ISSN 10227954, Russian Journal of Genetics, 2011, Vol. 47, No. 3, pp. 279–288. © Pleiades Publishing, Inc., 2011. Original Russian Text © V.V. Ashapkin, L.I. Kutueva, B.F. Vanyushin, 2011, published in Genetika, 2011, Vol. 47, No. 3, pp. 320–331.
MOLECULAR GENETICS
Is the Cytosine DNA Methyltransferase Gene MET1 Regulated by DNA Methylation in Arabidopsis thaliana Plants?1 V. V. Ashapkin, L. I. Kutueva, and B. F. Vanyushin A.N. Belozersky Institute of PhysicoChemical Biology, Moscow State University, Moscow, 119991 Russia email:
[email protected] Received May 19, 2010
Abstract—The methylation patterns of the MET1 gene in organs of Arabidopsis thaliana were studied by South ern blot hybridization of DNA samples hydrolyzed with differentially methylationsensitive restriction endonu cleases. A highly methylated on internal cytosine residue CCGG site was found 1.5 kb upstream of the gene, whereas CCGG sites located in more proximal parts of the 5'flanking region and the gene itself are essentially unmethylated. This methylation pattern was observed in different organs of plants belonging to two different ecotypes as well as in different transgenic plant lines. The methylation level of a CCGG site in exon 3 (2.1 kb from the gene’s 5'end) occurred to be variable between different transgenic plant lines and two ecotypes studied. Transcription levels of the MET1 gene vary slightly in organs of wildtype plants without any obvious correlation with its methylation. The transgenic antisense MET1 constructs expressed in plant genome do affect both MET1 methylation and its transcription but again without any obvious correlation. The comparative investigation of transcription levels of different genes of cytosine DNA methyltransferase family MET (MET1, MET2a, MET2b, MET3) and their methylation patterns shows that there may exist some mechanisms defending the most actively transcribed gene MET1 of this family from methylation mediated silencing. In contrast to DRM2 gene we could not find any adenine methylated GATC sites in the MET1 gene. DOI: 10.1134/S1022795411020037 1
DNA methylation takes part in the tissuespecific gene transcription control and parental imprinting, keeps expression of repeated sequences within an acceptable range for the plant wellbeing and inacti vates all kinds of potentially dangerous elements in the plant genome [1, 2]. The silent state of genes usually correlates with methylation of their promoters, whereas the promoter hypomethylation is as a rule associated with transcription reactivation. In contrast to animals, where methylation deals mostly with the symmetric sequences CpG, plant DNA is extensively methylated at two types of symmetric sequences, CpG and CpNpG, as well as at asymmetric ones CpNpN (where N is A, C or T). Plants have three rather well studied gene families that code for cytosine DNA methyltransferases. The genes of MET family code for DNA methyltrans ferases most related to mammalian Dnmt1 [3]. These are the major CpGspecific maintenance DNA meth yltransferases, that seem to be an essential part of the mechanisms determining the developmental phase transitions in the plant life cycle. A second class of DNA methyltransferases, the CMT or “chromometh ylase” family, is unique to plants [4]. All chromometh ylases known today contain a bromo adjacent homol ogy (BAH) domain, a feature also common to Dnmt1/MET methyltransferases. This domain is
1 The article was translated by the authors.
believed to be implicated in conjugation of DNA methylation to its replication and transcriptional reg ulation [5]. The DNA methyltransferases of both fam ilies are involved in the hereditary maintaining the symmetric methylation patterns, the MET family enzymes (at least its major member MET1) acting on hemimethylated CpG sites [6], whereas the chro momethylases methylating symmetrical trinucleotide CpNpG sites [7, 8]. A third family of plant DNA meth yltransferases is the “domain rearranged methyltrans ferases” or DRM, most related to animal de novo meth yltransferases Dnmt3 [9]. DRM’s role as major de novo plant DNA methyltransferases was reasonably well con firmed by genetic analysis of relevant mutants [10, 11] and directly demonstrated in an elegant series of in vitro and in vivo experiments [12]. DNA methyltransferases of both plant specific families DRM and CMT seem to be required for the proper maintenance of the methyla tion patterns at asymmetric and CpNpG sites. The met1 lossoffunction mutations essentially eliminate the CpG methylation throughout the genome. However, such mutants were also found to have significantly reduced CpNpG and asymmetric methylation levels at some loci [13, 14]. DNA methyl transferase MET1 is fully active in drm1 drm2 cmt3 tri ple mutant strains, defective on major DNA methyl transferases of two other families. Nevertheless it is not sufficient for the maintenance of DNA methylation at the CpNpG and asymmetric sites. It seems most likely
279
280
ASHAPKIN et al.
