Biol. Chem., Vol. 377, pp. 605 - 610, October 1996 · Copyright © by Walter de Gruyter & Co · Berlin · New York

Review Genetic Analysis of Genomic Methylation Patterns in Plants and Mammals

Jeffrey A. Yoder and Timothy H. Bestor* (1975) pointed out that heritable methylation patterns Department of Genetics and Development, College of could allow clonal transmission of states of gene expres- Physicians and Surgeons of Columbia University, sion during development; programmed changes in meth- 701 W. 168th St., New York, NY 10032, USA ylation patterns would allow expression (or establish re- pression) of the appropriate sets of genes during cellular * Corresponding author differentiation. In the years since those proposals were made, two important predictions were tested and con- firmed: promoter methylation does repress transcription, While it is now accepted that methylation of cytosine and methylation patterns are indeed subject to clonal in- residues plays a role in various epigenetic phenomena heritance (reviewed in Jost and Saluz, 1993). Furthermore, in mammals and flowering plants, the involvement of many tissue-specific genes were shown to undergo de- methylation patterns in the regulation of normal dev- methylation at the time of activation, or to be differentially elopment has remained a controversial and essenti- methylated in expressing and non-expressing tissues. ally untested issue in the 20 years since such a role was It was unfortunate that most of the data that bore on the first proposed. Antisense suppression of a DMA meth- question of methylation and development were fragmen- yltransferase in Arabidopsis and characterization of tary and correlative and therefore incapable of proving methylation-defective mutants of Arabidopsis have (or disproving) causation. It was also noticed by many shown that perturbations of methylation patterns dis- that small-genome metazoa such as Drosophila and rupt the development of plants, and targeted mutation Caenorhabditis elegans develop in the absence of detect- of the murine gene that encodes the one known form able modified bases in their DNA. The significance of DNA of DNA methyltransferase has shown that methylation modification in developmental gene control has often, and is required for cellular differentiation, genomic imprin- with good reason, been called into question. ting, and X chromosome inactivation in mammals. Rigorous tests of causation require genetic analysis; if Ectopic expression of homeotic genes and homeotic informative developmental abnormalities were to result transformations of floral organs in methylation-defec- from mutations in genes whose sole function lay in the tive plants suggest that (in plants and perhaps mam- establishment or maintenance of methylation patterns, a mals) heritable methylation patterns reinforce and causative role in development might be inferred. Such may have supplanted heritable gene control mediated data may be emerging from antisense experiments and by chromosomal proteins of the Polycomb and tri- characterization of methylation mutants in the flowering thorax groups. It is also possible that the developmen- plant Arabidopsis thaliana. Intensive studies of mouse tal abnormalities are the result of ectopic gene expres- strains that bear targeted mutations in the gene for the one sion caused by activation of transcription from nearby known form of mammalian DNA methyltransferase have parasitic sequence elements that are normally repres- shown that methylation patterns are involved in a number sed by methylation. Application of modern methods of of prominent epigenetic phenomena. The results of these genetic analysis promises to give definite answers to genetic studies confirm that methylation patterns perform long-standing questions as to the roles and signifi- essential roles, but have not yet revealed whether the pri- cance of genomic methylation patterns in normal dev- mary function of methylation patterns is developmental elopment and genome defense. gene control or defense against parasitic sequence ele- Key words: Antisense / Development / DNA methyltrans- ments. ferase / 5-methylcytosine. Demethylation and Arabidopsis Development

Introduction Richards and colleagues mutagenized>Arao/cfops/s with ethylmethane sulfonate and screened selfed progeny The existence of m5C in DNA of mammals and flowering for demethylation of centromeric repeat sequences, plants has been known for 50 years, and hypotheses as to which are normally heavily methylated in plants and mam- the function of m5C have proliferated in the 21 years that mals. Several demethylating mutant lines were recovered have elapsed since Riggs (1975) and Holliday and Pugh and termed ddm (for decreased DNA methylation). All mu- 606 JA YoderandT.H. Bestor

