A cytosine DNA methyltransferase containing a chromodomain,
Zea methyltransferase2
(
Zmet2
), was cloned frommaize. The sequence of
ZMET2
is similar to that of the Arabidopsis chromomethylases
CMT1
and
CMT3
, with C-termi-nal motifs characteristic of eukaryotic and prokaryotic DNA methyltransferases. We used a reverse genetics approachto determine the function of the
Zmet2
gene. Plants homozygous for a
Mutator
transposable element insertion intomotif IX had a 13% reduction in methylated cytosines. DNA gel blot analysis of these plants with methylation-sensitiverestriction enzymes and bisulfite sequencing of a 180-bp knob sequence showed reduced methylation only at CpNpGsites. No reductions in methylation were observed at CpG or asymmetric sites in heterozygous or homozygous mutantplants. Our research shows that chromomethylase
Zmet2
is required for in vivo methylation of CpNpG sequences.
INTRODUCTION
Cytosine methylation is a DNA modification that is corre-lated with gene expression in eukaryotes. In plants, hyper-methylation is associated with inactivation of transposableelements (Chandler and Walbot, 1986; Schwartz and Dennis,1986; Banks et al., 1988), instances of imprinting and para-mutation (Kermicle, 1996; Walker, 1998), and transgene si-lencing (Linn et al., 1990; Ingelbrecht et al., 1994). Control ofgene expression mediated by DNA methylation also mayplay a role in plant development (Kakutani et al., 1995;Finnegan et al., 1996).
DNA methylation patterns are established and maintainedby DNA methyltransferases, which catalyze the transfer ofthe methyl group from
S
-adenosylmethionine (SAM) to theC-5 position in the pyrimidine ring of cytosine. In plants,5-methylcytosine residues are found predominantly at sym-metric CpG and CpNpG sequences (Gruenbaum et al.,1981). Methylated cytosines are found at a lower frequencyat asymmetric positions such as CpTpT and CpApT. At leastthree types of methyltransferase activity are likely to exist inplants. De novo methyltransferase activity is required to es-tablish methylation at unmethylated sites and to propagatemethylation at asymmetric sites. Separate CpG and CpNpGmethyltransferase activities maintain symmetric methylationpatterns during the process of DNA replication and cell divi-sion. Although the CpG and CpNpG methyltransferase ac-tivities likely are distinct, evidence in pea suggests thepossibility that these activities may be conferred by prod-
ucts of a single gene (Pradhan and Adams, 1995; Pradhanet al., 1998). CpNpG methylation is common in plants butinfrequent or absent in mammals.
The focus of this study was the analysis of a putativemaize methyltransferase gene homologous with the CMTfamily of chromomethylase genes in Arabidopsis. Chro-momethylases were first identified in a database search forgenes containing chromodomains (Henikoff and Comai,1998; Rose et al., 1998). Chromodomains have been foundin several proteins involved in the chromatin-based regula-tion of gene expression and may be critical for the targetingof these proteins within the nucleus (Cavalli and Paro, 1998).Chromomethylases contain conserved methyltransferasedomains at the C-terminal end, whereas the N-terminal por-tion of chromomethylases differs substantially from that ofother methyltransferases. Chromomethylases lack
�
750amino acids in their N-terminal domain relative to the
Dnmt1
class of methyltransferases, which is exemplified by theMET1 (Finnegan and Dennis 1993) and
Zmet1
genes inplants. Chromomethylases are similar in size to the
Dnmt3
class of de novo
methyltransferases, which is representedby the
Zmet3
and
Drm2
genes in plants (Cao et al., 2000),but they differ substantially on the basis of the organizationof the conserved methyltransferase motifs and the structureof the N-terminal portion of the gene. Chromomethylaseshave not been identified in the genomic sequences of anyorganism other than plants. To date, no function has beenassigned to the chromomethylase class of methyltrans-ferases. The
CMT1
chromomethylase of Arabidopsis hasbeen deemed nonessential because several Arabidopsisecotypes contain genes with a retroelement insertion that
1
To whom correspondence should be addressed. E-mail [email protected]; fax 608-262-5217.
1920 The Plant Cell
Figure 1. Maize Zmet2 Encodes a Chromomethylase.
(A) The conserved methyltransferase motifs of the ZMET2 and ZMET5 inferred amino acid sequences are aligned with the Arabidopsis chromo-methylases CMT1, CMT2, and CMT3. Black shading indicates identical residues, and gray shading indicates similarity. Dashes in the sequences rep-resent gaps introduced by CLUSTAL W to optimize the alignments. Alignments were processed by BOXSHADE. The locations of the six conservedmethylase motifs are indicated above the sequences. The chromodomain is located upstream of and adjacent to motif IV. The location of the Mu in-sertion in the zmet2-m1::Mu allele is indicated by an arrowhead above the sequences.(B) Relationships of maize and Arabidopsis chromomethylases. The aligned conserved methyltransferase motifs of ZMET2, ZMET5, CMT1,CMT2, and CMT3 (from the beginning of motif I to the end of motif X) were analyzed using PHYLIP. The resulting tree is shown with bootstrapvalues indicated at the nodes.
Chromomethylases Maintain CNG Methylation 1921
disrupts the coding region or a frameshift mutation that re-sults in truncated proteins (Henikoff and Comai, 1998). Noobvious differences in DNA methylation have been corre-lated with these alleles.
We have characterized one member of a small gene fam-ily of chromomethylases in maize,
Zea methyltransferase2
(
Zmet2
). In this study, we show that this methyltransferase isrequired for the methylation of CpNpG sequences.
