat at 10q26 and forFSHD genes proximal to the D4Z4 repeat region in4q35 (20). The simplest explanation of how the lossof many, but not all, D4Z4 repeats in the subtelo-meric region of one 4q arm leads to this dominantphenotype is that it causes inappropriate activationof one or more genes in cis as a result of the loss ofhe postulated heterochromatinization in this region1,4,5,8,16,19,23). We have examined methylation ofhe D4Z4 repeats because of the frequent associationf high levels of vertebrate DNA methylation withranscriptionally silent heterochromatin (10) andhe finding that experimentally induced, tandemepetition of genes in mice often leads to hypermeth-lation in conjunction with transcriptional silencing25,26). Furthermore, local methylation of transcrip-ion control elements has been demonstrated to re-ult in a more highly condensed chromatin configu-ation and concomitant inhibition of gene expression11).

By Southern blot analysis of digests made witharious CpG methylation-sensitive restriction endo-ucleases that recognize 19 sites in the 4q D4Z4.3-kb monomer, we found that normal somatic tis-

AL

ues (brain, lung, liver, heart, and spleen) areostly, although not completely, methylated in this

epeat. In normal somatic cell populations, the high

Page 8

hpStv3rtyoy

F TH

level of methylation of most CpGs in the D4Z4 re-peat is illustrated by our finding that .70% of the

ybridizing fragments in SmaI digests of diverseostnatal somatic tissues are larger than 12 kb inouthern blots hybridized under stringent condi-ions (Fig. 5 and data not shown). This indicates thatery high percentages of the 11 SmaI sites per.3-kb repeat in these tissues are methylated. Theesidual unmethylated sites in the D4Z4 repeats ofissues appear to often be interspersed with meth-lated sites as indicated by the appreciable amountsf the digestion products expected for partial meth-lation at SmaI, EagI, MluI, and SacII sites (Figs. 1,

2, 4, and 5). Analysis of DNA methylation by bisul-fite-based genomic sequencing is necessary to deter-mine whether some CpG sites in normal somaticDNAs are more prone to residual hypomethylationthan others. However, the very high percentage ofC 1 G and CG dinucleotides (73% C 1 G; 325 CpGsper 3.3 kb), the predominant target for vertebrateDNA methylation, in the D4Z4 repeat ensures ahigh m5C content in this repeat when the majority ofits CpGs are methylated.

In contrast to normal postnatal somatic tissuesand normal cell lines (LCLs), sperm from normalindividuals and LCLs from ICF patients werelargely hypomethylated in their D4Z4 repeats.Sperm-like hypomethylation of these subtelomericrepeats and of satellite 2 DNA sequences in juxta-centromeric heterochromatin regions of chromo-somes 1 and 16 (1qh and 16qh) in somatic ICF cells(14,27) can be ascribed to ICF-specific mutations inone of the three known DNA methyltransferasegenes, DNMT3B (13). These mutations result in asmall but reproducible decrease in the ratio ofgenomic m5C to C (0.043 in ICF brain vs 0.0455 innormal brain (14)) just as sperm DNA has a lowerm5C content than various postnatal somatic tissues(15). It remains to be determined whether hypom-ethylation of D4Z4 repeats in sperm might resultfrom a temporary decrease in DNMT3B activity dur-ing spermatogenesis.

We tested whether FSHD cells, which have abnor-mally low numbers of copies of D4Z4 repeats at one4q35 region, are hypomethylated at these repeatsdue to the smaller number of tandem copies. Ourmost definitive results are from one of the two FSHDLCLs, for which we know the D4Z4 copy number atthe disease-conferring, deletion-containing 4q35 re-

METHYLATION O

gion and at the other three D4Z4 subtelomeric re-gions not associated with FSHD (the unaffected al-lelic 4q35 and the two allelic 10q26 regions).