Discriminates unmethylated tations were recessive and fell into two complementation and hemimethylated CpG sites groups, DDM1 and DDM2 (Vongs et a/., 1993). DDM1 has been carefully characterized and found to display a num- ber of remarkable characteristics. Developmental defects were seen in ddml homozygotes; these included reduced Nuclear localization (GK) repeats Catalytic center apical dominance (branching produced two stems rather than a shoot and a stem, so that the plants took on a stunt- Fig. 1 Organization of DMA Methyltransferases of Higher Eukaryotes. ed, bushy appearance), partially fused carpals and other The organization of MET1 from Arabidopsis is similar to that floral abnormalities, and sterility or reduced fertility. The of DMA methyltransferases from mammals, chicken, and sea phenotypes were highly unusual in that they were variable urchin, except that the Zn is not discernible in the but became more severe with each meiotic generation, plant . The protein comprises a C-terminal catalytic do- and unique in that the phenotypes persisted after the re- main that is closely related to bacterial (cytosine-5)-methyltrans- cessive mutation had been removed by backcrossing to ferases; this is joined by a lysine/glycine repeat motif to a large N- the wildtype parental strain. The phenotypes were so per- terminal domain that has roles in nuclear import, targeting to replication foci during S phase (Leonhardt etal., 1992), and the in- sistent as to allow identification and mapping of the loci hibition of de novo methylation (Bestor, 1992). AZn-binding site is (unlinked to DDM1) that were responsible for different as- shared with the mammalian homologue of Drosophila trithorax pects of the phenotype (E. Richards, personal communi- (Bestor and Verdine, 1994) and there is a sequence related to cation). chicken Polybromo-1 (GenBank X90849). These two proteins are The progressively increasing severity of the phenotypes related to Drosophila proteins that are involved in the clonal pro- was associated with progressive demethylation of the ge- pagation of states of gene activity. nome; early passage plants showed conspicuous de- methylation only of highly reiterated sequences, while single-copy sequences became demethylated at later formations of floral organs; these included transforma- passages. While fully recessive when maintained with a tions of sepals to petals, sepals to carpels, and stamens to wildtype DDM1 allele, removal oiddml by backcrossing petals. Several different homeotic transformations were to the parental strain did not cause prompt reversion of observed within the flowers of a single plant, and a con- phenotype; the plants displayed characteristic ddml phe- sistent feature of the results of the antisense experiments, notypes for several generations and phenotypic reversion and those of the ddml studies, was the pronounced and restoration of normal m5C levels occurred only after pleiotropy of the phenotypes. Like ddml plants, the anti- several generations in the absence of ddml. It is appa- sense plants showed increasingly severe phenotypes with rently unprecedented that a recessive allele should pro- each generation of selfing, and removal of the transgene duce a phenotype that persists for several meiotic genera- by backcrossing did not erase the phenotype until several tions after removal of the mutant allele. generations had elapsed. The of the DDM1 gene has not yet been identi- fied. Enzyme assays and biochemical tests have indicated that extractable DMA methyltransferase activity and S- Does Cytosine Methylation Play a Role in the adenosyl L-methionine are not altered in ddml homozy- Normal Development of Plants? gotes (Kakutani etal., 1995). DDM1 may encode a factor that regulates DMA methyltransferase, or a chromosomal Studies of ddml and of DNA methyltransferase antisense protein which alters the accessibility of target sites. DDM1 plants have shown that characteristic developmental de- has been mapped to a small interval of distal chromosome fects result from perturbations of methylation patterns. 5, which will aid its isolation by positional cloning meth- These data have been interpreted to mean that pro- ods. grammed changes in methylation patterns are involved in The laboratories of Finnegan (Finnegan etal., 1996) and plant development, and methylation gradients through Dellaporta (Ronemus et a/., 1996) have used antisense elements of the plant were proposed to be involved in me- technology to suppress DMA methyltransferase in Ara- ristem determinancy (Ronemus etal., 1996). This interpre- bidopsis] the enzyme targeted in these studies (MET1; tation is consistent with the original predictions of Riggs Finnegan and Dennis, 1993) had been cloned by similarity (1975) and Holiday and Pugh (1975). Finnegan etal. (1996) with the mammalian enzyme (Dnmt; Bestor ef a/., 1988), to found that floral homeotic genes were ectopically expres- which it is closely related in sequence and domain or- sed in leaf tissue, and that the DNA methyltransferase an- ganization (Figure 1). The antisense constructs produced tisense phenotype resembled that of loss-of-f unction mu- phenotypes that resembled those seen in ddml homozy- tations at the CURLY LEAF (CLF) locus. CLF is a member gotes, but were more severe and had some additional fea- of the Polycomb group of genes, which in diptera and tures. Both antisense studies produced plants in which mammals are involved in the regulation of homeotic gene flower placement and branching patterns were abnormal. expression. Polycomb proteins (and those of a second Both groups observed a loss of apical dominance, stunt- group known as the trithorax group) form complexes on ing, and delayed flowering, as in ddml homozygotes. DNA that are subject to mitotic inheritance (Paro, 1995), Finnegan etal. (1996) observed striking homeotic trans- and it has been suggested that heritable methylation pat- Genomic Methylation Patterns in Plants and Mammals 607 terns reinforce and may have largely supplanted regula- coat and internal organs, and such animals appear even tion by heritable protein-DNA complexes in higher eu- on certain inbred backgrounds. The similarity of this case karyotes (Bestor and Verdine, 1994). This proposal is sup- to the ddm 1 and antisense studies in plants is striking, and ported by the observation of sequence motifs shared by is a reminder that developmental defects can result from eukaryotic DMA methyltransferases and members of the the epigenetic activation of parasitic sequence elements, Polycomb/trithorax group (Figure 1; Bestor, 1995) and which in turn activate nearby cellular genes. The genomes predicts a specific role for methylation patterns: suppres- of both plants and mammals contain tens of thousands of sion of transcription of genes that have spatially-restricted parasitic sequence elements, most of which contain au- patterns of expression. Notice that this proposed Poly- tonomous promoters and/or enhancers (Charlesworth comb-like role is limited to stabilization of the repressed era/., 1994). state, and no statement as to causation can be made. It can be argued that the developmental abnormalities While the extant data are consistent with a develop- that ensue in plants after disruption of methylation pat- mental role, there is an alternative interpretation of the re- terns do not settle the old question of methylation and de- sults which does not posit a role in normal development. velopment; either of the two salient hypothetical functions The genomes of flowering plants (and of most higher eu- of methylation (developmental gene control and genome karyotes) contain large numbers of transposable elements defense) are compatible with the available experimental and proviruses which contain promoters and enhancers evidence. that are normally silenced by cytosine methylation (Bestor, 1990; Bestor and Coxon, 1993; Bestor and Tycko, 1996; Martienssen and Richards, 1995). Demethylation and ac- Perturbations of Methylation Patterns and tivation of the promoters of parasitic sequence elements Mammalian Development can drive ectopic expression of nearby genes (Figure 2; Michaud ef a/., 1994); ectopic activation of genes that af- Jaenisch, Li, and colleagues have used homologous re- fect development decisions will naturally cause develop- combination in embryonic stem (ES) cells to construct a mental abnormalities. Several mutations in mice have re- series of mutations of increasing severity at the murine cently been found to be caused by insertion of retroviruses Dnmt gene on proximal chromosome 9, which encodes of the intracisternal A particle (IAP) class; enhancers and the one known form of DMA methyltransferase. All alleles promoters within the viral LTR can drive high level expres- are embryonic lethals; the least severe allele (a hypomor- sion of adjacent genes. All the retrovirus insertion alleles phic allele that produces a near-wildtype enzyme at 5% of show highly variable pleiotropic phenotypes, and most normal levels; Li ef a/., 1992,1993a) prevents homozygous show imprinted expression. The IAP insertion allele at mutant embryos from developing past the 25 somite stage agouti that has been characterized in most detail, Aiapy, is characteristic of 9.5 day embryos. Two more severe alleles controlled by methylation (Michaud etal., 1994); when the halt development at the presomite stage, or 8 - 8.5 days (Li retroviral DMA is unmethylated the viral promoter causes etal., 1993a; Tucker ef a/., 1996). Very large maternal sto- the agouti gene to expressed at high levels, and the mice res of DNA methyltransferase in the oocyte (Carlson etal., are grossly obese, diabetic, prone to a variety of tumors, 1992) prevent the enzyme from becoming limiting until day and their coats are lemon-yellow in color. When the retro- 6 or 7. The embryos die when m5C levels decline to ~ 30% viral DMA is methylated the agouti gene is expressed nor- of wildtype. The early lethality of the Dnmt mutations in mally and the animals are of wildtype phenotype. Many mice restricts observations to a short period of embryonic animals show variegated expression of A'apy in both the development.

B Methylation patterns regulate Methylation represses gene expression during development parasitic sequence elements

CH3 CH, CH3

I Programmed demethylation or Forced demethylation in Forced demethylation in ddml and antisense plants w ddm I and antisense plants Υ

Fig. 2 Ectopic Gene Activation by Unscheduled Demethylation or by Proximity to an Active Parasitic Sequence. The two salient hypothetical functions of methylation patterns are developmental gene control and suppression of parasitic sequence elements. The Figure shows that developmental abnormalities resulting from forced demethylation in plants are compatible with either hypothesis. In (A) demethylation of a gene that affects development leads to its premature activation; this hypothesis requires that demethylation be necessary for stage-specific expression of developmental genes. In (B) demethylation has caused the activation of a transposable element, whose enhancers activate transcription of nearby genes. This hypothesis does not require that methylation patterns be involved in the normal regulation of cellular genes. The lethality of mutations at Dnmt have obviated similar experiments in mice, but the hypotheses are valid for both plants and mammals. 608 J.A. YoderandT.H. Bestor