that the partial loss of CpNpG and asymmetric meth ylations in met1 mutants is not directly caused by pro miscuous enzymatic activity of the MET1 enzyme itself but rather is a secondary effect caused by the pri mary loss of CpG methylation. On the other hand in a number of the Arabidopsis transgenic antisenseMET1 plant lines and met1 mutant lines the SUP and AG genes were found to be greatly hypermethylated at both CpNpG and asymmetric sites, while overall DNA methylation level in these plant lines was decreased to ~10% of wildtype plants level [15]. One possible explanation is that factors controlling fidelity of genomic methylation are themselves regulated by DNA methylation. A second more trivial possibility is that residual DNA methyltransferase activity in these hypomethylation mutants is in some way hyper acti vated when overall methylation is too low, resulting in ectopic methylation of some genes. We have studied earlier the effects of MET1 inacti vation on DRM2 gene methylation in a number of transgenic lines of Arabidopsis containing parts of MET1 gene coding sequence as antisense constructs [16]. Variable patterns of methylated cytosine residues were found in wildtype and transgenic plants. Inter estingly, along with cytosine methylation the DRM2 gene was found to be methylated at some adenine res idues (Gm6ATC); the adenine methylation degree of DRM2 gene being variable in wildtype and transgenic plants. Thus MET1 activity definitely affects methyla tion and probably expression of the DRM2 gene. Whether the expression of the MET1 gene itself is regulated by DNA methylation, so far as we are aware, never was studied. Recent progress in genomewide technologies has lead to the first versions of Arabidop sis DNA methylome [17, 18]. About a third of all expressed genes were found to be methylated, but MET1 gene appeared to belong to the unmethylated two thirds. That could well be regarded as a negative answer to the question above, but as a matter of fact, we have already done an investigation of cytosine methylation pattern of Arabidopsis MET1 gene by Southern blot hybridization earlier and found it to contain some methylated sites [19]. Namely, we have detected some methylated sites in the 5'flanking region and in the first exons of MET1 gene, whereas the remaining best part of its coding region was found to be totally unmethylated. This methylation pattern appeared to be just opposite to a common one usually found in expressed genes. Unfortunately, the sequence of the MET1 gene having been still unknown at that time, we could draw only a tentative picture of its methylation [19]. Here we investigated the methylation patterns and expression of the MET1 gene in various organs of Ara bidopsis thaliana plants, including several transgenic plant lines containing parts of the MET1 gene coding sequence as antisense constructs. We have also done a comparative analysis of DNA methylation patterns of the other MET family DNA methyltransferase genes.
MATERIALS AND METHODS Plants The cultivation of Arabidopsis thaliana plants (EnkheimT and Columbia ecotypes) and their trans formation were described earlier [16]. The transgene presence in these plants was verified by Southern blot hybridization, and plants containing one or two copies of transgene were used for further experiments. The recombinant constructs were obtained by cloning parts of MET1 coding sequence in antisense orienta tion under copperinducible promoters in plasmid vectors pPMB7066 and pPMB768 and subcloning the resulting chimeric constructs into a binary vector plas mid pPMB765 [20, 21]. Three segments of MET1 coding sequence were used: a 5'end proximal 0.6 kbp fragment of the Arabidopsis thaliana MET1 gene (from nucleotide –48 to nucleotide 592 relative to the first nucleotide of exon 1) from the plasmid pMAT1 [19] and fragments of the fulllength MET1 cDNA sequence from the recombinant plasmids pYc2 (~2 kbp of the 5'end proximal part) and pYc8 (~2.6 kbp of the 3'end proximal part) [22]. Two vari ants of copperinducible promoters used were: PMB7066, driving a copperinducible transcription in leaves of transgenic plants and a constitutive transcrip tion in their roots [20] and PMB768, silent in leaves and fully copperinducible in roots [21]. The respec tive homozygous lines produced were divided into six groups denoted as pYc27066, pYc2768, pYc87066, pYc8768, MAT7066, MAT768, indicating the kind of the MET1 target sequence and the variant of pro moter used. Individual plant lines in each of these groups were identified by consecutive numbers. For copperinduction of transgenes the respective populations of 14dayold plants were grown for four weeks being watered with a 10 μM CuSO4 solution. Analysis of DNA Methylation Patterns The MET1Pro fragment corresponding to the 0.6 kbp 5'flanking sequence of MET1 gene (from nucle otide –594 to nucleotide 49 relative to first nucleotide of exon 1) was produced by PCR amplification of the respective segment of A. thaliana genomic DNA and cloned in a plasmid vector pGEM3Zf+ (Promega). The cloned fragments of plasmids pMET1Pro, pYc2 and pYc8 were excised with appropriate restriction endo nucleases and purified by preparative electrophoresis in lowmelting point agarose gels, labeled with [α 32P]dATP using MegaPrime kit (Amersham) and used as hybridization probes in Southern blot hybridization experiments. The DNA samples were isolated from freshly harvested plant organs essentially as described [19]. In a typical experiment the DNA samples were first digested with a rarecutter restriction endonu clease HindIII, BamHI or EcoRV, 5 activity units per μg DNA, for 2 hrs at 37°C in a buffer solution recom mended by the supplier (Fermentas). The resulting
RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
2011
IS THE CYTOSINE DNA METHYLTRANSFERASE GENE MET1 REGULATED