Southern blot analysis of MluI and SmaI digests ofDNA from this LCL and one other FSHD LCL andfrom normal LCLs revealed no detectable D4Z4 hy-pomethylation associated with the FSHD genotype.If there were such hypomethylation, we should havebeen able to detect it although only about 2% of thecopies of D4Z4 were located on the disease-linked4q35 allelic region in the LCL from the FSHD pa-tient for whom the D4Z4 copy number at each 4q35and 10q26 locus was known. The next most informa-tive type of FSHD samples were the blood DNAs.These had the advantage of coming from unculturedcells and so were not subject to culture-associatedchanges in genomic methylation (15). For the FSHDblood samples, the number of copies of D4Z4 on thedisease-associated allelic region was known but themore difficult-to-obtain information about the copynumber (usually more than 20) at the other 4q andthe 10q positions was unavailable. Again, there wasno correlation of methylation status with D4Z4 ge-notype. Lastly, skeletal muscle biopsy samples wereanalyzed. These are the most relevant to the syn-drome because it is this tissue that is affected in thesyndrome, and there are frequently tissue-specificdifferences in DNA methylation. Unfortunately,data on D4Z4 copy numbers were not available forthese patients. Nonetheless, we showed that at thetested D4Z4 CpG sites, there was a high degree ofmethylation and no consistent FSHD-linked in-crease in hypomethylated copies of D4Z4. However,because we have no information about D4Z4 copynumbers in these muscle biopsy samples, we cannotbe certain that the deletion-linked D4Z4 repeats at4q35 are just as highly methylated as those else-where.

Our study demonstrates that, in normal postnatalsomatic tissues, D4Z4 is highly methylated in accordwith the hypothesized heterochromatic nature ofD4Z4 in normal cell populations. Our data indicatethat there is no strong association between hypom-ethylation of these repeats and the FSHD syndrome.This is consistent with the finding that patients withthe ICF syndrome have hypomethylated D4Z4 re-peats but were not reported to display symptoms ofFSHD except for one case of muscle hypotonia ob-served in an ICF patient in the neonatal period andat age 6, that was not of the type associated withFSHD (28). A caveat in the comparison of ICF andFSHD is that ICF is a very rare disease (less than 40

329E FSHD REPEAT

reported patients over 20 years) which quite oftencauses death by age 14 (29) while the onset of symp-toms in FSHD is usually in the second decade (30).

Page 9

N ET

Although it appears unlikely that D4Z4 hypomethy-lation is a frequent characteristic of FSHD, D4Z4methylation might be relevant to maintenance ofchromatin structure at 4q35 and 10q26. DNA meth-ylation can help lock in a constitutively heterochro-matic structure, e.g., in tandemly repeated satelliteDNA sequences, preventing a small percentage ofcopies of these chromosomal sites from becomingdecondensed (14). D4Z4 methylation in conjunctionwith a high number of tandem copies of this repeatmay help stabilize a heterochromatic structure at4q35 in normal somatic cell populations. The conse-quence of a lack of such methylation might be thatusually only a small percentage of D4Z4 repeat re-gions reversibly lose heterochromaticity while a de-crease in copy number beyond a threshold level mayhave much greater consequences to chromatin struc-ture and thereby lead to FSHD.

ACKNOWLEDGMENTS

We are grateful to the late Dr. Kiichii Arahata and Dr. PeterLunt for sharing FSHD cell lines and unpublished informationand Dr. Christoph Grunau for helpful discussions about his DNAmethylation database www.methdb.de. This research was sup-ported in part by NIH Grant CA81506 and FSH Society GrantFSHS-MB-06.

REFERENCES

1. Lunt P. Neuromuscular diseases: From basic mechanisms toclinical management. In Monographs in Clinical Neuro-science (Deymeer F, Ed.). Basel: Karger, pp 44–60, 2000.

2. Upadhyaya M, Maynard J, Rogers MT, Lunt PW, Jardine P,Ravine D, Harper PS. Improved molecular diagnosis of fa-cioscapulohumeral muscular dystrophy (FSHD): Validationof the differential double digestion for FSHD. J Med Genet34:476–479, 1997.

3. Song M, Goto K, Lee JH, Matsumura T, Sahashi K, ArahataK. Facioscapulohumeral muscular dystrophy (FSHD). Neu-roScience News 3:28–33, 2000.