Despite the small developmental window, remarkable Demethylation and Remethylation in Plants and findings have been made by the Jaenisch and Li laborato- Mammals ries with Dnmt mutant mouse embryos. First, the mutation is lethal, thereby demonstrating that while cytosine meth- A remarkable finding in the ddml and DMA methyltrans- ylation may be dispensable to flies and nematodes it is ferase antisense studies was the low rate of de novo essential for mammals (Li et a/., 1992). Second, three im- methylation \nArabidopsis. Once reduced, m5C levels re- printed genes that are normally expressed from only one mained very low and were not restored several genera- allele (the maternal or paternal allele) are either expressed tions after the ddm 1 mutation or the antisense constructs from both alleles or from neither (Li et al., 1993b). This con- were removed by segregation. This is in sharp contrast to firmed that allele-specific methylation underlies the phe- the current picture in mammals (Monk ef al., 1987; Chaillet nomenon of genomic imprinting. Third, the Xist gene, ef a/., 1991), where the genome is thought to be largely de- which is normally expressed only from the inactive X chro- methylated at two developmental stages (the primordial mosome, is expressed on all X chromosomes in embryos germ cell and the blastocyst). Waves of de novo methyla- that have reduced methylation levels (Beard ef a/., 1995). tion are believed to establish new methylation pattern du- Xist is known to be involved in X inactivation, and it is be- ring gametogenesis and early postimplantation develop- lieved that demethylation causes the cell to inactivate ment. This is supported by the knock-in study that was both X chromosomes in females and the sole X chromo- described previously; Dnmt mutant ES cells regained some in males. Fourth, EScellshomozygousforanyofthe much of their genomic m5C upon introduction of a Dnmt Dnmt mutations grew normally as undifferentiated ES cDNA (Tucker ef a/., 1996). Either plants and mammals use cells but underwent apoptotic cell death when induced to very different means of establishing methylation patterns, differentiate (Li et a/., 1992). In the case of the two more or the demethylation and remethylation thought to occur severe alleles, no differentiated derivatives are seen, while during mammalian reproduction are less sweeping than ES cells homozygous for the least severe allele were capa- believed. There are in fact a number of cases in mice ble of some differentiation (Tucker ef a/., 1996). and humans in which methylation patterns have been ob- Jaenisch, Li, and colleagues used a novel knock-in pro- served to be transmitted through the germline (Allen ef al., cedure to insert a DMA methyltransferase cDNA within one 1990; Silva and White, 1988; Tremblay ef a/., 1995), as is of the mutant alleles in homozygous Dnmt mutant ES cells the case in plants. However, the extant data suggest that (Tucker ef a/., 1996). Chimaeric mice derived from the methylation patterns are less dynamic in plants than in knock-in ES cells developed normally, and it was found mammals. that centromeric satellite DMA and sequences related to Moloney MuLV had undergone de novo methylation. How- ever, methylation at imprinted alleles was only partial and How Many DNA Methyltransferases? was not complete until the genes had been transmitted through the germ line. This led the authors to conclude One of the outstanding questions in the DNA methylation that the methylation status of imprinted genes could be field concerns the regulation of de novo methylation, and established only in the germline, while the methylation the most popular answer to this question invokes a family status of non-imprinted genes could be set up during of sequence-specific de novo methyltransferases that act embryonic development. However, repetitive sequences during gametogenesis and early postimplantation devel- rather than non-imprinted cellular genes were examined, opment (Jähner and Jaenisch, 1984). The methylation and the methylation status of the imprinted genes was as- patterns shaped by these developmentally-regulated en- sayed only at a few restriction sites. zymes would be maintained by the DNA methyltrans- The reasons for the lethality of mutations at Dnmt are ferase encoded by the Dnmt gene. The existence of sepa- not known, but ectopic X inactivation should be sufficient rate de novo and maintenance is well-accepted to kill cells, and the loss of imprinting may also contribute. but not at all well-proven; there is simply no direct evi- There may be additional, undiscovered defects induced dence for developmentally-regulated DNA methyltrans- by reduced methylation. The inability of Dnmt mutant ES ferases of the predicted character. Furthermore, Dnmt has cells to differentiate removed the opportunity to observe both de novo and maintenance activity (Bestor, 1992). developmental abnormalities, while the relative tolerance Jaenisch, Li, and colleagues found that atargeted deletion of plants to perturbations of methylation patterns made of an exon that encodes the catalytic center of Dnmt al- developmental effects conspicuous and informative. While lows homozygous mutant cells to maintain small amounts extremely useful in many ways, early lethality reduced the of m5C in their DNA, which does require that at least one utility of Dnmt mutant mice in addressing the long-stand- additional enzyme exist in mammals (cited in Tucker ef al., ing question of the role of programmed methylation and 1996). However, an activity in lysates of the Dnmt null mu- demethylation in the control of mammalian development. tant cells was found to catalyze methyl transfer to the arti- ficial poly d(l-C), and therefore cannot be highly sequence-specific. Results of genetic studies suggest that there is a single key DNA methyltransferase in plants. Plants are known to Genomic Methylation Patterns in Plants and Mammals 609 encode two or three enzymes that can be identified as Beard, C., Li, E., and Jaenisch, R. (1995). Loss of methylation ac- DMA methyltransferases on the basis of sequence analy- tivates Xist in somatic but not embryonic cells. Genes & Devel. sis (Finnegan and Dennis, 1993). The antisense experi- 9,2325-2334. ments targeted only one of these, MET1 (the enzyme most Bestor, T.H. (1990). DNA methylation: How a bacterial immune function has evolved into a regulator of gene expression and closely related to the product of the mammalian Dnmt 5 genome structure in higher eukaryotes. Phil. Trans. Royal Soc. gene). While the loss of m C from the DMA of the antisense Lond.B326,179-187. plants could have been due to a loss of either de novo or Bestor, T.H. (1992). Activation of mammalian DNA methyltrans- maintenance activity, the slow remethylation that follows ferase by cleavage of a Zn-binding regulatory domain. EMBO removal of the construct by segregation in sexual crosses J. 77,2611-2618. confirms that MET1 is the predominant agent in both Bestor, T.H. (1995). Biochemical basis of allele-specific gene ex- maintenance and de novo methylation, although it is clo- pression in genomic imprinting and X inactivation. In: Genomic sely related to what is believed to be the maintenance en- Imprinting: Causes and Consequences. R. Ohlsson, K. Hall, and M. Ritzen, eds. (Cambridge, UK: Cambridge University zyme of mammals. The other enzymes encoded by the Press). Arabidopsis genome must fulfill minor or specialized roles. Bestor, T.H., and Coxon, A. (1993). The pros and cons of DNA 5 It is not unlikely that the residual m C observed in the DMA methylation. Curr. Biol. 3,384 - 386. of Dnmt null ES cells (Tucker et a/., 1996) is the product of Bestor, T.H., and Tycko, B. (1996). Creation of genomic methyla- analogous mammalian enzymes. tion patterns. Nature Genetics 72,363 - 367. There is no compelling conceptual requirement for a Bestor, T.H., and Verdine, G.L. (1994). DNA methyltransferases. family of ofe novo methyltransferase. The separation of Curr. Op. Cell Biol. 6,380-389. Dnmt into regulatory and catalytic domains opens the way Bestor, T.H., Laudano, A., Mattaliano, R., and Ingram, V. (1988). Cloning and sequencing of a cDNA encoding DNA methyl- for regulation by sequence-specific accessory factors, of mouse cells. The carboxyl-terminal domain of and it remains quite possible that de novo methylation is the mammalian enzyme is related to bacterial restriction meth- cued not by the sequence context of CpG sites but by yltransferases. J. Mol. Biol. 203,971 -983. interactions of repeated sequences or certain types of Carlson, L.L., Page, A.W., and Bestor, T.H. (1992). Localization structural features in DMA (Bestor and Tycko, 1996). and properties of DNA methyltransferase in preimplantation Specific chromosomal proteins may also trigger de novo mouse embryos: Implications for genomic imprinting. Genes & methylation of CpG dinucleotides in their vicinity. So little Devel. 6,2536-2541. is known of the regulation of de novo methylation that Chaillet, J.R., Vogt,T.R, Beier, D.R., and Leder, P. (1991). Parental- specific methylation of an imprinted transgene is established few candidate mechanisms can be excluded. However, it during gametogenesis and progressively changes during de- seems that acceptance of the central role of a family of se- velopment. Cell 66,77 - 83. quence-specific, developmentally-regulated DNA meth- Charlesworth, B., Sniegowski, P., and Stephan, W. (1994). The yltransferases should not precede the identification and evolutionary dynamics of repetitive DNA in eukaryotes. Nature isolation of at least one such enzyme. 377,215-220. Finnegan, E.J., and Dennis, E.S. (1993). Isolation and identifica- tion by sequence homology of a putative cytosine methyltrans- ferase from Arabidopsis thaliana. Nucleic Acids Res. 2 7,2383 - Conclusions 2388. Finnegan, E.J., Peacock, W.J., and Dennis, E.S. (1996). Reduced It has been said that biochemistry cannot prove causation, DNA methylation in Arabidopsis thaliana results in abnormal and that genetics cannot prove mechanism. Both genetic plant development. Proc. Nat. Acad. Sei. USA93,8449-8454. and biochemical methods are being applied to answer the Holliday, R., and Pugh, J.E. (1975). DNA modification mecha- question of the biological roles of methylation patterns in nisms and gene activity during development. Science 787, higher eukaryotes, and answers to old questions can at 226-232. Jähner, D., and Jaenisch, R. (1984). DNA methylation in early last be seen to be taking form. mammalian development. In: DNA Methylation, A. Razin, H. Cedar, and A. Riggs, eds. (Berlin, Germany: Springer-Verlag). Jost, J.-P, and Saluz, H.-P. (1993). DNA Methylation: Molecular Acknowledgements Biology and Biological Significance. (Basel, Switzerland: Birkhäuser). We thank Drs. Jean Finnegan, Rudolf Jaenisch, En Li, and Eric Kakutani, T, Jeddeloh, JA, and Richards, E.J. (1995). Richards for discussions and communication of unpublished Characterization of an Arabidopsis thaliana DNA hypomethy- data. Supported by NIH grants GM00616, CA60610, and lation mutant. Nucleic Acids Res. 2,130 -137. AI40021toT.H.B. Leonhardt, H., Page, A.W., Weier, H.-UI., and Bestor, T.H. (1992). A targeting sequence directs DNA methyltransferase to sites of DNA replication in mammalian nuclei. Cell 77,865 - 874. Li, E., Beard, C., Forster, A.C., Bestor, T.H., and Jaenisch, R. References (1993a). DNA methylation, genomic imprinting, and mamma- lian development. Cold Spring Harb. Symp. Quant. Biol., LVIII, Allen, N.D., Morris, M.L, and Surani, M.A. (1990). Epigenetic con- 297-305. trol of transgene expression and imprinting by genotype-spe- Li, E., Beard, C., and Jaenisch, R. (1993b). Role for DNA methyl- cific modifiers. Cell 61,853 - 861. ation in genomic imprinting. Nature 366,362-365. 610 JA YoderandT.H. Bester