hydrolyzed DNA was precipitated with two volumes of –20°C cold ethanol in the presence of 2 M ammo nium acetate, rinsed three times with 70% ethanol, dried and solubilyzed in water, 10 μl per 1 μg, divided into 1 μg portions and digested with one of the second ary methylationsensitive restriction endonucleases: HpaII—cuts unmethylated CCGG sites but is inhib ited by methylation of either of its C residues; MspI— an isoschizomeric endonuclease that cuts both CCGG and Cm5CGG sites but is inhibited by methylation of external C residue; HhaI—cuts GCGC sites but is inhibited by methylation of either of its C residues; DpnI—cuts Gm6ATC sites but not the adenineunm ethylated GATC sites, irrespective of their C methyla tion status; MboI—an isoschizomeric endonuclease with opposite sensitivity to adenine methylation, that cuts GATC sites but not the Gm6ATC ones, again irre spective of their C methylation status; Sau3A— another isoschizomeric endonuclease that cuts both GATC and Gm6ATC sites but not the GATm5C ones. The Southern blot hybridization was done as described previously [16]. The restriction fragments of the MET1 gene produced upon DNA cleavage with restriction endonucleases were compared to a detailed map of their corresponding recognition sites deduced from the known nucleotide sequence of the MET1 gene (NCBI Entrez GeneID: 834975; TAIR:AT5G49160). The same methods were used to study the methylation patterns of other MET family genes. Analysis of Gene Expression Levels of the MET1 and other genes mRNAs in sam ples of total highmolecular weight RNA isolated from different plant organs were studied by Northern blot hybridization with respective 32Plabeled probes. Total RNA was prepared from freshly harvested plant organs by a modification of the acid guanidine thiocyanate phenolchloroform extraction method [23]. RNA sam ples (5 μg per lane) were fractionated on formaldehyde containing agarose gels [24]. The RNA were transferred from gels to Hybond–N+ membranes by vacuum blot ting and hybridized to 32Plabeled probes accordingly with the supplier’s directions (Amersham). RESULTS OrganSpecific Patterns of MET1 Gene Methylation and Expression in WildType Plants The MET1 gene fragments produced upon genomic DNA cleavage with the restriction endonu cleases having various methylation sensitivity are shown on the upper part of Fig. 1. To make the under standing of these results more comfortable a map of respective sites and positions of the restriction frag ments detected on Southern blots relative to MET1 gene sequence is presented on the Fig. 1 lower part. All RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
281
positions here and downwards are shown in kilo base pairs (kbp) relative to transcription initiation site. As can be seen on Fig. 1, one 4.3 kbp fragment is pro duced upon the MET1 gene 5'end proximal part cleav age both with single restriction endonuclease HindIII and its combination with the restriction endonuclease HpaII. This proves the only CCGG site present in this part of the MET1 gene (~1.5 kbp upstream of transcrip tion initiation site) to be methylated at the internal C res idue. A shorter fragment (~3 kbp) is a predominant prod uct of DNA cleavage with a HindIII + MspI combina tion, that is obviously a result of nearly full cleavage of the CCGG site mentioned above. Thus this site is fully methylated on the internal cytosine residue and essen tially unmethylated on the external one. Three pYc2 hybridizing fragments are seen upon DNA cleavage with restriction endonuclease BamHI: a most strong one of ~1.4 kbp obviously corresponds to the gene seg ment between BamHI sites (0.9) and (2.3). Its high intensity appears to be a trivial consequence of its full overlapping with the pYc2 probe sequence, whereas only short end sequences of neighbor fragments over lap the probe and hence produce weaker bands. The major 1.4 kbp BamHI fragment is replaced with a shorter one of ~1.2 kbp upon DNA hydrolysis with a BamHI + HpaII combination as a consequence of full cleavage of CCGG site (2.1) by HpaII. The same 1.2 kbp fragment is seen upon DNA cleavage with a BamHI + MspI combination. Hence, the CCGG site (2.1) in the third exon of MET1 gene is unmethylated on both cytosine residues. Another fragment of ~3.9 kbp length produced by BamHI + HpaII combi nation is obviously a result of DNA cleavage on sites CCGG (–3.0) and BamHI (0.9). Hence the CCGG site (–3.0), in contrast to lower CCGG site (–1.5), is unmethylated on both cytosine residues. A shorter (~2.4 kbp) fragment replaces the 3.9 kbp fragment when the combination BamHI + MspI used, evidently a result of CCGG (–1.5) and BamHI (0.9) sites cleav age. Thus CCGG sites (–3.0) and (2.1) are fully unm ethylated on both cytosine residues in leaf DNA, whereas the CCGG site (–1.5) is fully methylated on the internal C residue and nearly unmethylated on the external one. The methylation patterns of the 5'end proximal part of the MET1 gene in stem and flower DNAs are essentially the same. The only notable dif ference seems to be a larger proportion of DNA mole cules methylated on the external C residue of CCGG site (–1.5) as evidenced by a higher relative quantity of MspI resistant 4.3 kbp HindIII (Fig. 1a, lanes 8, 12) and 3.9 kbp BamHI fragments (lanes 10, 14). Thus, the only heavily methylated part of the 5'proximal region of MET1 gene is represented by the CCGG site (–1.5) fully methylated on the internal C residue and variably methylated on the external one in all Arabidopsis organs investigated: leaf, stem, flower (Fig. 1a) as well as in root, green silique and mature seed (not shown). A major long fragment (~5.3 kbp) and a barely vis ible short one of ~0.9 kbp are produced upon the 2011
282
ASHAPKIN et al. (a) 1 2 3 4 5 6 7 8 9 10 11 12 13 14
(b) 1 2 3 4 5 6 7 8 9 10 11 12 13 14
23.1 9.42 6.56 4.36 23.1 9.42 6.56 4.36 2.32 2.03
2.32 2.03
0.56
0.56
(c) B (0.9) B (2.3) H (1.7; 2.0) H (2.9)
B (–7.7) H (–2.6) –3.0
–1.5
B (5.6)
2.1 2.6
4.5 4.8
B (14.6) H (8.2)
5.8 6.6–6.9
1 kbp pYc2 cDNA probe pYc8 cDNA probe
HindIII HindIII + HpaII
4.3
0.9
4.3
0.5
1.6
1.8 1.3
0.5
3.2 HindIII + MspI
1.6
4.3
3.9
BamHI + HpaII
3.9
2.4
1.8 1.0
1.4
8.6
BamHI
BamHI + MspI
5.3
3.3
1.2
1.9 2.2
1.1
1.2
1.9
0.8
RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
2011
IS THE CYTOSINE DNA METHYLTRANSFERASE GENE MET1 REGULATED