4. Winokur ST, Bengtsson U, Feddersen J, Mathews KD, We-iffenbach B, Bailey H, Markovich RP, Murray JC, WasmuthJJ, Altherr MR, Schutte, BC. The DNA rearrangement as-sociated with facioscapulohumeral muscular dystrophy in-volves a heterochromatin-associated repetitive element: Im-plications for a role of chromatin structure in thepathogenesis of the disease. Chromosome Res 2:225–234,1994.

5. Hewitt JE, Lyle R, Clark LN, Valleley EM, Wright TJ,Wijmenga C, van Deutekom JC, Francis F, Sharpe PT,Hofker M, et al. Analysis of the tandem repeat locus D4Z4associated with facioscapulohumeral muscular dystrophy.

330 TSIE

Hum Molec Genet 3:1287–1295, 1994.

6. Lyle R, Wright TJ, Clark LN, Hewitt JE. The FSHD-associ-ated repeat, D4Z4, is a member of a dispersed family of

homeobox-containing repeats, subsets of which are clusteredon the short arms of the acrocentric chromosomes. Genomics28:389–397, 1995.

7. Winokur ST, Bengtsson U, Vargas JC, Wasmuth JJ, AltherrMR, Weiffenbach B, Jacobsen SJ. The evolutionary distri-bution and structural organization of the homeobox-contain-ing repeat D4Z4 indicates a functional role for the ancestralcopy in the FSHD region. Hum Molec Genet 5:1567–1575,1996.

8. Deidda G, Cacurri S, Piazzo N, Felicetti L. Direct detectionof 4q35 rearrangements implicated in facioscapulohumeralmuscular dystrophy (FSHD). J Med Genet 33:361–365,1996.

9. Cryderman DE, Tang H, Bell C, Gilmour DS, Wallrath LL.Heterochromatic silencing of Drosophila heat shock genesacts at the level of promoter potentiation. Nucleic Acids Res27:3364–3370, 1999.

10. Gama-Sosa MA, Wang RY-H, Kuo KC, Gehrke CW, EhrlichM. The 5-methylcytosine content of highly repeated se-quences in human DNA. Nucleic Acids Res 11:3087–3095,1983.

11. Jones PL, Veenstra GJ, Wade PA, Vermaak D, Kass SU,Landsberger N, Strouboulis J, Wolffe AP. Methylated DNAand MeCP2 recruit histone deacetylase to repress transcrip-tion. Nature Genet 19:187–191, 1998.

12. Kondo T, Bobek MP, Kuick R, Lamb B, Zhu X, Narayan A,Bourc’his D, Viegas-Pequignot E, Ehrlich M, Hanash SM.Whole-genome methylation scan in ICF syndrome: Hypo-methylation of non-satellite DNA repeats D4Z4 and NBL2.Hum Molec Genet 9:597–604, 2000.

13. Xu G, Bestor TH, Bourc’his D, Hsieh C, Tommerup N,Hulten M, Qu S, Russo JJ, Viegas-Pequignot E. Chromo-some instability and immunodeficiency syndrome caused bymutations in a DNA methyltransferase gene. Nature 402:187–191, 1999.

14. Tuck-Muller CM, Narayan A, Tsien F, Smeets D, Sawyer J,Fiala ES, Sohn O, Ehrlich M. DNA hypomethylation andunusual chromosome instability in cell lines from ICF syn-drome patients. Cytogen Cell Gen 89:121–128, 2000.

15. Ehrlich M, Gama-Sosa M, Huang L-H, Midgett RM, KuoKC, McCune RA, Gehrke C. Amount and distribution of5-methylcytosine in human DNA from different types oftissues or cells. Nucleic Acids Res 10:2709–2721, 1982.

16. Lee JH, Goto K, Matsuda C, Arahata K. Characterization ofa tandemly repeated 3.3-kb KpnI unit in the facioscapulo-humeral muscular dystrophy (FSHD) gene region on chro-mosome 4q35. Muscle Nerve 2:S6–S13, 1995.

17. Ji W, Hernandez R, Zhang X-Y, Qu G, Frady A, Varela M,Ehrlich M. DNA demethylation and pericentromeric rear-rangements of chromosome 1. Mutat Res 379:33–41, 1997.