Li, E., Bester, T.H., and Jaenisch, R. (1992). Targeted mutation of Ronemus, M.J., Galbiati, M., Ticknor, C., Chen, J., and Dellaporta the DNA methyltransferase gene results in embryonic lethality. S.L. (1996). Demethylation-induced developmental pleiotropy Cell 69,915-926. in Arabidopsis. Science 273,654 - 657. Martienssen, R.A., and Richards, E.J. (1995). DMA methylation in Silva, A.J., and White, R. (1988). Inheritance of allelic blueprints eukaryotes. Curr. Opin. Genet. Dev. 5,234 - 242. for methylation patterns. Cell 54,145 -152. Michaud, E., van Vugt, M., Bultman, S., Sweet, H., Davisson, M., Tremblay, K.D., Saam, J.R., Ingram, R.S., and Tilghman, S.M. and Woychik, R. (1994). Differential expression of a new domi- (1995). A paternal-specific methylation imprint marks the alle- nant agouti allele (A'apy) is correlated with methylation state and les of the mouse H19 gene. Nature genet. 9,407 - 413. is influenced by parental lineage. Genes & Devel. 8, 1463 - Tucker, K.L., Beard, C., Dausman, J., Jackson-Grusby, L, Laird, 1472. P.W., Lei, H., and Jaenisch, R. (1996). Germ-line passage is re- Monk, M., Boubelik, M., and Lehnert, S. (1987). Temporal and re- quired for establishment of methylation and expression pat- gional changes in DMA methylation in the embryonic, extraem- terns of imprinted but not of nonimprinted genes. Genes & bryonic and gem cell lineages during mouse embryo develop- Devel. 70,1008-1020. ment. Development 99,371 -382. Vongs, A., Kakutani, T, Martienssen, R.A., and Richards, E.J. Paro, R. (1995). Propagating memory of transcriptional states. (1993). Arabidopsis thaliana DNA methylation mutants. Trends Genet. 71,295 - 297. Science 260,1926-1928. Riggs, A.D. (1975). X inactivation, Differentiation, and DMA meth- ylation. Cytogenet. Cell Genet. 74,9-11. Biol. Chem., Vol. 377, pp. 611 -617, October 1996 · Copyright ©by Walter de Gruyter& Co-Berlin-New York

Review

Hydroxynitrile of Higher Plants

Harald Wajant1'* and Franz Effenberger2 During cyanogenesis, cyanohydrins are decomposed 1 to sugar, HCN and a carbonyl compound by a two step 1nstitute of Cell Biology and Immunology, Allmandring 31, mechanism. First the glycoside is cleaved in an occa- and 2 sionally specific ß-glucosidase mediated reaction into Institute of Organic Chemistry, Pfaffenwaldring 55, carbohydrate and the corresponding a-hydroxynitrile. University of Stuttgart, D-70569 Stuttgart, Germany Subsequently the a-hydroxynitrile decomposes to the appropriate carbonyl compound and HCN. The latter step * Corresponding author occurs (base-catalyzed) spontaneously or enzymatically by action of an -hydroxynitrile (HNL) (Conn, 1981). Both products of decomposition of -hydroxynitriles, as Release of HCN from cyanogenic glycosides is due to well as the compound on its own, have a toxic potential to the cleavage of the carbohydrate moiety by ß-glucosi- organisms suggesting that the degradation of cyanogenic dases to yield the corresponding a-hydroxynitrile, glycosides acts as a defense mechanism against herbivo- which dissociates spontaneously into HCN and a car- res (Nahrstedt, 1985). In some cases it has been shown bonyl compound, or by action of an a-hydroxynitrile that cyanogenic glycosides can be metabolized to non- lyase (HNL). A short review of the regulation of the ca- cyanogenic compounds without liberation of HCN to the tabolism of cyanogenic glycosides during cyanogene- environment. The existence of this non-cyanogenic meta- sis and germination of cyanogenic plants is given. The bolic pathway suggests an additional role for cyanogenic major biochemical properties of HNLs purified from glycosides as N-storage compounds (Lieberei et a/., 1985; various species of higher plants are summarized. Selmaref a/., 1988; Swain and Poulton, 1994). Thereafter the phylogenetic relationship, molecular Free HCN can usually only occur upon tissue disrup- structure and catalytic mechanism of these enzymes tion; raising the question: how can cyanogenesis be pre- are discussed. Finally we give an overview of recent vented in the healthy, undamaged plant? Two simple progress in the use of HNLs as biocatalystsforthe syn- mechanisms seem conceivable: first, compartmentaliza- thesis of optically active cx-hydroxynitriles which are tion of substrates (cyanogenic glycoside and a-hydroxy- important building blocks in the fine chemical and nitrile, and catabolizing enzymes (ß-glucosidase, HNL), or pharmaceutical industries. secondary inhibition of these enzymes by endogenous Key words: Biocatalyst / Cyanogenesis / Cyanohydrins / factors. In all cyanogenic plants investigated to date, com- Hydroxynitrile. partmentalization at the tissue or subcellular level seems to be the means of choice to prevent cyanogenesis. In the seeds of rosaceous stone fruits which are highly cyanoge- Introduction nic, the substrates amygdalin and are located within the parenchyma cells of the cotyledons whereas the The release of HCN (cyanogenesis) is not only widely dis- catabolizing enzymes amygdalin , prunasin hy- tributed in higher plants (Spermatophyta), including im- drolase and mandelonitrile lyase are exclusively found in portant food plants like cassava and sorghum, but is in ad- the procambium (Swain and Poulton), 1994; Swain et a/., dition found in several species of ferns, bacteria, fungi and 1992; Wu and Poulton, 1991). Hence, it seems that cyano- insects (Conn, 1981). Of course, the highly toxic HCN genesis in undamaged seeds is prevented by compart- does not occur free in plants but is stored as cyanogenic mentalization on the tissue level. However, in developing precursors. These precursors are usually a-hydroxyni- seedlings of sorghum, which is taken as a cyanogenic mo- triles which are stabilized by O-ß-glycosidic linkage to del organism, compartmentalization occurs on the cellular saccharides (cyanogenic glycosides). Free hydroxynitri- as well as on the subcellular level; thus, dhurrin is restric- les (cyanohydrins) are fairly stable under acidic conditions ted to the vacuoles (Saunders and Conn, 1977) of epider- but rather unstable at pH values above 5. The sugar moiety mal cells while dhurrinase and HNL are located in the me- in cyanogenic glycosides is usually D-glucose, but di-, tri- sophyll tissue associated with the chloroplast or restricted and tetraglycosides have also been described. Typically to the cytoplasm, respectively (Kojima ef a/., 1979; Thayer R1 and R2 are dissimilar leading to a chiral carbinol carbon and Conn, 1981). oftheaglycon(Seigler, 1991). For several combinations of R1 and R2 both epimers are found in nature, but such forms do not usually occur in the same species (Seigler, 1991). 612 H. Wajant and R Effenberger