MET1 gene 3'proximal part cleavage with restriction endonuclease HindIII (Fig. 1b, lane 1). As is quite evi dent from the restriction map of MET1 gene the long fragment corresponds to the gene sequence between HindIII sites (2.9) and (8.2), whereas the short one to the gene sequence between HindIII sites (2.0) and (2.9). A number of short fragments are seen upon DNA cleavage with a HindIII + HpaII combination, a major one of about 1.6 kbp and few weaker ones of about 1.8, 1.3, 0.8 and 0.5 kbp (Fig. 1b, lane 2). The first one is obviously a product of DNA cleavage on sites HindIII (2.9) and CCGG (4.5). Hence, the CCGG site (4.5) in the center of exon 8 is essentially or fully unmethylated on both C residues. Skipping the tiresome details, the thorough comparison of all the fragments produced upon DNA hydrolysis with endo nuclease combinations used in this study shows that CCGG sites (2.6) and (4.5) are unmethylated on both cytosine residues, whereas the CCGG sites (4.8) and (5.8) are partially methylated on the internal cytosine residue and unmethylated on the external one. A group of closely spaced CCGG sites in the 3'flanking region (6.6–6.9) seems to be fully unmethylated. The methylation patterns of the MET1 gene 3'end proximal part in stem and flower DNAs are quite sim ilar to that in leaf DNA. We have used the EcoRII and MvaI isoschizomeric pair of restriction endonucleases to study the MET1 gene methylation of CpNpG type. Both recognize and cleave the same pentanucleotide sequence CCWGG (where W is A or T). Cleavage with EcoRII is inhibited by internal cytosine methylation, whereas MvaI cleaves DNA irrespective of cytosine methylation. All CCWGG sites in the MET1 gene region were found to be unmethylated (not shown). Hence, involvement of the CpNpGtype methylation in regulation of the MET1 gene transcription seems to be quite unlikely. In order to understand whether the minor variabil ity in the methylation levels of some CCGG sites observed between genomic DNA of different plant organs is of any functional significance we studied the transcription levels of MET1 gene in different organs from the same plants. As is evident from results shown on Fig. 2, the gene seems to be actively expressed in all organs studied. Moderate variations in MET1 mRNA levels observed in different organs can probably be
283
pYc8 probe flower
stem
leaf
root
~5 kb
Fig. 2. Northern blot detection of MET1 mRNA in organs of A. thaliana var. EnkheimT.
accounted for by differences in the cell proliferative activities. This is in perfect accord with the current view of MET1 as a principal, if not sole, CpGspecific maintenance cytosine DNA methyltransferase [1, 2]. We could not see any evident correlation between peculiarities of the MET1 gene methylation and levels of its expression. It seems quite probable that expres sion of this major maintenance DNA methyltrans ferase coordinated with DNA replication in plant cells is achieved by a hierarchy of various molecular mech anisms, DNA methylation being only a part of this system. Effects of the Antisense MET1 Constructs on the MET1 Gene Methylation and Expression As a part of further investigation of the DNA meth ylation role in regulation of MET1 gene expression we have studied, whether the experimental manipulations with MET1 gene expression could affect its methyla tion patterns. To this end we have produced a collec tion of transgenic plants containing different parts of MET1 gene coding sequence as antisense constructs under the control of copperinducible promoters. In general, the observed patterns of the MET1 gene methylation in leaf DNA of the wildtype plants and transgenic plants produced from them appeared quite similar to those described above, though for technical reasons the plants of another ecotype (Columbia) were used. The only notable difference was the partial methylation of the CCGG site (2.1) in Columbia as opposed to its fully unmethylated state in EnkheimT. Interesting, the methylation level of this particular site
Fig. 1. Methylation patterns of the MET1 gene in organs of A. thaliana var. EnkheimT. The fragments of MET1 gene hybridizing to a 5'end proximal probe pYc2 (a) and a 3'end proximal probe pYc8 (b) are shown. The positions of all sites, detected on Southern blots (a) and (b), and hybridization probes are shown on the lower part (c). The lanes on (a) and (b) contain: 1—leaf DNA digested with HindIII; 2—leaf DNA digested with HindIII plus HpaII; 3—leaf DNA digested with HindIII plus MspI; 4—leaf DNA digested with BamHI; 5—leaf DNA digested with BamHI plus HpaII; 6—leaf DNA digested with BamHI plus MspI; 7—stem DNA digested with HindIII plus HpaII; 8—stem DNA digested with HindIII plus MspI; 9—stem DNA digested with BamHI plus HpaII; 10—stem DNA digested with BamHI plus MspI; 11—flower DNA digested with HindIII plus HpaII; 12—flower DNA digested with HindIII plus MspI; 13—flower DNA digested with BamHI plus HpaII; 14—flower DNA digested with BamHI plus MspI. The positions and length of marker phage λ DNA HindIII restriction fragments are shown on the left; (c)—Schematic representation of MET1 gene methylation pattern. Restriction sites are: H (HindIII), B (BamHI), black circles—CCGG sites fully methylated at the internal C, grey circles—CCGG sites partially methylated at the internal C, open circles—unmethylated CCGG sites. Exons of MET1 are shown by rectangles, introns—by thin lines between. Gene map positions of restriction fragments detected on Southern blots (a) and (b) are shown below. Major restriction fragments are shown by the thick lines, minor fragments—by the thin ones. RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
2011
284
ASHAPKIN et al. (a) II M
I H
M
H
(b) 1 2
III H
M
4.1
(c) 3 4
8.3
2.7
1.4
3.6
(d)
E (–1.5)
1.0 E (6.8)
–1.5 –0.5–0.1 0.5
2.1 2.6
4.5 4.8 5.5 5.8 6.6–6.7
1 kbp
MET1Pro probe
8.3
EcoRV + HpaII + HhaI + Sau3A + DpnI
4.1 3.6 1.4
2.7 1.0 8.3
Fig. 3. Methylation patterns of the 5'end proximal part of the MET1 gene in transgenic lines of A. thaliana var. Columbia. The fragments of the MET1 gene 5'end proximal region were analyzed by Southern blot hybridization to a MET1 promoter probe, MET1Pro, shown on (d). (a)—Three types (IIII) of the methylation patterns of CCGG sites: H—HpaII, M—MspI. (b)— Methylation pattern of GCGC sites—HhaI. (c)—Methylation pattern of GATC sites: 1—EcoRV alone, its combinations with: 2—DpnI, 3—MboI, 4—Sau3A. (d)—Schematic representation of MET1 gene methylation patterns. Positions of relevant restriction sites are shown by E (EcoRV) or graphic symbols: circles—CCGG sites, ovals—GATC sites, or squares—GCGC sites. Exons of MET1 gene are shown by rectangles, introns—by thin lines between. Fully methylated HpaII (–1,5) is shown by a black circle, partially methylated HpaII sites—by grey circles, unmethylated sites—by open symbols. Positions of the restric tion fragments detected on Southern blots (a), (b) and (c) are shown below.