RESULTS
Zmet2
Is a Chromomethylase
Zmet2
was discovered in an expressed sequence tagsearch of the Pioneer Hi-Bred International databases usingconserved methyltransferase domains as the query se-quence. The full-length cDNA sequence of
Zmet2
was ob-tained by 3
�
and 5
�
rapid amplification of cDNA ends, andthe sequence was analyzed against the GenBank andSWISS-PROT databases. The full-length genomic sequencewas obtained by screening a genomic library.
Screening of the genomic library also resulted in the recov-ery of a class of clones with significant similarity to
Zmet2
. Ge-nomic sequences of the homologous clones were obtained,and the cDNA sequences were confirmed by sequencingrapid amplification of cDNA ends products. This second geneis designated
Zmet5
. ZMET2 and ZMET5 belong to a class ofDNA methyltransferases termed chromomethylases (Henikoffand Comai, 1998). The
Zmet2
and
Zmet5
nucleotide se-quences are 90% identical over the coding regions.
Amino acid sequence alignments of ZMET2 and ZMET5with the Arabidopsis chromomethylases CMT1, CMT2, andCMT3 (Figure 1A) showed conservation in motifs I, IV, VI,VIII, IX, and X (motifs defined by Posfai et al., 1989; Kumaret al., 1994). Phylogenetic analysis indicated that the ZMET2and ZMET5 proteins are more closely related to CMT1 andCMT3 (GenBank accession numbers provided in Methods)than to CMT2 (Figure 1B). Alignments of ZMET2 and CMT1revealed 44% amino acid identity and 57% similarity. CMT1and ZMET2 have 87% amino acid similarity across the sixconserved functional domains.
The N-terminal domain of ZMET2
is smaller than thosefound in the
dnmt1
class of maintenance methyltrans-ferases, but it does contain putative nuclear localization sig-nals, as defined by Raikhel (1992). A chromodomain ispresent between motifs I and IV, and the amino acid se-
quence and position are conserved between ZMET2 andCMT1. The inferred ZMET2 protein, using the first predictedtranslation start site located within a consensus Kozak se-quence (Kozak, 1991), is 912 amino acids in length with apredicted mass of 101 kD.
The ZMET5 and CMT protein sequences were tested forthe presence of recognizable domains by using both thePFAM and SMART protein prediction World Wide Web serv-ers (Schultz et al., 2000). In addition to containing a chromo-domain and a conserved methyltransferase domain, allchromomethylases contain a bromo adjacent homology(BAH) domain. The location of this domain is indicated inFigure 1C. BAH domains have been implicated in linkingDNA methylation, replication, and transcriptional regulationin mammals (Callebaut et al., 1999). ZMET1 and MET1 bothcontain two BAH domains in the N-terminal regulatory re-gion. This may indicate common mechanisms controllingthese two classes of methyltransferases or targeting them inthe nucleus. This finding also supports the phylogenetic evi-dence suggesting that the chromomethylases and MET1-type enzymes have a common ancestor (Cao et al., 2000).
Mutant Analysis Reveals That
Zmet2
Is Required For CpNpG Methylation
A reverse genetics approach was used to determine thefunction of
Zmet2
. An F2 family segregating for a
Mutator
transposable element (
Mu
) insertion in
Zmet2
was identifiedfrom Pioneer Hi-Bred International’s TUSC (Trait Utility Sys-tem for Corn) populations by using a polymerase chain re-action primer for
Mu
and a gene-specific primer for
Zmet2
.This allele is called
zmet2-m1
::
Mu
. The
Mu
element is in-serted into exon 18, which encodes motif IX. To determinethe likely effect of the
Mu
insertion, the aberrant transcriptresulting from this insertion event was sequenced. The ab-errant transcript contains a stop codon after amino acid 833(Figure 1A). The resulting protein lacks motif X, which is re-quired for SAM binding (Cheng et al., 1993), and is expectedto lack enzymatic function.
Twelve individual plants from an F4-derived F5 family,composed of three wild-type plants, seven plants heterozy-gous for the
zmet2-m1
::
Mu
allele, and five plants homozy-gous for the
zmet2-m1
::
Mu
allele, were analyzed by HPLCto assess the effect of the mutation on global methylationlevels. A 12.6% decrease in 5-methylcytosine was observedin plants homozygous for
zmet2-m1
::
Mu
relative to siblingshomozygous for wild-type
Zmet2
(Table 1). Heterozygous
Figure 1.
(continued).
(C)
Diagrams of methyltransferase proteins. The chromomethylases of maize and Arabidopsis are shown with Zmet1 and Zmet3. The location ofthe chromodomains (CD), the conserved methyltransferase motifs (I, IV, and VI to X), the BAH domains, and the ubiquitin-associated domains(UBA) are indicated by different shading.