Flavoprotein HNLs Nonflavoprotein HNLs

HNLs have been purified and characterized from almost Nonflavoprotein HNLs have been isolated from three fa- a dozen cyanogenic plants. Traditionally, HNL has been milies of dicotyledons (Linaceae, Euphorbiaceae, Ola- divided into two groups based on FAD content. The flavo- caceae) (Hughes et a/., 1994; Kuroki and Conn, 1989; protein HNLs, all isolated from species of the rosaceae Wajant and F rster, 1996; Xu ei a/., 1988), one family of family, have (R)-mandelonitrile as natural substrate and monocotyledones (Gramineae) (Bove and Conn, 1961) are monomeric glycoproteins with a molecular mass of and one family of ferns (Polypodiaceae) (Wajant et a/., 58 - 66 kDa (HNLs from Prunoideae) and 75 - 82 kDa 1995b). These enzymes require between 100- and 1500- (Maloideae) (Poulton, 1988). HNLs are major fold purification to reach homogeneity. In contrast to the proteins of the rosaceous stone fruits and need only flavoprotein group of HNLs, the nonflavoprotein HNLs 5-50 fold purification to reach homogeneity (Table 1), comprise enzymes forming a rather heterogeneous group (Poulton, 1988). Their overall similarity in biochemical pro- regarding their biochemical properties (Table 1). The perties and their confined occurence to a defined sys- thoroughly investigated HNL from Sorghum bicolar tematic group of plants (Gerstner et al., 1968) indicate that (SbHNL) has (S)-p-hydroxy-mandelonitrile as its natural flavoprotein HNLs are a narrow phylogenetically related substrate and is a heterotetrameric glycoprotein of 110 class of enzymes. This evidence is also supported by im- kDa (Smitskamp-Wilms ef a/., 1991; Jansen ef a/., 1992; munological data which shows a high degree of serologi- Wajant and Mundry, 1993). The HNL from Ximenia ameri- cal cross-reactivity between various flavoprotein HNLs cana (XaHNL) converts (S)-mandelonitrile and is a mono- and their cognate antisera (Gerstner und Pfeil, 1972). meric glycoprotein of 38 kDa (Kuroki et a/., 1989). The he- Despite possessing a flavo prosthetic group, flavoprotein terogeneity of the nonflavoprotein group of HNLs can be HNLs, like other HNLs, do not catalyze a net reduction- strikingly clearly shown in the acetone cyanohydrin lyase oxidation reaction, raising the question of the functional (ACL) subgroup of HNLs. While the ACLs from Linum usi- role of FAD in this group of enzymes. Cleavage of the tatissimum (LuACLor LuHNL) accept various aliphatic (R)- holoenzyme of HNL from almond (PaHNL) at acidic pH cyanohydrins as their substrate (Albrecht ef a/., 1993), the and subsequent reconstitution of apoenzyme and FAD two narrowly related ACLs from Manihot esculenta correlates with a reversible loss of enzyme activity sug- (MeACL or MeHNL) and Hevea brasiliensis (HbACL or gesting that FAD is required for maintenance of the overall HbHNL) convert exclusively (S)-cyanohydrins (Wajant conformational structure of the active molecule (Becker et ef a/., 1995a; F rster ef a/., 1996; Klempier ef a/., 1993). a/., 1963, B rwald and Jaenicke, 1978; Jorns, 1979). These enzymes are also clearly different concerning mo- lecular weight and subunit composition. Moreover, using specific antisera against each of these ACLs no cross re- activity can be observed (Table 2), (Wajant and F rster,

Table 1 Properties of HNLs.

Enzyme source Natural substrate Kinetic properties R/S Molecular mass FAD Carbo- Reference speci- (kDa) hy-

•Fi^NI*·«/ /-Ι^Ë + Ë

KM(ITIM) l/max Native Subunit (μίτιοΙ/Γηίη/ΓηΙ)

Prunus sp. (R)-mandelonitrile 0.093 - R 55-80 55-80 yes yes Xu, eta/., 1986 (Rosaceae) Sorghum bicolor (S)-p- 0.55 S 105 33 no yes Bove and (Gramineae) hydroxy- and Conn, 1961 mandelonitrile 22 Linum acetone 2.5 1.1 R 82 42 no no Xu,efa/.,1988 usitatissimum cyanohydrin (Linaceae) Manihot esculenta acetone 105-119 - S 92-124 28-30 no no Hughes ef a/., (Euphorbiaceae) cyanohydrin 1994 Hevea brasiliensis acetone 115 S 105 30 no no Wajant and (Euphorbiaceae) cyanohydrin F rster, 1996 Ximenia americana (S)- 0.28 S 38 38 no yes Kuroki and (Olacaceae) mandelonitrile Conn, 1989 Phlebodium (R)- 0.83 60.1 R 168 20 no no Wajant ef a/., aureum mandelonitrile 1995a (Polypodiaceae) Hydroxynitrile Lyases of Higher Plants 613

Table 2 Serological Cross-Reactivity of HNLs (Wajant and various phylogenetically distinct groups of molecules. Förster, 1996; Wajant et a/., 1995a). According to the biochemical and serological data, we

PaHNL SbHNL LuHNL MeHNL HbHNL PhaHNL propose that each of the following HNLs define the proto- type of a distinct group of HNLs: SbHNL, LuHNL, MeHNL anti PaHNL +++----- (together with HbHNL), XaHNL, PhaHNL and PaHNL (to- anti SbHNL -+++---- gether with other rosaceous HNLs). However, to decide anti LuHNL - - +++ - anti MeHNL ---+++ +++ whether these groups result from convergent or divergent anti HbHNL - +++ +++ molecular evolution, the primary and three dimensional anti PhaHNL ----- +++ structure have to be compared.