seemed to vary between wildtype and different trans genic plants. To get a more detailed picture we studied the fragments cleaved out from respective part of the MET1 gene with restriction endonucleases EcoRV, HpaII and MspI. The fragments produced were visual ized by Southern blot hybridization with a MET1Pro probe complementary to a short 0.64 kb sequence just upstream of the MET1 gene first exon. A pair of hybridizing bands of about 4.1 and 3.6 kbp were pro duced from each DNA sample investigated upon DNA hydrolysis with the EcoRV + HpaII combina tion. The longer one is a product of DNA cleavage on sites EcoRV (–1.5) and CCGG (2.6) in DNA mole
cules having the CCGG site (2.1) methylated, whereas the shorter one results from DNA cleavage on sites EcoRV (–1.5) and CCGG (2.1) in DNA molecules having the last one unmethylated (Fig. 3d). As both fragments cover the entire sequence of the MET1Pro probe, the relative intensities of corresponding 4.1 and 3.6 kbp bands can be used as a quantitative measure of DNA molecules methylated and unmethylated at the CCGG site (2.1). Three major patterns of the corre sponding blot hybridization bands, denoted as I, II and III on Fig. 3a, have been found in different plant lines analyzed. The first one (I) characterized by an approximately equal intensities of 4.1 and 3.6 kbp
RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
2011
IS THE CYTOSINE DNA METHYLTRANSFERASE GENE MET1 REGULATED
fragments was found in the wildtype plants, as well as in plants of two transgenic lines, MAT70663 and pYc270664, grown under inducing conditions (watering with 10 μM CuSO4 solution for four weeks). Thus, the methylation level of the CCGG site (2.1) in these plants appears to be about 50%. As can be judged from results of DNA cleavage with the EcoRV + MspI combination, there is some methylation on the exter nal C residue also, but a weaker one (lane M). The sec ond methylation pattern (II) characterized by a lower intensity of 3.6 kb band produced both with HpaII and MspI digestions was found in plants of the same trans genic lines MAT70663 and pYc270664 mentioned above but grown under noninducing conditions, as well as in plants of another transgenic line, pYc27685 irrespective of growth conditions. In these plants the CCGG site (2.1) appears to be nearly fully methylated on both cytosine residues. The third methylation pat tern (III) characterized by a lower intensity of the long 4.1 kbp fragment was found in plants of a single trans genic line, pYc870661, irrespective of growth condi tions. In these plants the CCGG site (2.1) appears to be essentially unmethylated on both cytosine residues. To render the study of the MET1 cytosine methyla tion patterns more complete, we have also analyzed the methylation state of GCGC sites. To this end the cleavage fragments produced upon hydrolysis of DNA with endonucleases HhaI and EcoRV were visualized by blot hybridization with labeled MET1Pro and pYc8 probes. No fragments other than expected from com plete cleavage of the gene at all GCGC sites were observed. A typical pattern of MET1Prohybridizing bands is shown in Fig. 3b. Both fragments expected, 1.4 kbp from DNA cleavage on sites EcoRV (–1.5) and HhaI (–0.1) and 2.7 kbp from its cleavage on HhaI sites (–0.1) and HhaI (2.6) are clearly visible. A visibly higher intensity of the shorter fragment results from its more extensive overlap with the MET1Pro probe used. The absence of longer fragments in this digest proves the DNA sample to be fully cleaved on both HhaI sites (–0.1) and (2.6). Hence these sites are fully unmethy lated on both cytosine residues. The same is true for a close trio of GCGC sites in exon 11 (data not shown). The possible methylation of MET1 gene on ade nine and cytosine residues at GATC sites was analyzed with a trio of isoschizomeric endonucleases DpnI, MboI and Sau3A, differently affected by these types of DNA methylation. We have not found any indications of MET1 cleavage by endonuclease DpnI. The 8.3 kbp EcoRV DNA restriction fragment containing the entire MET1 gene sequence (Fig. 3d) is fully resistant to digestion with DpnI (lanes 1 and 2 in Fig. 3c), though it contains more than 20 GATC sites. Thus, none of these sites are adeninemethylated. This con clusion is confirmed by complete cleavage of these sites by an adenine methylationsensitive endonu clease MboI (lane 3 on the same Fig. 3c). Hence, in contrast to DRM2 [16] the MET1 gene appears to be completely unmethylated on adenine residues at RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
1
pYc8 probe 2 3 4
285
5
~5 kb
Fig. 4. Northern blot detection of MET1 mRNA in trans genic plant lines of A. thaliana var. Columbia: Lane 1— wild type plants; 2—MAT70663 line plants grown under noninducing conditions; 3—same plants grown under inducing conditions; 4—pYc870661 line plants grown under noninducing conditions; 5—same plants grown under inducing conditions.