1922 The Plant Cell
plants had significantly (
�
�
0.001) less methylation than didthe homozygous normal plants. The reduction in methyla-tion of the heterozygous class relative to the homozygousnormal class was
�
27% of the reduction in methylation ob-served in the homozygous
zmet2-m1
::
Mu
class.Restriction enzyme analysis of the same DNA samples
was used to test for site-specific changes in DNA methyla-tion. DNA gel blots of DNA cut with methylation-sensitive re-striction enzymes and probed with repetitive sequencesrevealed significant reductions in cytosine methylation at
m
CpCpG sites (Figure 2) and
m
CpA/TpG sites (data notshown). Plants homozygous for the
zmet2-m1
::
Mu
allelehad the largest reduction in methylation, whereas plantsheterozygous for
zmet2-m1
::
Mu
were intermediate in theirdigestion patterns. No changes were observed at
m
CpGsites, indicated by unchanged HpaII (Figure 2) and HhaI(data not shown) restriction profiles. Genomic bisulfite se-quencing of the 180-bp knob sequence (Figure 3) confirmedthat only CpNpG methylation was reduced in the
zmet2-m1
::
Mu
mutant. The 180-bp knob sequence was meth-ylated at both symmetric and asymmetric cytosines. Thebisulfite sequencing analysis revealed that asymmetricmethylation was not reduced by this mutation.
Our analysis of the
zmet2-m1
::
Mu
allele indicates that re-ductions in methylation are restricted to CpNpG sites. TheHPLC data indicate an overall reduction in global methyla-tion of 12.6%. Calculations based on work by Gruenbaumet al. (1981) indicate that CpNpG methylation in wheat seed-lings accounts for
�
30 to 40% of total cytosine methylation.The 12.6% reduction in total cytosine methylation in the ho-mozygous
zmet2-m1
::
Mu
plants explains a 30 to 50% re-duction in methylation at CpNpG sites.
Complete reduction in methylation at a given methylatedsite was observed rarely, but nearly every site analyzedshowed some reduction. This indicates that the partial reduc-tion in CpNpG methylation is caused by random reduction ofmethylation at most or all CpNpG sites, rather than completereduction at some sites with no reduction at others. Our anal-ysis did not allow us to determine if there are sectors of meth-ylation types within the leaf tissue analyzed. That is, we donot know at present whether DNA from certain cells is com-pletely devoid of CpNpG methylation whereas others havenormal amounts of methylation, or conversely, if CpNpGmethylation is lost randomly at sites along a DNA strand.
There are several potential explanations for an incompletereduction in CpNpG methylation in homozygous
zmet2-m1
::
Mu
plants. The most likely explanation is that
Zmet5
, ahomolog of
Zmet2
, has at least partial overlapping functionand expression. This notion is supported by overlapping ex-pression profiles of
Zmet2
and
Zmet5
(data not shown) and ahigh degree of conservation among the proteins. A secondpossibility is that an enzyme unrelated to
Zmet2
is capable ofmaintaining CpNpG methylation at a reduced frequency. Thecloned MET1 gene from pea displayed both CpG and CpA/TpG methyltransferase activities in vitro (Pradhan and Adams,1995; Pradhan et al., 1998). No CpCpG activity was found forthis protein, however. In addition, MET1 antisense plantsshowed some reduction at CpNpG sites (Finnegan et al.,1996), supporting the possibility that this class of methyltrans-ferases may have CpNpG activity in vivo. Finally, it is possiblethat the
zmet2-m1
::
Mu
insertion does not reduce the activityof the enzyme completely, although our sequence analysis ofthe altered transcript indicates that the protein will lack a criti-cal domain and provides little support for this possibility.
Restriction analysis and HPLC quantitation support an in-termediate level of methylation reduction in heterozygous
zmet2-m1
::
Mu
plants relative to homozygous normal plants(Tables 1 and 2). This intermediate level of methylation in theheterozygote class was unexpected for the
zmet2-m1
::
Mu
allele, which should have lost enzymatic function. One pos-sible explanation for the intermediate methylation of the het-erozygote is that the amount of CpNpG methylation in thegenome is stoichiometrically determined by the amount ofZMET2 protein. To test whether the amount of ZMET2 stoichi-ometrically determines the level of methylation in cells,hypoploids, euploids, and hyperploids, with one, two, andthree copies of chromosome 10L, respectively, were gener-ated using a stock with a maize A chromosome arm translo-cated to a B chromosome fragment containing the Bchromosome centromere.
Zmet2
maps distal to R on chro-mosome 10L (data not shown) and should be contained onthe translocation chromosomes. Methylation levels of allthree classes were not significantly different from each other(data not shown). Assuming that altered copy number ofchromosome arms containing
Zmet2
translates into varyinglevels of ZMET2
enzyme, this result indicates, but does notprove conclusively, that enzyme amount likely does not limitmethylation levels.
Table 1.
Cytosine Methylation Levels in an F4-Derived F5 Family Segregating for
zmet2-m1
::
Mu
a
Genotype Plants Total 5-
me
Cytosine (%)
b
Wild-Type Levels (%) Decrease (%)
Wild type 3 24.80 a 100.0 0.0Heterozygous
zmet2-m1
::
Mu
7 23.96 b 96.6 3.4Homozygous
zmet2-m1
::
Mu
5 21.68 c 87.4 12.6
a
The 5-methylcytosine content of DNA extracted from tissue of immature leaves was determined by reverse phase HPLC. Percentages of 5
m
Ccontent [5
m
C/(5
m
C
�
C)] were calculated from concentrations determined from integration of peak areas and comparisons with known stan-dards.
b
Letters following means indicate groupings that were significantly different (� � 0.001).