18. Zhang XY, Loflin PT, Gehrke CW, Andrews PA, Ehrlich M.Hypermethylation of human DNA sequences in embryonalcarcinoma cells and somatic tissues but not in sperm. Nu-cleic Acids Res 15:9429–9449, 1987.

19. Lemmers RJ, van der Maarel SM, van Deutekom JC, van

AL

der Wielen MJ, Deidda G, Dauwerse HG, Hewitt J, HofkerM, Bakker E, Padberg GW, Frants RR. Inter- and intra-chromosomal sub-telomeric rearrangements on 4q35: Impli-cations for facioscapulohumeral muscular dystrophy

Page 10

F TH

(FSHD) aetiology and diagnosis. Hum Molec Genet 7:1207–1214, 1998.

20. van Geel M, Heather LJ, Lyle R, Hewitt JE, Frants RR, deJong PJ. The FSHD region on human chromosome 4q35contains potential coding regions among pseudogenes and ahigh density of repeat elements. Genomics 61:55–65, 1999.

21. Wijmenga C, van Deutekom JC, Hewitt JE, Padberg GW,van Ommen GJ, Hofker MH, Frants RR. Pulsed-field gelelectrophoresis of the D4F104S1 locus reveals the size andthe parental origin of the facioscapulohumeral musculardystrophy (FSHD)-associated deletions. Genomics 19:21–26,1994.

22. Orrell RW, Tawil R, Forrester J, Kissel JT, Mendell JR,Figlewicz DA. Definitive molecular diagnosis of facioscapu-lohumeral dystrophy. Neurology 52:1822–1826, 1999.

23. Tupler R, Berardinelli A, Barbierato L, Frants R, Hewitt JE,Lanzi G, Maraschio P, Tiepolo L. Monosomy of distal 4q doesnot cause facioscapulohumeral muscular dystrophy. J MedGenet 33:366–370, 1996.

24. Gabriels J, Beckers MC, Ding H, De Vriese A, Plaisance S,van der Maarel SM, Padberg GW, Frants RR, Hewitt JE,Collen D, Belayew A. Nucleotide sequence of the partiallydeleted D4Z4 locus in a patient with FSHD identifies a

METHYLATION O

putative gene within each 3.3 kb element. Gene 236:25–32,1999.

25. Lau S, Jardine K, McBurney MW. DNA methylation pattern

of a tandemly repeated LacZ transgene indicates that mostcopies are silent. Dev Dyn 215:126–138, 1999.

26. Sutherland HG, Kearns M, Morgan HD, Headley AP, MorrisC, Martin DI, Whitelaw E. Reactivation of heritably silencedgene expression in mice. Mamm Genome 11:347–355, 2000.

27. Jeanpierre M, Turleau C, Aurias A, Prieur M, Ledeist F,Fischer A, Viegas-Pequignot E. An embryonic-like methyl-ation pattern of classical satellite DNA is observed in ICFsyndrome. Hum Molec Genet 2:731–735, 1993.

28. Schuffenhauer S, Bartsch O, Stumm M, Buchholz T, Petro-poulou T, Kraft S, Belohradsky B, Meitinger T, Wegner R.DNA, FISH and complementation studies in ICF syndrome:DNA hypomethylation of repetitive and single copy loci andevidence for a trans acting factor. Hum Genet 96:562–571,1995.

29. Brown DC, Grace E, Summer AT, Edmunds AT, Ellis PM.ICF syndrome (immunodeficiency, centromeric instabilityand facial anomalies): Investigation of heterochromatin ab-normalities and review of clinical outcome. Hum Genet 96:411–416, 1995.

30. Lunt PW, Jardine PE, Koch MC, Maynard J, Osborn M,Williams M, Harper PS, Upadhyaya M. Correlation betweenfragment size at D4F104S1 and age at onset or at wheel-

331E FSHD REPEAT

chair use, with a possible generational effect, accounts formuch phenotypic variation in 4q35-facioscapulohumeralmuscular dystrophy (FSHD). Hum Molec Genet 4:951–958,1995.