GATC sites. The cytosine methylationsensitive endonuclease Sau3A completely cleaves all GATC sites present in MET1 gene sequence. Hence these sites are not methylated on cytosine residues also. Thus, the methylation level of the CCGG site (2.1) in the third exon of the MET1 gene seems to be the only trait variable between leaf DNA of different plant lines used in our investigation. In order to understand whether this variability is of any functional signifi cance, we have compared the MET1 transcription lev els in plants having different methylation patterns at this site. The maximum level of MET1 mRNA was observed in wildtype plants (Fig. 4, lane 1). Let us remind that the methylation level of CCGG site (2.1) in these plants is about 50% (Fig. 3a, I). The plants of a transgenic line MAT70663 with a high methyla tion level of CCGG site (2.1) (Fig. 3a, II) have a some what reduced level of MET1 mRNA (Fig. 4, lane 2). The same plants grown under inducing conditions have an even more reduced level of MET1 mRNA (Fig. 4, lane 3), though the methylation level of the CCGG site (2.1) in these plants is quite similar to that of the wildtype plants (Fig. 3a, I). Plants of the only transgenic line, that have been found to be largely unm ethylated at the CCGG site (2.1) (Fig. 3c, III) irrespec tive of growth conditions, namely pYc870661, have a MET1 mRNA level (Fig. 4, lanes 4, 5) intermediate between noninduced (Fig. 4, lane 2) and induced (Fig. 4, lane 3) plants of the transgenic line MAT 70663. Comparative Investigation of Methylation Patterns of Different MET1 Family Genes None of our experiments described above showed any evident correlation between MET1 gene tran scription and its methylation patterns. Does it mean that DNA methylation does not play any significant role in regulation of MET1 expression? We believe there is no definite answer to this question. On the one hand we have quite sound reasons to suggest MET1 2011
286
ASHAPKIN et al. –1.5
–0.5–0.1
0.5
2.1 2.6
4.5 4.8
5.5 5.8 6.6–6.7
MET1
1 kbp
MET2a
MET2b
MET3
Fig. 5. Methylation pattern of the MET family genes of A. thaliana. Positions of relevant restriction sites are shown by: circles— CCGG sites, ovals—GATC sites, or squares—GCGC sites. Exons are shown by rectangles, introns—by thin lines between. Fully methylated sites are shown by black symbols, partially methylated ones—by grey symbols, unmethylated ones—by open symbols. Positions of partially methylated sites taken from BSSeq methylome (http://epigenomics.mcdb.ucla.edu/DNAmeth/) are shown by asterisks. Positions of the direct repeat sequences are shown by dotted lines.
gene to be reliably protected from transcriptioninhib itory effects of DNA methylation in normal condi tions by some specific mechanisms. We could not cre ate experimental conditions where these mechanisms would malfunction, but this does not mean that it is principally impossible. An indirect evidence in favor of this hypothesis could come from comparative analysis of DNA methylation patterns of other MET family genes that being quite close homologs of MET1 never theless differ from it by far lower transcription levels. It is worth adding that biological significance of these genes in Arabidopsis genome, to say nothing about their specific functions, is still unknown.
We have studied the methylation patterns of all MET family genes on CCGG and GCGC sites by methods described above. The final results of this comparative analysis are presented on Fig. 5. As can be seen the MET2a (TAIR:AT4G14140) and MET2b (TAIR:AT4G08990) genes, commonly considered to be a result of resent gene duplication, contain a vast highly methylated region including most of their cod ing sequences except for exon 1. The extensive methy lation of these genes in all sequence contexts (CpG, CpNpG and CpNpN) seems to be in perfect accord with a rather low level of their transcription in plant organs. Still another gene of the MET family, namely
RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
2011
IS THE CYTOSINE DNA METHYLTRANSFERASE GENE MET1 REGULATED
MET3 (TAIR:AT4G13610), has a rather unusual methylation pattern. It consists of a completely unm ethylated 5'flanking region and essentially unmethy lated coding sequence except for 3'end proximal exons XI–XIII, containing a cluster of methylated CpG, CpNpG and CpNpN sites. As far as we are aware, the expression of the MET3 gene is still essen tially unstudied. According to our own preliminary data its level is slightly higher than that of MET2a, b genes, but by far lower as compared to MET1 gene. DISCUSSION Our early deductions concerning the MET1 gene methylation pattern [19] fully conform to the results of the present more extensive investigation. A new finding deserving special attention is that the immediate 5' flanking region of the MET1 gene seems to be totally unmethylated (three GCGC sites between (–0.2) and (–0.1) and GATCG site (0.5)). Quite probably that unmethylated region contains all the major transcription regulatory elements. The invariantly methylated CCGG site (–1.5) resides inside a dispersed repeat sequence located further upstream between –2 and –0.65 kbp of the MET1 gene (denoted by a dotted line on Fig. 5). Whether this sequence contains some regulatory ele ments relevant to MET1 transcription still remains to be seen. The copies of this repeat sequence are present in intergenic regions at variable