Chromomethylases Maintain CNG Methylation 1923
A second possible explanation for a partial reduction ofmethylation in heterozygous zmet2-m1::Mu plants is thatthe mutant protein acts as a competitive inhibitor of the nor-mal ZMET2 protein, and possibly the ZMET5 protein, pro-ducing a dominant negative effect. Because methyltrans-ferases are thought to function enzymatically as monomers,a dominant negative effect would have to be caused by tar-get site competition rather than production of nonfunctionalenzyme complexes. The target site could be either a proteincomplex associated with DNA or a direct association withthe DNA itself. zmet2-m1::Mu is likely to produce a proteinthat contains the domains involved in targeting and DNAbinding, including the chromodomain and the BAH domain,but it lacks the SAM binding domain, thereby abolishing en-zyme function. Under the dominant negative hypothesis, re-duction in methylation would result from the localization ofthe mutant protein to hemimethylated sites without complet-ing the methylation reaction. The mutant enzyme would re-main associated with the target site, precluding an interactionof the hemimethylated site with a functional methyltrans-ferase. This would result in a reduction in methylation even ifsome amount of functional protein were present in the cell.
Inheritance of Methylation Status
To test the stability of methylation levels over generations,homozygous zmet2-m1::Mu mutant plants derived from het-erozygous, F3-derived plants were compared with homozy-gous zmet2-m1::Mu mutant plants derived from severalgenerations of self-pollination of homozygous mutant plants.The percentage of methylated cytosines was consistentamong all homozygous mutant progeny regardless of pedi-gree and did not decrease upon self-pollination of homozy-gous mutants (data not shown). Unfortunately, becausethere is a methylation reduction in heterozygotes, it is im-possible to assess a “first generation” effect because thenature of the reverse genetics approach used does not pro-duce sufficient DNA or sibling controls for a valid analysis ofthe founder heterozygous mutant plant. Therefore, we can-not conclude unambiguously that reduction in methylationdoes not increase in the generation subsequent to the pri-mary mutation event. It is only possible to conclude that thelevel of methylation is relatively consistent thereafter.
To determine the extent of remethylation when the zmet2-m1::Mu mutation is removed by segregation, we producedbackcross plants with a homozygous zmet2-m1::Mu plantas a grandparent. The inbred line, Mo17, was crossed to ahomozygous zmet2-m1::Mu plant, and the resulting F1 plantwas then backcrossed to the Mo17 parent line. Restrictionenzyme analysis of backcross progeny indicated that all in-dividuals without the Mu insertion displayed substantialremethylation of repetitive centromeric sequences. Further-more, analysis of the same DNA samples by HPLC indicatedthat genomic levels of cytosine methylation in homozygousnormal backcross progeny were less than those of Mo17,
Figure 2. Gel Blot Analysis of Repetitive DNA Methylation Patterns.
Decreased methylation is observed in mutant plants (� �) relative tononmutant plants (� �) digested with MspI, which is sensitive tomethylation at mCpCpG sequences. No changes in methylation pat-terns at mCpG sites are observed in mutant plants, as indicated bythe lack of digestion with HpaII. Plants heterozygous for zmet2-m1::Mu (� �) also show decreases at mCpCpG sites. DNA gel blotswere hybridized with probes for repetitive DNA: the 9-kb 26s-5.8s-17s ribosomal repeat (A), the 5s ribosomal repeat (B), and the cen-tromeric repeat pSau3A9 (C).
1924 The Plant Cell
the nonmutant parent, but significantly greater than those ofheterozygous backcross siblings (Table 2). These data areconsistent with the hypothesis that methylation levels arepartially, but not completely, restored in the first generationof homozygote wild-type progeny obtained from a homozy-gous mutant parent.
The observation that remethylation occurs in normal prog-eny of zmet2-m1::Mu plants indicates either that ZMET2 hasin vivo, de novo activity and is responsible for the establish-ment of CpNpG methylation patterns or that a separate denovo methyltransferase functions only early in developmentand that Zmet2 is responsible for maintaining these pat-terns. Our current data do not allow us to determine whichof these possibilities is correct.
DISCUSSION
Our data indicate that chromomethylases function in vivo tomaintain CpNpG symmetrical methylation patterns. This isconsistent with two observations. First, CpNpG is found inmost angiosperm and gymnosperm genomes but is limited
in frequency in organisms other than plants. Second, chro-momethylases have been found in plant species ranging frommonocots to dicots but have not been found in the genomesof any other organisms. Therefore, chromomethylases, whichapparently evolved after the divergence of plants from otherorganisms, offer plant genomes a second means to propa-gate methylated cytosines. The conservation of chromometh-ylase function across species as diverse as Arabidopsis andmaize suggests that these genes provide a function that of-fers an evolutionary advantage to the organism.
The conserved domains of chromomethylases may pro-vide insight into the purpose of these enzymes. In additionto the conserved methyltransferase domains, chromometh-ylases contain a chromodomain and a BAH domain. Chro-modomains are found in several proteins involved inchromatin-level repression of transcription (Cavalli andParo, 1998). For two of these proteins, Polycomb and HP1,the chromodomain is critical for the proper targeting. Thechromodomains of the Drosophila melanogaster dosagecompensation proteins MOF and MSL-3 are involved inbinding to noncoding RNA molecules (Akhtar et al., 2000).The interaction of chromodomains with RNA may be themechanism for targeting proteins containing chromodomains
Figure 3. The DNA Methylation Patterns of Both Strands of the 180-bp Knob Repeat Were Determined by Direct Genomic Bisulfite Sequencing.