1996). Recently, Wajant et a/. (1995b) described a man- Primary Structure of HNLs delonitrile lyase (MDL) from the fern Phlebodium aureum (PhaHNL). Like the MDLs isolated from rosaceae species, In recent years the flavoprotein HNL from Prunus serotina this enzyme is an (R)-HNL. But in contrast to the rosa- (Cheng and Poulton, 1993) and the nonflavoprotein HNLs ceous enzymes, this MDL is not a flavoprotein. It is also from Sorghum bicolor (Wajant ef a/., 1994), Manihot escu· clearly different in size and subunit composition and again lenta (Hughes eta/., 1994), Hevea brasiliensis (Hasslacher there is no serological relationship to other HNLs (Table 2), etal., 1996) and Linum usitassimum (own unpublished re- (Wajant et a/., 1995a). Taking into consideration the heter- sults) have been cloned and their primary sequence analy- ogeneity in biochemical properties descr. bed above and zed with regard to sequence homologies to other proteins. the lack of serological cross reactivity among most of the While HbHNL and MeHNL share 74% identity, there is no HNLs, the possibility arises that HNLs can be divided into sequence homology between them and any other cloned

Table 3 Synthesis of (R)-Cyanohydrins 2 by Enzyme Catalyzed Addition of HCN to 1. 0 (fl)-hydroxynitrile V" Rr-// . ur*M ·/*«··»Ivase f\ ' — C + HON — w \ R V ' H H CN 1 (R)-2

Aldehydes 1 / Cyanohydrins(R)-2

R = Conv. (%) ee(%) Source Reference

C6H5 98.9 n.d. Prunus amygdalus Wehtjeefa/.,1990 81 94 Prunus amygdalus Brusseeefa/., 1990 72 92 Prunus amygdalus Ognyanovefa/,,1991 Linum usitatissimum Albrecht ef a/., 1993 96 >99 Prunus amygdalus Effenberger,1994 82 99 Phlebodium aureum Wajant ef a/., 1995b

3,4-(CH2O2)C6H3 50 93 Prunus amygdalus Brusseeefa/., 1990 35 90 Prunus amygdalus Ognyanovefa/,,1991 60 99 Prunus amygdalus Jäger, 1992

2-thienyl 67 96 Prunus amygdalus Eichhorn, 1995 92 99 Phlebodium aureum Wajant ef a/., 1995b

n-C3H7 100 92 Prunus amygdalus Brusseeefa/., 1990 100 95 Prunus amygdalus Huuhtanen and Kanerva, 1 992 n.d. n.d. Linum usitatissimum Albrecht ef a/., 1993 98.6 98 Prunus amygdalus Effenberger,1994 42 55 Phlebodium aureum Wajant ef a/., 1995b

CH3CH=CH 99 69 Prunus amygdalus Brusseeefa/., 1990 n.d. a Linum usitatissimum Albrecht ef a/., 1993 68 97 Prunus amygdalus Effenberger,1994

cC6H11 96 0 Prunus amygdalus Brusseeefa/., 1990 72 96 Prunus amygdalus Ognyanovefa/,,1991 95 97 Prunus amygdalus Effenberger and Stelzer, 1 993

a n.d.;[a]D = - 614 H. Wajant and F. Effenberger

HNL. Moreover, there is no obvious homology among However, most intriguing were the homologies found for other HNLs. SbHNL. Analysis of the SbHNL revealed extensive homo- Comparison of the deduced aa sequence of PsHNL logies to serine carboxypeptidases which belong to the with other known sequences in various databases shows structurally well investigated group of á/β hydrolase fold moderate homologies to various , especially enzymes (Wajant et al., 1994). Interestingly, sites of func- to members of the dehydrogenase family. In particular tional importance in carboxypeptidases are preferably these homologies comprise a region with putative FAD conserved, e.g. N- motifs, and ac- binding properties (Cheng and Poulton, 1993). For HbHNL tive site residues. The of carboxypeptidases and MeHNL limited but significant homologies to two and other á/β is a which is also proteins of unknown function from rice have been found found in the subtilisin and chymotrypsin groups of serine (Hasslacher ei a/., 1996). The mRNA of these proteins ac- proteases as well as in cysteine proteases (Liao and cumulates concomitantly with the occurence of induced Remmington, 1990; Liao et al., 1992). Based on the com- resistance after leaf infiltration with Pseudomonas syrin- parison of the 3-dimensional structure of various á/β hy- gae pv. syringae suggesting a putative role of these pro- drolase fold enzymes, Ollis et al. (1992) proposed that teins in aquired resistance (Reimmann ei a/., 1995). Re- these enzymes have diverged from a common ancestor markably, the deduced aa sequence of LuHNL shows no and that they have seemingly evolved so as to conserve homology with the H N Ls from cassava and rubber tree de- the positions of key catalytic residues (divergent). spite having the same natural substrate (own unpublished Despite this fact, there is no overall similarity between results). This lack of sequence homology nicely matches this group of enzymes and other enzymes that have cata- with the aforesaid discrepancy in biochemical properties. lytic triads. In particular, considering that the overall fold of

Table 4 Synthesis of (S)-Cyanohydrins 2 by Enzyme Catalyzed Addition of HCN to Aldehydes 1. OH Ï (S)-hydroxynitrile Υ // lyase ^ R C + HCN R^ ^'" CN H Ç 1 (R)-2

Aldehydes 1 / Cyanohydrins (S)-2

R = Conv. (%) ee(%) Source Reference

C6H5 94 Hevea brasiliensis Klempiereia/.,1993 91 97 Sorghum bicolor Effenberger, 1994 34 90 Manihot esculenta (natural enzyme) Wajant ei a/., 1995a 100 98 Manihot esculenta (recombinant enzyme) F rsterei a/., 1996 67 >99 Hevea brasiliensis (recombinant enzyme) Schmidt ei a/., 1996

3-C6H50-C6H4 20 Hevea brasiliensis Klempiereia/.,1993 93 96 Sorghum bicolor Effenberger, 1994 41 23 Manihot esculenta (recombinant enzyme) Roos, 1995 9 99 Hevea brasiliensis (recombinant enzyme) Schmidt ei a/., 1996

3-HO-C6H4 97 91 Sorghum bicolor Effenberger, 1994 25 50 Manihot esculenta (recombinant enzyme) Roos, 1995

3-thienyl 85 97 Sorghum bicolor Effenberger, 1994 90 38 Manihot esculenta (natural enzyme) Wajant ei a/., 1995a 98 98 Manihot esculenta (recombinant enzyme) F rsterei a/., 1996 49 99 Hevea brasiliensis (recombinant enzyme) Schmidt ei a/., 1996

2-furyl 80 80 Sorghum bicolor Effenberger, 1994 71 88 Manihot esculenta (recombinant enzyme) Roos, 1995 55 98 Hevea brasiliensis (recombinant enzyme) Schmidt ei al. ,1996