genomic locations, all of them being highly methylated. It is interesting that this repeat sequence upstream of MET1 gene exactly corresponds to pronounced peaks of siRNAs produc tion (http://neomorph.salk.edu/epigenome.html). We wonder whether this highly methylated sequence could serve as an insulator mechanism preventing interfering aberrant readthrough transcription from sequences upstream of the MET1 gene promoter. A second new finding is that some sites at the 3'end prox imal part of the MET1 gene are partially methylated. Recently highresolution mapping of methylated sites in the whole genome of Arabidopsis thaliana was described [14, 17, 18, 25]. Unlike our finding the 5meth ylcytosine immunoprecipitation (mCIP) method [17, 18] did not reveal any significant methylation of the MET1 gene except for a short region just upstream of CCGG (–1.5) site (online Arabidopsis methylome at http://epige nomics.mcdb.ucla.edu/DNAmeth/). It seems that sensi tivity of the mCIP method is not high enough for detec tion of rare methylated sites in the MET1 coding region. This assumption is readily supported by results of a bisulfite conversion ultrahighthroughput sequencing (BSSeq) [14, 25]. In perfect agreement with our results, BSSeq method showed a number of significantly methylated sites in the 5'end proximal part of the MET1 coding region as well as some slightly methy lated sites in its 3'end proximal part (denoted by asterisks in Fig. 5). Unfortunately, methylation of the dispersed repeat element containing the CCGG site (–1.5) in the MET1 promoter region (–0.65 and RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
287
above) could not be analyzed by BSSeq method. An important finding of this method is detection of a clus ter of methylated sites immediately downstream of this repeat (between (–0.65) and (–0.54)) and just upstream of the unmethylated proximal part of the promoter region detected in our study (Fig. 5). One must admit that the genomewide approaches to anal ysis of DNA methylation, though giving a wealth of information concerning general tendencies of gene methylation, are still at their infancy and suffer from a number of important limitations. All DNA methy lome versions obtained so far are combined static snapshots of genome methylation of multiple cell types at certain developmental stages. The justified conclusions concerning regulation of individual genes by cellspecific variations in DNA methylation are, therefore, very limited. The mCIP studies have two additional technical shortages, namely low sensitivity (they seem to detect only about half of all methylated cytosine residues [25]) and the intrinsic inability to identify the sequence context of individual methylated sites. The sequencing of bisulfite treated DNA on a genomewide scale does not have these shortages, but it is very expensive and laborintense and still not available for most laboratories. Thus, investigation of individual genes still remains the main approach in the studies of biological significance of developmental and tissuespecific variations in DNA methylation. We have not found any correlation between MET1 gene methylation on cytosine residues and its tran scription. It is not surprising for normal plant organs since both methylation and expression levels of this constitutive gene vary in rather close limits. But in an experimental situation, when the transcription level of MET1 gene was lowered manifolds with transgenic antisense constructs, its methylation pattern was min imally affected without any obvious correlation with transcription levels. Thus MET1 gene seems to be surely protected from methylation in plant cells pro viding its stable transcription in all organs. Interestingly both significantly cytosine methylated segments of MET1 gene are associated with the dispersed repeat sequences: first one between –2.0 and –0.67 kbp of 5'flanking region. Second one—between 1.8 and 2.38 kbp inside the 5'end proximal part of the gene. The methylated sites detected in our study are located inside these repeat elements, though genomewide bisulfite sequencing shows some spreading of methylation down stream of the one in 5'flanking region (four methylated CpG sites between –0.65 and –0.54) and upstream of the one in the 5'end proximal part of the gene (cluster of methylated CpG sites between 1.0 and 1.8) (http://epige nomics.mcdb.ucla.edu/DNAmeth/). Quite different from the upstream repeat sequence, found at ubiquitous locations throughout the genome of A. thaliana, the down stream one is present inside the coding regions of the MET family genes only. Recently repeatdirected recruitment of nonCpG methylation and siRNA spreading upstream of an 2011
288
ASHAPKIN et al.
A. thaliana regulatory gene SDC was described [26]. The dispersed repeat element in the 5'flanking region of MET1 gene may well be a part of similar mecha nism.
12.
ACKNOWLEDGMENTS Plasmids pYc2 and pYc8 were a generous gift of Doctors E. Jean Finnegan and Elizabeth S. Dennis (CSIRO, Plant Industry, Canberra, Australia); the copperinducible vector system for plant transforma tion was a generous gift of Doctors Vadim L. Mett and Paul H.S. Reynolds (Horticulture and Food Research Institute of New Zealand). This work was supported by Russian Foundation of Fundamental Research (Grant 050448071); and Leading Scientific Schools of Russia (Grant NSH – 3444.2008.4).
13.
14.
15. 16.