CpG, CpNpG, and asymmetric methylation all were detected in the knob sequence of plants wild type for Zmet2. CpG dinucleotides are indi-cated by ovals, and CpNpG trinucleotides are indicated by rectangles. The symbols above and below the alignment indicate the amount of DNAmethylation observed in wild-type and homozygous zmet2-m1::Mu plants. When only one symbol is shown, the methylation was the same inwild-type and zmet2-m1::Mu plants. When the methylation status differed, two symbols are shown, with the top symbol showing the methylationstatus of that base in zmet2-m1::Mu plants and the bottom symbol showing the methylation status of that base in wild-type siblings. The onlydifferences in methylation were found at CpNpG sequences. No changes in CpG or asymmetric methylation were observed.
Chromomethylases Maintain CNG Methylation 1925
to specific regions of chromosomes. In plants, some RNAmolecules have been shown to induce DNA methylation ofhomologous sequences (Wassenegger et al., 1994; Joneset al., 1999; Mette et al., 2000; Wassenegger, 2000). Chro-momethylases may be the enzymes responsible for RNA-directed DNA methylation in plants. Recently, Swi6, anHP1 homolog, was shown to interact directly with meth-ylated Lys-9 of histone H3 (Rea et al., 2000; Nakayama et al.,2001). This finding suggests the alternative possibility thatchromomethylases are targeted by their chromodomain toheterochromatic regions marked by H3 Lys-9 methylation.
A BAH domain also is found in the N-terminal portion ofchromomethylases. It is interesting that all Dnmt1/Met1 DNAmethyltransferases and chromomethylases contain BAH do-mains. The Dnmt1/Met1 proteins contain two BAH domains,whereas chromomethylases contain only one BAH domain.This common feature may suggest a similar function or tar-geting for these two groups of methyltransferases. One func-tion proposed for the BAH domain is to link DNA methylationto replication. These two classes of methyltransferases bothmay be involved in maintaining symmetric methylation pat-terns, with the Dnmt1/Met1 class acting on hemimethylatedCpG sites and the chromomethylases methylating hemimeth-ylated CpNpG sites soon after replication.
The distribution of CpNpG methylation in the genome alsomay provide insight into the function of chromomethylases.In general, methylation is found at lower levels in expressedgenes and single-copy sequences than in repetitive se-quences. However, CpG methylation is found in some re-gions of promoters and genes, depending on the gene andthe stage of development. CpNpG methylation is almostnever found in genic regions and appears to be restricted torepetitive DNA, with the highest abundance in heterochro-matic regions. Therefore, the purpose of CpNpG methyla-
tion may be to reinforce the heterochromatic state and assistin the silencing of transposable and retrotransposable ele-ments. The function of the chromodomain may be to targetCpNpG methylation to these repeats. The increase in methyla-tion across these repeats likely accelerates primary sequencedegeneration attributable to mutation via spontaneous deami-nation of methylated cytosines. The addition of CpNpG methyl-ation is expected to nearly double the number of methylatedcytosines in regions susceptible to this type of methylation,thereby nearly doubling the frequency of C-to-T transition mu-tations caused by deamination of cytosines.
In summary, we have determined that the maize geneZmet2 is required for the methylation of CpCpG and CpA/TpGsites in vivo. Plants containing a Mu transposable element in-sertion that disrupts Zmet2 have an �13% reduction in meth-ylation. The observations that the chromomethylase family ofgenes is unique to plants, that chromomethylase genes are re-quired for CpNpG methylation, and that CpNpG methylation ismuch more abundant in plants than in other organisms allsupport the notion that the chromomethylase family of genesevolved to provide and maintain CpNpG methylation in plants.Future research will determine whether the abundance of Cp-NpG methylation in plants plays a unique role in the control ofgene expression and genome stability or if this family of meth-yltransferases evolved as a redundant mechanism to CpGmethylation and other silencing pathways.
METHODS
Cloning and Sequencing of zmet2
A partial cDNA clone (CGET064) from an immature tassel cDNA librarywas identified in the Pioneer Hi-Bred International expressed se-quence tag collection. The sequence of this clone, which representsthe 3� end of the transcript, was used to design forward and reverseprimers for 5� and 3� rapid amplification of cDNA ends (RACE). RACEwas conducted using the Marathon cDNA Amplification Kit (Clontech,Palo Alto, CA) according to the manufacturer’s protocols on cDNAprepared from Mo17 10-day-old seedling mRNA. Total RNA was ex-tracted using Trizol (Gibco BRL) according to the manufacturer’s pro-tocol, and mRNA was isolated using oligo(dT)-cellulose columns(Pharmacia). RACE products were isolated, and ends were sequencedusing Marathon primers and gene-specific primers (zmet2-RT, 5�-CTACAACATCATAGTTGGGCAGAGG-3�; and zmet2-5F, 5�-TAAAGG-GCGTGAGGGTTGGA-3�). The remaining sequence was obtained frompolymerase chain reaction (PCR) products by primer walking (zmet2-1R,5�-CCAGCTCAGCTCAGATCTGTCATCCTTT-3�; zmet2-1F, 5�-TGG-TTGCTATGGTCTGCCACAGTTCAG-3�; and zmet2-3R, 5�-TCTCTA-ATTTTCTGCGGGCAG-3�; zmet2-8R, 5�-GCAATCAAGCACATT-GTCGTTCTTTTCCTC-3�; zmet2-9R, 5�-TTCTTTGCGGCAGTGCTG-CG-3�; and zmet2-9F, 5�-GAAGAAGAGGGTGGGGAGAAGGAACG-3�).Two sequencing passes were made on the cDNA ends, and foursequencing passes were made on the intervening regions. PCRproducts were sequenced using Big Dye terminator cycle sequencingon an ABI sequencer (Perkin-Elmer Applied Biosystems) at the Univer-sity of Wisconsin Biotechnology Center Sequencing Facility.
