tebrate Skull (Oxford Univ. Press, Oxford, 1937), S. L. Flan, J. Zool. 225, 577 (1991), J. A. McClan, J. Morphol. 64 21 1 (1939); E. McCrady Jr., Am. Anat. Mem. No. 16 (1938); , E. G. Wever, C. W. Bray, J. Exp 2001. 75, 503 (1937). 8. S. W. Herr~ng, A m 2001. 33, 472 (1993) W. Maer, Gegenbaurs iWorpho1. Jahrb. 133, 123 (1 987). 9. T. Rowe, W. D Carlson, W. W. Bottorff Thrinax- odon: Digital Atlas of the Skull (Unv, of Texas Press, ed. 2, Austn, 1995), ava~labe on CD-ROM. 10 P Alberch, S. J. Gould, G. F. Oster. D. B. Wake, Paleobiology 5, 296 (1979); W. L. Fink, ibid. 8, 254 (1 982); A G. Kluge, in Ontogeny and Systematics , C J. Humphr~es, Ed, (Columb~a Univ. Press, New York, 1988), pp. 57-81, S. J. Gould, Onrogeny and Phylogeny (Belknap, Harvard Univ. Press, Cam- bridge, MA, 1977). 11. G. G. Slmpson, Am. J. Sci. 214, 259 (1927) 12. C. Walsh and C. L. Cepko, Science 255,434 (1992) 13. A. B. Butler. Brain Res Rev. 19, 66 (1 994); R. G Northcutt, Am. 2001. 24, 710 (1984); P. S. Ulnsk~, In The Ecology and Biology of Mammal-Like Reptiles, N. Hotton Ill, P. D. MacLean, J. J. Roth. E. C. Roth Eds. (Sm~thsonianInstitution Press, Washington, DC, 1986), pp. 149-171. 14. M. E. Desmond and A. G. Jacobson, Dev. Biol. 57, 188 (1 977); R Jelinek and T. Pex~eder, Folia Mar- phol. 17, 184 (1970); l. C. Nafiagas, J. Philipp. Is/. Med. Assoc. 5, 251 (1925); L. H. Weed, Contrib. Embiyol. 9, 425 (1920) 15. P. S. Ulnski, J. Comp. Neurol. 142, 33 (1971) 16. D. A. N. Hoyte, lnt. Rev. Gen. Exp. Zool 2, 345 (1966); D R. Carter, J. Biomech. 20, 1095 (1987), S. A. Newman and W. D. Comper, Development 11 0,l (1 990). 17. 0. R Langworthy, J. Comp. Neurol. 46, 201 (1 928); O. Larsel, E. McCrady Jr., A. A. Zmmermann, ibid. 63, 95 (1935). 18. R. A Lende, Science 141, 730 (1963) 19. H. J. Jerison. Evolution of the Brain and Intelligence (Academc Press. New York. 1973), In The Neocor- tex, B L. Findlay et a/. Eds. (Plenum, New York. 19901, pp 5-1 9; J. Hopson, 13/01, Reptil. 9, 39 (1 979). 20. J. Allman, Neuroscience 2, 257 (1 990). 21. J.-H Gao eta/., Science 272, 545 (1 996). 22. W. T. Thatch, H. P Goodkin, J. G Keat~ng, Annu. Rev. Neurosci. 15, 403 (1992). 23 B K Hall, Evolutionaiy Developmental Biology (Chapman and Hall, London, 1992); J. Hanken and B K. Hal, Eds., The Skull, vos. 1, 2, and 3 (Unv, of Chlcago Press, Ch~cago, IL, 1993). 27 March 1996; accepted 31 May I996 Demethylation-Induced Developmental (1 1). In one experiment, a total of nine primary transforma~its (TO generation) Pleiotropy in Arabidopsis were recovered; progeny tests indicated that SIX lines that were f~~rther characterized Michael J. Ronemus, Massimo Galbiati, Christine Ticknor, contained s~nele-~OCLIS transferred DNA (T- Jychian Chen, Stephen L. Dellaporta* The function of DNA methylation in higher plants was investigated by expression of a complementary DNA encoding a cytosine methyltransferase (METI) from Arabidopsis thaliana as an antisense RNA in transgenic plants. This expression resulted in a 34 to 71 percent reduction in total genomic cytosine methylation. Loss of methylation was ob- served in both repetitive DNA and single-copy gene sequences. Developmental effects included altered heterochrony, changes in meristem identity and organ number, and female sterility. Cytosine demethylation prolonged both vegetative and reproductive phases of development. These findings implicate DNA methylation in establishing or maintaining epigenetic developmental states in the meristem. DNA) insertions (1 2). Methylation patterns in repetitive DNA sequences were examined by Southern (DNA) hybridization (13). Genomic DNAs were digested with the isochizo~ners Hpa I1 or Msp I (Fig. 2, upper panels) and probed with a centromeric repeat or a 55 ribosomal DNA sequence; both repeats are lnethylated in wild-type genornic DNA (3). Hpa I1 digestion is inhibited if either cytosine in the CCGG target site IS methylated; Msp I call cleave C5'"CGG but not ""CCGG (1 4). W~th both , , probes, Hpa I1 digest~on revealed a high ex- tent of demethylation in three of six antisense Plant genomes contain relatively large the mouse gene (Fig. 1) (5, 6). The MET1 lines (Tr244, 246, and 248: designated amounts of the modified nucleotide gene represents one member of a small gene "strong") and in the dilml mutant. Three 5-methylcytosine (51nC) (1 ). Despite evi- family in Arabidopsis (5) that maps to posi- antisense lines (Tr242, 243, and 245: desig- dence im~licating cvtosine methvlation in tion 68.9 on chro~nosome5, nollallelic to nated "weak") showed near wild-tvve levels - , plant epigenetic phenomena, such as re- peat-induced gene silencing (RIGS), cosup- pression, and inactivation of transposable elements (2), the role of cytosine methyl- ation in plant developmental processes is not clear. In Ambidopsis, ddm (decrease In DNA methylation) mutants have been iso- lated with reduced levels of cytosine meth- ylation in repetitive DNA sequences, al- though these mutations do not result in any detectable change in DNA methvltrans- ferase enzyrnatic"activity (3, 4). Afier sev- eral generations of self-pollination, ddm mutants exhibit a slight delay (1.7 days) in flowering, altered leaf shape, and an in- crease in cauline leaf 11~11nber (4). To address the role of DNA methylation in plant development, we ~ ~ s e d an antisense strategy to interfere with METI, a DNA ~l~ethyltra~lsferase (MTase) gene of Arabi- dopsis, previously cloned by homology to M. J. Ronemus M. Galbiati, C. Tlcknor, S. L. Delapolta Department of B~ology, Yale Univers~ty, New Haven, CT 06520-8104, USA. J. Chen. Academia Sln~ca, Tapei 11529. Tawan. "To whom correspondence should be addressed. the ddml locus (7). The MET1 gene is expressed in seedling, vegetative, and floral tissues; in the inflorescence, expression is seen at highest levels in meristelnatic cells by in situ RNA hybr~dization (8). To inhib- it expression of the METl gene, we intro- d~~ced an antisense construct (Fig. I), con- sisting of a 4.3-kb MET1 cDNA in the antisense orie~ltatioll under the control of a constitutive viral promoter (CaMV 355) 19)- into Arabidobsis strain Columbia ( 10) , of methylation. Msp I digestion was Inore complete in strong antisense lines than in wild-type or weak antisense lines. These re- sults indicate that strong antisense lines con- tain substantial de~nethylatio~i of these repeated sequences at both C5"'CGG and ""CCGG sites (15). DNA methylation was examined at four single-copy gene sequences (16) (Fig. 2, low- er panels). Substantial demethylation of all fo~~r genes was seen onlv In the strong anti- , , by ~~obitcterium'-mediated transforma;ion senseUlines; the ddml mutant showed little Fig. 1. The predicted FOCI Ac~d~c MET7 gene product and antsense construct The predlcted gene product of the METl locus IS a 1534-amno acld pro te~n w~th a h~gh degree of homology to the mouse MTase part~cuary In the catalfl~c and NH,-term1 Eco RI Eco RI nal foc~ target~ng do malns (5 30) The METl ant~sense construct IS shown below A 4 3-kb MET1 cDNA spannlng the postons ~ndcated was Inserted In the reverse orlentailon wth respect to the CaMV 35s promoter In the pMON530 T-DNA vector (9) 654 SCIENCE VOL 273 2 AUGUST 1996