REFERENCES 1. Vanyushin, B.F. and Ashapkin, V.V., Peculiarities of DNA Methylation in Plants, in: Progress in DNA Methylation Research, New York: Nova Science, 2007, pp. 1–90. 2. Vanyushin, B.F. and Ashapkin, V.V., DNA Methylation in Plants, New York: Nova Science, 2009. 3. Finnegan, E.J. and Kovac, K.A., Plant DNA Methyl transferases, Plant. Mol. Biol., 2000, vol. 43, pp. 189–210. 4. Henikoff, S. and Comai, L., A DNA Methyltransferase Homologous with a Chromodomain Exists in Multiple Polymorphic Forms in Arabidopsis, Genetics, 1998, vol. 149, pp. 307–318. 5. Callebaut, I., Courvalin, J.C., and Mornon, J.P., The BAH (BromoAdjacent Homology) Domain: A Link between DNA Methylation, Replication and Transcriptional Regu lation, FEBS Lett., 1999, vol. 446, pp. 189–193. 6. Kishimoto, N., Sakai, H., Jackson, J., et al., Site Spec ificity of the Arabidopsis METI DNA Methyltrans ferase Demonstrated through Hypermethylation of the SUPERMAN Locus, Plant. Mol. Biol., 2001, vol. 46, pp. 171–183. 7. Bartee, L., Malagnac, F., and Bender, J., Arabidopsis Cmt3 Chromomethylase Mutations Block NonCG Methylation and Silencing of an Endogenous Gene, Genes Dev., 2001, vol. 15, pp. 1753–1758. 8. Lindroth, A.M., Cao, X., Jackson, J.P., et al., Require ment of Chromomethylase3 for Maintenance of CpXpG Methylation, Science, 2001, vol. 292, pp. 2077–2080. 9. Cao, X., Springer, N.M., Muszynskil, M.G., et al., Conserved Plant Genes with Similarity to Mammalian de novo DNA Methyltransferases, Proc. Natl. Acad. Sci. USA, 2000, vol. 97, pp. 4979–4984. 10. Cao, X. and Jacobsen, S.E., Role of the Arabidopsis DRM Methyltransferases in de novo DNA Methyla tion and Gene Silencing, Curr. Biol., 2002, vol. 12, pp. 1138–1144. 11. Cao, X., Aufsatz, W., Zilberman, D., et al., Role of the DRM and CMT3 Methyltransferases in RNA
17.
18.
19.
20.
21.
22.
23.
24. 25. 26.
Directed DNA Methylation, Curr. Biol., 2003, vol. 13, pp. 2212–2217. Wada, Y., Ohya, H., Yamaguchi, Y., et al., Preferential de novo Methylation of Cytosine Residues in Non CpG Sequences by a Domains Rearranged DNA Methyltransferase from Tobacco Plants, J. Biol. Chem., 2003, vol. 278, pp. 42386–42393. Cao, X. and Jacobsen, S.E., LocusSpecific Control of Asymmetric and CpNpG Methylation by the DRM and CMT3 Methyltransferase Genes, Proc. Natl. Acad. Sci. USA, 2002, vol. 99, suppl. 4, pp. 16491–16498. Cokus, S.J., Feng, S., Zhang, X., et al., Shotgun Bisul phite Sequencing of the Arabidopsis Genome Reveals DNA Methylation Patterning, Nature, 2008, vol. 452, pp. 215–219. Jacobsen, S.E., Sakai, H., Finnegan, E.J., et al., Ectopic Hypermethylation of FlowerSpecific Genes in Arabidopsis, Curr. Biol., 2000, vol. 10, pp. 179–186. Ashapkin, V.V., Kutueva, L.I., and Vanyushin, B.F., The Gene for Domains Rearranged Methyltransferase (DRM2) in Arabidopsis thaliana Plants Is Methylated at Both Cytosine and Adenine Residues, FEBS Lett., 2002, vol. 532, pp. 367–372. Zhang, X., Yazaki, J., Sundaresan, A., et al., Genome Wide HighResolution Mapping and Functional Analysis of DNA Methylation in Arabidopsis, Cell, 2006, vol. 126, pp. 1189–1201. Zilberman, D., Gehring, M., Tran, R.K., et al., Genome Wide Analysis of Arabidopsis thaliana DNA Methylation Uncovers an Interdependence between Methylation and Transcription, Nat. Genet., 2007, vol. 39, pp. 61–69. Kutueva, L.I., Ashapkin, V.V., and Vanyushin, B.F., The Methylation Pattern of a Cytosine DNAMethyl transferase Gene in Arabidopsis thaliana Plants, Bio chem. Mol. Biol. Int., 1996, vol. 40, pp. 347–353. Mett, V.L., Lochhead, L.P., and Reynolds, P.H.S., CopperControllable Gene Expression System for Whole Plants, Proc. Natl. Acad. Sci. USA, 1993, vol. 90, pp. 4567–4571. Mett, V.L., Podivinsky, E., Tennant, A.M., et al., A Sys tem for TissueSpecific CopperControllable Gene Expression in Transgenic Plants: NoduleSpecific Antisense of Asparate AminotransferaseP2, Transgen. Res., 1996, vol. 5, pp. 105–113. Finnegan, E.J. and Dennis, E.S., Isolation and Identi fication by Sequence Homology of a Putative Cytosine Methyltransferase from Arabidopsis thaliana, Nucleic Acids Res., 1993, vol. 21, pp. 2383–2388. Chomczynski, P. and Sacchi, N., SingleStep Method of RNA Isolation by Acid Guanidinium Thiocyanate– Phenol–Chloroform Extraction, Anal. Biochem., 1987, vol. 162, pp. 156–159. Sambrook, J., Fritsch, E.F., and Maniatis, T., Molecu lar Cloning: A Laboratory Manual, New York: Cold Spring Harbor Lab., 1989, 2ed. Lister, R., O’Malley, R.C., TontiFilippini, J., et al., Highly Integrated SingleBase Resolution Maps of the Epige nome in Arabidopsis, Cell, 2008, vol. 133, pp. 523–536. Henderson, I.R. and Jacobsen, S.E., Tandem Repeats Upstream of the Arabidopsis Endogene SDC Recruit NonCG DNA Methylation and Initiate siRNA Spreading, Genes Dev., 2008, vol. 22, pp. 1597–1606.
RUSSIAN JOURNAL OF GENETICS
Vol. 47
No. 3
2011