Table 2. Remethylation of Backcross Progeny from a Heterozygous zmet2-m1::Mu Plant Backcrossed to the Nonmutant Mo17 Parental Linea
a Mean cytosine methylation levels are given for each family type.Percentages of 5mC content [5mC/(5mC � C)] were calculated fromconcentrations determined from integration of peak areas and com-parisons with known standards.b n, number of individual plants tested. Two duplicates of each sam-ple were analyzed by HPLC.c Letters following means indicate groupings that were significantlydifferent (� � 0.01).
1926 The Plant Cell
Identification of the zmet2-m1::Mu Mutant Allele
A mutant containing a Mu transposable element insertion was iden-tified in a collection of indexed mutagenized F2 families derived fromseveral Mu active stocks (Bensen et al., 1995). The mutant was iden-tified using a Mu-specific primer (5�-AGAGAAGCCAACGCCA(A/T)CGCCTC(C/T)ATTTCGTC-3�) and zmet2 gene-specific primers(zmet2-1F, 5�-TGGTTGCTATGGTCTGCCACAGTTCAG-3�; and zmet2-1R, 5�-CCAGCTCAGCTCAGATCTGTCATCCTTT-3�). Because theMutator population is quite variable, heterozygous zmet2-m1::Mu F2seed was advanced by selfing to produce the F4-derived F5 segre-gating family used in this analysis.
DNA Extraction and Gel Blot Analysis for Genotyping and Methylation Analysis
The fifth to seventh immature leaf tips were collected from 15 plantsof the F4-derived F5 segregating family and frozen immediately indry ice. Tissue was ground in liquid nitrogen, and DNA was extractedas described (Saghai Maroof et al., 1984). The genotype at Zmet2was determined by digesting DNA (10 �g) with BamHI and EcoRI,which cut on each side of the Mu insertion. The digested DNA waselectrophoresed through a 0.8% agarose–0.5 � Tris-borate-EDTAgel. Gels were treated with 0.25 N HCl for 15 min, denatured in 0.2 NNaOH and 0.6 M NaCl for 30 min, and then neutralized in 0.5 M Trisand 1.5 M NaCl for 30 min. DNA was transferred to an Immobilon ny-lon membrane (Millipore, Bedford, MA) with 5 � SSC (1 � SSC is0.15 M NaCl and 0.015 M sodium citrate). Blots were dried at 80Cfor 1.5 hr. Prehybridization was performed in 5 � SSC, 50 mM Tris,pH 8.0, 0.2% SDS, 10 mM EDTA, 1 � Denhardt’s solution (1 � Den-hardt’s solution is 0.02% Ficoll, 0.02% polyvinylpyrrolidone, and0.02% BSA), and 0.1 mg/mL single-stranded sheared herring DNAovernight (8 to16 hr) at 65C.
Hybridization conditions were similar to prehybridization condi-tions except for the addition of 5% dextran sulfate to the hybridiza-tion solution. Probes (25 to 50 ng) (clone CGET064 for genotyping)were labeled with 32P-dCTP (50 �Ci) by using random priming. Afterovernight hybridization at 65C, blots were washed two times (0.15 �SSC and 0.1% SDS) for 30 to 45 min at 65C. Hybridized blots wereexposed to Kodak BioMax film. DNA gel blot analysis with methyla-tion-sensitive restriction enzymes was conducted on a subset of theplants using the same protocols used for genotyping except that5 �g of DNA was digested. Enzymes included in the study were thedifferentially methylation-sensitive isoschizomers HpaII-MspI andEcoRII-BstNI as well as other methylation-sensitive enzymes(BamHI, BglII, HhaI, PstI, PvuII, SacI, and ScrFI). Blots were hybrid-ized with probes for repetitive-sequence regions of the maize ge-nome, a 9-kb clone for the maize 26s-5.8s-17s repeat (McMullen etal., 1991), the 5s ribosomal subunit clone (Mascia et al., 1981), andcentromere probe pSau3A9 (Jiang et al., 1996).
HPLC Analysis
HPLC was conducted according to protocols described previously(Gehrke et al., 1984). Duplicate preparations for each of 15 plantswere analyzed. Five homozygous zmet2-m1::Mu plants, seven het-erozygous zmet2-m1::Mu plants, and three homozygous normalplants were analyzed. For the remethylation study, duplicate prepa-
rations from three plants of each class were analyzed. HPLC analysiswas conducted at the University of Wisconsin Biotechnology Center.A volume of 50 �L was injected into a Brownlee Laboratories Spheri-5RP-8 column (Alltech, Deerfield, IL). Nucleosides were separated ata flow rate of 0.75 mL/min. All samples were analyzed on a BeckmanInstruments System Gold chromatograph, and nucleosides were de-tected at A254 and A280. Nucleoside and nucleotide standards (Sigma)were used to determine nucleoside peak positions and to createstandard curves to determine nucleoside concentration. The ratio of5-methylcytosine to total cytosine was calculated relative to knownstandards as the mean value from the two wavelength readings, andstatistical analysis was conducted using SAS (SAS Institute, Cary,NC). The significance of the variance among genotypic classes wastested using the variance from plants nested within genotypes as theerror term. Pairwise differences among classes were assessed usingt tests, with the plants nested within genotypes also used as the errorterm.