(CH3)2CH 81 Hevea brasiliensis Klempiereia/., 1993 91 95 Manihot esculenta (recombinant enzyme) F rsterei a/., 1996

n-C3H7 80 Hevea brasiliensis Klempiereia/., 1993 92 90 Manihot esculenta (recombinant enzyme) Roos, 1995

cCeHll 100 92 Manihot esculenta (recombinant enzyme) F rsterei a/., 1996 94 99 Hevea brasiliensis (recombinant enzyme) Schmidt ei a/., 1996 Hydroxynitrile Lyases of Higher Plants 615 these enzymes, the order of the catalytic triad residues in mes was the preparation of (R)-mandelonitrile from benz- the primary sequence and the placement of residues for- and HCN, with emulsin as the source of the en- ming the are different, suggests that these zyme (Rosenthaler, 1908). A more general procedure for groups of enzymes have evolved independently from dif- the preparation of (R)-cyanohydrins (Becker et a/., 1965), ferent ancestral molecules (convergent evolution). Crys- using the isolated enzyme from bitter almonds (PaHNL), tallographic studies and the order of the catalytic triad re- resulted in fairly good optical yields for the natural sub- sidues in the primary sequence of SbHNL clearly indicate strate (86% ee), but for most unnatural sub- that this HNL is also an á/β hydrolase fold enzyme. strates the optical yields were low. A decisive improve- Mutational analysis of MeHNL (Wajant and Pfizenmaier, ment of optical yields for a wide variety of aldehydes as 1996; in press) and HbHNL (Hasslacher ei a/., 1996), as substrates was the discovery that the undesirable non well as inhibitor studies, suggests that these HNLs also enzymatic addition of HCN to aldehydes is suppressed in utilize a catalytic triad. Moreover, the order of the catalytic organic (Effenberger ef a/., 1987). Even for alde- triad residues in MeHNL speaks well for an á/β hydrolase hydes that are poor substrates for the enzyme, the enzy- fold structure of this enzyme. Mutational analysis and matic reaction predominates over the chemical addition crystallographic studies should reveal whether other and (R)-cyanohydrins with high ee values result (Table 3). HNLs are related by convergent or divergent molecular In Table 4, the application of various (S)-hydroxynitrile evolution to the á/β hydrolase fold group of enzymes. lyases for the synthesis of (S)-cyanohydrins is summari- zed. In the case of (S)-HNLs, the use of organic solvents also normally give better optical yields than the use of Application of HNLs in Stereoselective Organic water/alcohol mixtures as reaction medium. Syntheses Chiral cyanohydrins as á-substituted carboxylic acid derivatives have considerable synthetic potential for In recent years significant progress has been achieved in many important classes of biologically active compounds Stereoselective synthesis. The development and applica- with stereogenic centers. In Scheme 1 some examples of tions of chiral catalysts, as well as the application of enzy- the application of (R)-cyanohydrins in organic synthesis mes, were of particular interest. are presented. Most of the transformations shown in One of the first asymmetric syntheses effected by enzy- Scheme 1 occur without any racemization.

Me H NHMe Me

(fi)-adrenaline (fl)-pantolactone

OH OH H

R^CV'" H Ar I Me CN COOH L(-) (fl)-cyanohydrin (fl)-hydroxy carb- oxylic acids (Ar=Ph, R'=Me)

OS02X

CN (S)-

\/ C N NH H 2 (S)-aziridines (S)-amino acids Scheme 1 616 H. Wajant and F. Effenberger