Genomic Bisulfite Sequencing of the 180-bp Knob Repeat
DNA was denatured initially by adding 1.5 �L of 10 N NaOH to 8 �gof DNA in 50 �L of distilled, deionized water followed by incubationfor 15 min at 37C. After denaturation, 2.5 �L of 8-hydroxyquinolineand 450 �L of a 3.24 M urea–2 M sodium metabisulfite solution wereadded to the DNA solution and mixed gently. The solution was thendivided into 100-�L aliquots and overlaid with mineral oil. Thebisulfite modification was conducted using 20 alternating cycles ofdenaturation (95C for 1 min) and modification (55C for 15 min). Afterbisulfite modification, the DNA solutions were pooled and precipi-tated. The DNA was desalted with a Qiagen (Valencia, CA) PCR prep kit.
The 180-bp repeat was sequenced directly from the bisulfite-mod-ified genomic DNA using Big Dye terminator cycle sequencing on anABI sequencer according to the manufacturer’s instructions (Perkin-Elmer Applied Biosystems). Sequencing reactions were performedusing the buffer and enzyme supplied by the manufacturer in a 10-�Lvolume with 320 ng of DNA and 10 pg of primer. Sequencing reac-tions were conducted separately on each strand using the primers180-bp Forward (5�-CCACACAACCCCCATTTTT-3�) to sequenceone strand and 180-bp Reverse (5�-TCATACACCTCACCCCACAT-3�)to sequence the complementary strand.
Sequence Analysis
Sequence data were processed using tools available through theWorld Wide Web at http://dot.imgen.bcm.tmc.edu. CLUSTAL W wasused for multiple sequence alignments. Sequence alignments forpresentation were processed using BOXSHADE, which was availablethrough http://www.ch.embnet.org. Sequence comparisons and da-tabase searches using BLAST 2.0 were made through http://www.ncbi.nlm.nih.gov. Phylogenetic analysis was conducted usingthe PHYLIP programs available through http://bioweb.pastuer.fr.Phylogenetic trees were generated using the parsimony method.
GenBank Accession Numbers
The GenBank accession numbers are as follows; Arabidopsis chro-momethylases CMT1 (AF039367), CMT2 (CAA16759), and CMT3(AAG52543); Zmet1 (AF063403).
Chromomethylases Maintain CNG Methylation 1927
ACKNOWLEDGMENTS
We appreciate the assistance of Thomas Wright and his crew at theWest Madison Research Station, where the nursery is located. Weare grateful for the assistance of Lynn Hummel and Laura van Slyke,who helped to maintain plants in the greenhouse. We thank SteveJacobsen for ongoing discussions and review of the manuscript andfor sharing his data on plant chromomethylases. This work was sup-ported by Pioneer Hi-Bred International, Inc., and the University ofWisconsin Graduate School.
Received February 16, 2001; accepted May 25, 2001.
REFERENCES
Akhtar, A., Zink, D., and Becker, P.B. (2000). Chromodomains areprotein–RNA interaction modules. Nature 407, 405–409.
Banks J.A., Masson, P., and Fedoroff, N. (1988). Molecular mech-anisms in the developmental regulation of the Suppressor-muta-tor transposable element. Genes Dev. 2, 1364–1380.
Bensen, R.J., Johal, G.S., Crane, V.C., Tossberg, J.T., Schnable,P.S., Meeley, R.B., and Briggs, S.B. (1995). Cloning and charac-terization of the maize An1 gene. Plant Cell 7, 75–84.
Callebaut, I., Courvalin, J.C., and Mornon, J.P. (1999). The BAH(bromo-adjacent homology) domain: A link between DNA methyl-ation, replication and transcriptional regulation. FEBS Lett. 446,189–193.
Cao, X., Springer, N.S., Muszynski, M.G., Phillips, R.L., Kaeppler,S., and Jacobsen, S.E. (2000). Conserved plant genes with simi-larity to mammalian de novo DNA methyltransferases. Proc. Natl.Acad. Sci. USA 97, 4979–4984.
Cavalli, G., and Paro, R. (1998). Chromo-domain proteins: Linkingchromatin structure to epigenetic regulation. Curr. Opin. Cell Biol.10, 354–360.
Chandler, V.L., and Walbot, V. (1986). DNA modification of a maizetransposable element correlates with loss of activity. Proc. Natl.Acad. Sci. USA 83, 1767–1771.
Cheng, X., Kumar, S., Posfai, J., Pflugrath, J.W., and Roberts,R.J. (1993). Crystal structure of the HhaI DNA methyltransferasecomplexed with S-adenosyl-L-methionine. Cell 74, 299–307.
Finnegan, E.J., and Dennis, E.S. (1993). Isolation and identificationby sequence homology of a putative cytosine methyltransferasefrom Arabidopsis thaliana. Nucleic Acids Res. 10, 2383–2388.
Finnegan, E.J., Peacock, W.J., and Dennis, E.S. (1996). ReducedDNA methylation in Arabidopsis thaliana results in abnormal plantdevelopment. Proc. Natl. Acad. Sci. USA 93, 8449–8454.
Gehrke, C.W., McCune, R.A., Gama-Sosa, M.A., Ehrlich, M., andKuo, K.C. (1984). Quantitative reverse-phase high-performanceliquid chromatography of major and modified nucleosides in DNA.J. Chromatogr. 301, 199–219.
Gruenbaum, Y., Naveh-Many, T., Cedar, H., and Razin, A. (1981).Sequence specificity of methylation in higher plant DNA. Nature292, 860–862.
Henikoff, S., and Comai, L. (1998). A DNA methyltransferase
homolog with a chromodomain exists in multiple polymorphicforms in Arabidopsis. Genetics 148, 307–318.
Ingelbrecht, I., van Houdt, H., van Montagu, M., and Depicker, A.(1994). Posttranslational silencing of reporter transgenes intobacco correlates with DNA modification. Proc. Natl. Acad. Sci.USA 91, 10502–10506.
Jiang, J., Nasuda, S., Dong, F., Scherrer, C.W., Woo, S.S., Wing,R.A., Gill, B.S., and Ward, D.C. (1996). A conserved repetitiveDNA element located in the centromeres of cereal chromosomes.Proc. Natl. Acad. Sci. USA 93, 14210–14213.
Kakutani, T., Jeddeloh, J.A., and Richards, E.J. (1995). Character-ization of an Arabidopsis thaliana DNA hypomethylation mutant.Nucleic Acids Res. 23, 130–137.
Kermicle, J.L. (1996). Epigenetic silencing and activation of a maizer gene. In Epigenetic Mechanisms of Gene Regulation, V. Russo,R. Martienssen, and A. Riggs, eds (Cold Spring Harbor, NY: ColdSpring Harbor Laboratory Press), pp. 267–287.
Kozak, M. (1991). An analysis of vertebrate mRNA sequences: Inti-mations of translational control. J. Cell Biol. 115, 887–903.
Kumar, S., Cheng, X., Klimasauskas, S., Mi, S., Posfai, J., Roberts,R.J., and Wilson, G.G. (1994). The DNA (cytosine-5) methyltrans-ferases. Nucleic Acids Res. 22, 1–10.
Linn, F., Heidmann, I., Saedler, H., and Meyer, P. (1990). Epige-netic changes in the expression of the maize A1 gene in Petuniahybrida: Role of numbers of integrated gene copies and state ofmethylation. Mol. Gen. Genet. 222, 329–336.
Mascia, P.N., Rubenstein, I., Phillips, R.L., Wang, A.S., andXiang, L.Z. (1981). Localization of the 5S rRNA genes and evi-dence for diversity in the 5S rDNA region of maize. Gene 15, 7–20.
McMullen, M.D., Phillips, R.L., and Rubenstein, I. (1991). Molecu-lar analysis of the nucleolus organizer region in maize. In Chromo-some Engineering in Plants: Genetics, Breeding, and Evolution,P.K. Gupta and T. Tsuchiya, eds (New York: Elsevier), pp. 561–576.
Mette, M.F., Aufsatz, W., van der Winden, J., Matzke, M.A., andMatzke, A.J. (2000). Transcriptional silencing and promoter meth-ylation triggered by double-stranded RNA. EMBO J. 19, 5194–5201.
Nakayama, J., Rice, J.C., Strahl, B.D., Allis, D.C., and Grewal,S.I.S. (2001). Role of histone H3 lysine 9 methylation in epigeneticcontrol of heterochromatin assembly. Science 292, 110–113.
Posfai, J., Bhagwat, A.A., Posfai, G., and Roberts, R.J. (1989).Predictive motifs derived from cytosine methyltransferases.Nucleic Acids Res. 17, 2421–2435.
Pradhan, S., and Adams, R.L.P. (1995). Distinct CG and CNG DNAmethyltransferases in Pisum sativum. Plant J. 7, 471–481.
Pradhan, S., Cummings, M., Roberts, R.J., and Adams, R.L.P.(1998). Isolation, characterization and baculovirus-mediated expres-sion of the cDNA encoding cytosine DNA methyltransferase fromPisum sativum. Nucleic Acids Res. 26, 1214–1222.
Raikhel, N. (1992). Nuclear targeting in plants. Plant Physiol. 100,1627–1632.
Rea, S., Eisenhaber, F., O’Carroll, D., Strahl, B.D., Sun, Z.W.,
1928 The Plant Cell
Schmid, M., Opravil, S., Mechtler, K., Ponting, C.P., Allis, C.D.,and Jenuwein, T. (2000). Regulation of chromatin structure bysite-specific histone H3 methyltransferases. Nature 406, 593–599.
Rose, T.M., Schultz, E.R., Henikoff, J.G., Pietrokovski, S.,McCallum, C.M., and Henikoff, S. (1998). Consensus-degener-ate hybrid oligonucleotide primers for amplification of distantlyrelated sequences. Nucleic Acids Res. 26, 1628–1635.
Saghai Maroof, M.A., Soliman, K.M., Jorgensen, R.A., and Allard,R.W. (1984). Ribosomal spacer-length polymorphisms in barley:Mendelian inheritance, chromosome location, and populationdynamics. Proc. Natl. Acad. Sci. USA 81, 8014–8018.
Schultz, J., Copley, R.R., Doerks, T., Ponting, C.P., and Bork, P.
(2000). SMART: A web-based tool for the study of geneticallymobile domains. Nucleic Acids Res. 28, 231–234.
Schwartz, D., and Dennis, E. (1986). Transposase activity of the Accontrolling element in maize is regulated by its degree of methyla-tion. Mol. Gen. Genet. 205, 476–482.
Walker, E.L. (1998). Paramutation of the r1 locus of maize is associ-ated with increased cytosine methylation. Genetics 148, 1973–1981.
Wassenegger, M. (2000). RNA-directed methylation. Plant Mol.Biol. 43, 203–220.
Wassenegger, M., Heimes, S., Riedel, L., and Sanger, H.L. (1994).RNA-directed de novo methylation of genomic sequences inplants. Cell 76, 567–576.