References arbeit, Universität Stuttgart. Jansen, l., Woker, R., and Kula, M.-R. (1992). Purification and pro- Albrecht, J., Jansen, L, and Kula, M.-R. (1993). Improved purifica- tein characterization of hydroxynitrile lyases from Sorghum tion of an (R)-oxynitilasefrom Linum usitatissimum (flax) and in- and Almond. Biotechnol. Appl. Biochem. 75,90 - 99. vestigation of the substrate range. Biotechnol. Appl. Biochem. Jörns, M.S. (1979). Mechanism of by the flavoenzyme 77,191-203. oxynitrilase. J. Biol. Chem. 254,12145 -12152. Bärwald, K.-R., and Jaenicke, L (1978). D-Hydroxynitrile lyase: Klempier, N., Griengl, H., and Hayn, M. (1993). Aliphatic (S)- Involvement of the prosthetic flavin adenine dinucleotide in Cyanohydrins by enzyme catalyzed synthesis. Tetrahedron enzyme activity. FEBS Lett. 90,255 - 260. Lett. 34,4769 -4772. Becker, W., Benthin, U., Eschenhof, E., und Pfeil, E. (1963). Zur Kojima, M., Poulton, J.E., Thayer, S.S., and Conn, E.E. (1979). Kenntnis der Cyanhydrinsynthese II. Biochem. Z. 337, 156 - Tissue distributions of dhurrin and of enzymes involved in its 166. metabolism in leaves of Sorghum bicolor. Plant Physiol. 63, Becker, W., Freund, H., and Pfeil, E. (1965). Stereospecific syn- 1022-1028. thesis of D-hydroxy nitriles and optically active ethanolamines. Kuroki, G.W., and Conn, E.E. (1989). Mandelonitrile lyase from Angew. Chem. Int. Ed. Engl. 4,1079. Ximenia americana L.: Stereospecificity and lack of flavin Bove, C., and Conn, E.E. (1961). Metabolism of aromatic com- prosthetic group. Proc. Natl. Acad. Sei. USA 06,6978 - 6981. pounds in higher plants. J. Biol. Chem. 236,207 - 210. Liao, D.I., and Remington, S.J. (1990). Structure of wheat serine Brussee, J., Loos, W.T., Kruse, G.G., and van der Gen, A. (1990). carboxypeptidase II at 3.5-A resolution. J. Biol. Chem. 265, Synthesis of optically active silyl protected cyanohydrins. 6528-6531. Tetrahedron 46,979 - 986. Liao, D.I., Breddam, K., Sweet, R.M., Bullock, T, and Remington, Cheng, I.P., and Poulton, J.E. (1993). Cloning of cDNA of Prunus S.J. (1992). Refined atomic model of wheat serine carboxy- serotina (R)-(+)-mandelonitrile lyase and identification of a peptidase II at 2.2-Aresolution. Biochemistry 37,9796-9812. putative FAD-binding site. Plant Cell Physiol. 34,1139 -1143. Lieberei, R., Selmar, D., and Biehl, B. (1985). Metabolization of Conn, E.E. (1981). Cyanogenic glycosides. In: The Biochemistry cyanogenic glucosides in Hevea brasiliensis. PI Syst. Evol. 750, of Plants: A Comprehensive Treatise, Vol. 7: Secondary Plant 49-63. Products. P.K. Stumpf and E.E. Conn, eds. (New York: Nahrstedt, A. (1985). Cyanogenic compounds as protecting Academic Press), pp. 479 - 500. agents for organisms. PI Syst. Evol. 750,35 - 47. Effenberger, F. (1994). Synthesis and reactions of optically active Ognyanov, V.l., Datcheva, V.K., and Kyler, K.S. (1991). Prepara- cyanohydrins. Angew. Chem. Int. Ed. Engl. 33,1555 -1564. tion of Chiral Cyanohydrins by an Oxynitrilase-Mediated Effenberger, F., und Stelzer, U. (1993). Darstellung von (R)-2- Transcyanation. J. Am. Chem. Soc. 7 73,6992- 6996. (Sulfonyloxy)nitrilen und ihre Reaktionen mit Acetaten - Ollis, D.L, Cheah, E., Cygler, M., Dijkstra, B., Frolow, F., Franken, Konfigurationsumkehr optisch aktiver Cyanhydrine. Chem. S.M., Harel, M., Remmington, S., Silmann, l., Schräg, J., Ber. 726,779-786. Sussman, J.L., Veschueren, K.H.G., and Goldmann, A. (1992). Effenberger, F., Ziegler, T, and Förster, S. (1987). Enzyme-cataly- The / hydrolase fold. Protein. Engl. 5,197-211. zed cyanohydrin synthesis in organic solvents. Angew. Chem. Poulton, J.E. (1988). Localization and catabolism of cyanogenic Int. Ed. Engl. 26,458-460. glycosides. In: Cyanide Compounds in Biology. Ciba Foun- Eichhorn, J. (1995). Untersuchungen zur Substratabhängigkeit dation Symposium No. 140, D. Evered and S. Harnett, eds. der durch Oxynitrilasen katalysierten Cyanhydrinreaktion in (Chichester, UK: John Wiley and Sons), pp. 67 - 91. organischen Lösungsmitteln - enantioselektive Synthese von Reimmann, C., Hofmann, C., Mauch, F., and Dudler, R. (1995). (R)-Pantolacton und erythro-2-Methylaminopropanolen. Dis- Characterization of a rice gene induced by Pseudomonas sertation, Universität Stuttgart. syringae pv. syringae: requirement for the bacterial lemA gene Förster, S., Roos, J., Effenberger, F., Wajant, H., and Sprauer, A. function. Physiol. Mol. Plant. Pathol. 46,71 -81. (1996). The first recombinant hydroxynitrile lyase in the synthe- Roos, J. (1995). Untersuchungen des Substratspektrums der sis of (S)-cyanohydrins. Angew. Chem. Int. Ed. Engl. 35,437 - Hydroxynitril Lyase aus Manihot esculenta. Diplomarbeit, 439. Universität Stuttgart. Gerstner, E., Mätzke, V., und Pfeil, E. (1968). Zur chemischen und Rosenthaler, L (1908). Asymmetrie syntheses produced by en- biologischen Systematik der Rosaceen. Naturwissenschaften zymes. Biochem. Z. 14,238-253. 55,561-563. Saunders, JA, and Conn, E.E. (1977). Subcellular localization of Gerstner, E., und Pfeil, E. (1972). Zur Kenntnis des Flavinenzyms the cyanogenic glucoside of sorghum by autoradiography. Hydroxynitril-Lyase (D-Oxynitrilase). Hoppe Seyler's Z. Plant Physiol. 59,647-652. Physiol. Chem. 353,271 -286. Schmidt, M., Herve, S., Klempier, N., and Griengl, H. (1996). Hasslacher, M., Schall, M., Hayn, M., Griengl, H., Kohlwein, S.D., Preparation of optically active cyanohydrins using the (S)- and Schwalb, H. (1996). Molecular cloning of the full-length hydroxynitrile lyase from Hevea brasiliensis. Tetrahedron 52, cDNA of (S)-hydroxynitrile lyase from Hevea brasiliensis. J. 7833-7840. Biol. Chem.277,5884-5891. Seigler, D.S. (1991). Cyanide and cyanogenic glycosides. In: Hughes, J., Carvalho, J.P. de C., and Hughes, M.A. (1994). Herbivores: Their Interactions with Secondary Plant Meta- Purification, characterization, and cloning of a-hydroxynitrile bolites, G.A. Rosenthal and M.R. Berenbaum, eds. (New York: lyase from cassava (Manihot esculenta Crantz). Arch. Academic Press), pp. 35 - 77. Biochem. Biophys. 377,496 - 502. Selmar, D., Lieberei, R., and Biehl, B. (1988). Mobilization and uti- Huuhtanen, T.T., and Kanerva, L.T. (1992). Enzyme-catalysed lization of cyanogenic glycosides. Plant Physiol. 86,711-716. synthesis of optically active aliphatic cyanohydrins. Tetra- Smitskamp-Wilms, E., Brussee, J., van der Gen, A., van hedron Asymmetry 3,1223 -1226. Scharrenburg, G.J.M., and Sloothaak, J.B. (1991). Jäger, J. (1992). (R)-Oxynitrilase katalysierte Darstellung von (R)- Hydroxynitrile lyases from almond and sorghum as biocata- Cyanhydrinenhydroxysubstituierter Benzaldehyde-Synthese lysts. Reel. Trav. Chim. Pays-Bas 7 70,209 -215. von ß-Alkylaminoalkoholen aus (R)-Cyanhydrinen. Diplom- Swain, E., and Poulton, J.E. (1994). Utilization of amygdalin du- Hydroxynitrile Lyases of Higher Plants 617

ring seedling development of Prunus serotina. Plant Physiol. novel (R)-mandelonitrile lyase from the fern Phlebodium au- 706,437-445. reum. Plant Physiol. 109,1231 -1238. Swain, E., Li, C.P, and Poulton, J.E. (1992). Tissue and subcellu- Wajant, H., and Förster, S. (1996). Purification and characteriza- lar localization of enzymes catabolizing (R)-amygdalin in ma- tion of hydroxynitrile lyase from Hevea brasiliensis. Plant ture Prunus serotina seeds. Plant Physiol. 700,291 -300. Science 7 75,25-31. Thayer, S.S., and Conn, E.E. (1981). Subcellular localization of Wajant, H., and Pfizenmayer, K. (1996). Identification of potential dhurrin ß-glucosidase and hydroxynitrile lyase in the meso- active-site residues in the hydroxynitrile lyase from Manihot phyll cells of Sorghum leaf blades. Plant Physiol. 67,617-622. esculenta by site-directed mutagenesis. J. Biol. Chem., in Wajant, H., Mundry, K.-W. (1993). Hydroxynitrile lyase from press. Sorghum bicolor: a glycoprotein heterotetramer. Plant Science Wehtje, E., Adlercreutz, P., and Mattiasson, B. (1990). Formation 89,127-133. of C-C Bonds by Mandelonitrile Lyase in Organic Solvents. Wajant, H., Mundry, K.-W., and Pfizenmaier, K. (1994). Molecular Biotechnol. Bioeng. 36,39-46. cloning of hydroxynitrile lyase from Sorghum bicolor (L). Wu, H.C., and Poulton, J.E. (1991). Immunocytochemical loca- Homologies to serine carboxypeptidases. Plant Mol. Biol. 26, lization of mandelonitrile lyase in mature black cherry (Prunus 735-746. serotina Ehr.) seeds. Plant Physiol. 96,1329 -1337. Wajant, H., Förster, S., Böttinger, H., Effenberger, F., and Xu, L.L., Singh, B.K., and Conn, E.E. (1986). Purification and cha- Pfizenmaier, K. (1995a). Acetone cyanohydrin lyase from racterization of mandelonitrile lyase from Prunus lyonii. Arch. Manihot esculenta (cassava) is serologically distinct from other Biochem. Biophys. 250,322-328. hydroxynitrile lyases. Plant Science 708,1 -11. Xu, L.L., Singh, B.K., and Conn, E.E. (1988). Purification and cha- Wajant, H., Förster, S., Selmar, D., Effenberger, F., and racterization of acetone cyanohydrin lyase from Linum usitatis- Pfizenmaier, K. (1995b). Purification and characterization of a simum. Arch. Biochem. Biophys. 263,256 - 263.