R EPORTS dase-catalyzed oxidation of (Fig. Synthase: A 1A, 2) yields a 2-(␣-hydroxybenzyl)couma- ranone derivative, a hydrated form of , Oxidase Homolog which is then dehydrated (Fig. 1C, arrows a and b) (5). In this study, we establish that aurone biosynthesis is catalyzed by a ho- Responsible for Flower molog of plant (PPO). PPOs ubiquitously occur in higher plants and Coloration are responsible for the browning of plant tissues exposed to air. However, the physio- T. Nakayama,1* K. Yonekura-Sakakibara,2* T. Sato,1 S. Kikuchi,1 2 2 3 2 2 logical function of PPO in plants remains to Y. Fukui, M. Fukuchi-Mizutani, T. Ueda, M. Nakao, Y. Tanaka, be established (6). This report demonstrates 2 1 T. Kusumi, T. Nishino the participation of a PPO homolog in flower coloration. are plant flavonoids that provide yellow color to the flowers of some The yellow coloration of snapdragon flow- popular ornamental plants, such as snapdragon and cosmos. In this study, we ers is mainly provided by the glucosides of have identified an responsible for the synthesis of aurone from chal- aurones (aureusidin and bracteatin). Aureusidin cones in the yellow snapdragon flower. The enzyme () is a can be produced from either 2Ј,4Ј,6Ј,4-tetrahy- 39-kilodalton, -containing catalyzing the hydroxylation droxychalcone (THC) or 2Ј,4Ј,6Ј,3,4-pentahy- and/or oxidative cyclization of the precursor chalcones, 2Ј,4Ј,6Ј,4-tetrahy- droxychalcone (PHC), whereas bracteatin aris- droxychalcone and 2Ј,4Ј,6Ј,3,4-pentahydroxychalcone. The complementary es from PHC (7) (Fig. 1C, arrows e, f, and g). DNA encoding aureusidin synthase is expressed in the petals of aurone-con- A single enzyme, which should not be a perox- taining varieties. DNA sequence analysis revealed that aureusidin synthase idase, catalyzes dual chemical transformations, belongs to the plant polyphenol oxidase family, providing an unequivocal i.e., hydroxylation and oxidative cyclization (2Ј, example of the function of the polyphenol oxidase homolog in plants, i.e., flower ␣-dehydrogenation), of THC and PHC into au- coloration. reusidin and bracteatin, respectively (7). There-

Most floral colors present in nature arise from snapdragon (Antirrhinum majus) (Fig. 1B). 1Department of Biomolecular Engineering, Graduate (1). Genetic and biochemical Although the aurone biosynthetic gene(s) is School of Engineering, Tohoku University, Aoba-yama knowledge of biosynthesis in an attractive tool to engineer yellow flowers, 07, Sendai 980-8579, Japan. 2Institute for Fundamen- plants has provided a basis for controlling the biochemical and genetic details of aurone tal Research, Suntory Ltd., Wakayamadai 1-1-1, Shi- mamoto-cho, Mishima-gun, Osaka 618-8503, Japan. floral color through genetic engineering ap- biosynthesis have remained unclear (4). One 3Kobe Gakuin University, Kobe 651-2180, Japan. proaches (2). Aurones (Fig. 1A, 1) (3), a class mechanism proposed for aurone biosynthesis *To whom correspondence should be addressed. E- of plant flavonoids, confer bright yellow col- in plants (soy seedling) is a two-step path- mail: [email protected]; yskeiko@ or to flowers such as cosmos, coreopsis, and way, in which an H2O2-dependent peroxi- postman.riken.go.jp

Fig. 1. (A) General structures of aurones, 1, and chalcones, 2. The positional numbering in the aurones and chalcones is different. A and B shown in the aromatic rings indicate A- and B-rings in the flavonoid structures, respectively. (B) Photograph of yellow snapdragon flowers. (C) Possible pathways for aurone biosynthesis from chalcones in the snapdragon flowers. Dotted arrows indicate the putative pathway for aurone biosynthesis according to Rathmell et al. (5). Thick arrows indicate the pathway for aurone biosynthesis identified in this study.

www.sciencemag.org SCIENCE VOL 290 10 NOVEMBER 2000 1163 R EPORTS fore, we attempted to purify the enzyme respon- num majus (cv. Yellow Butterfly) that had the purified aureusidin synthase (39 kD). sible for aurone biosynthesis, which we call been prepared as described previously (11). It However, such a difference is consistently aureusidin synthase, from crude extracts of yel- contained an open reading frame of 1686 explained in terms of the proposed biogenesis low snapdragon flowers. base pairs encoding 562 amino acids. The of known plant PPOs; plant PPOs are gener- To purify a sufficient amount of the en- deduced sequence showed high ally synthesized as an ϳ65-kD precursor pro- zyme, we used 32 kg of snapdragon buds (8) similarity to plant PPOs such as those from tein (15), which is subsequently processed as starting material. Moreover, the addition of apple fruit (identity, 51%), grape berry into a mature PPO of ϳ40 kD after removal

THC to the buffers used during purification (47%), and potato tuber (39%) (Fig. 3) (13, of a 10-kD NH2-terminal peptide containing somewhat stabilized enzyme activity. Finally, 14). The predicted Mr (64 kD) of the AmAS1 transit sequences (15) and a 15-kD COOH- the enzyme (90 ␮g) was purified 107-fold to product was substantially larger than that of terminal peptide of unknown function (16). homogeneity by a combination of nine puri- fication steps in an activity yield of 1.4% (Fig. 2) (8). The mean value of the final specific activities under the standard assay conditions with THC as a substrate was 578 units mgϪ1 (9). However, the specific activ- ity of the enzyme and the actual degree of purification might have been considerably higher, given the instability of this enzyme during the purification. The relative molecu-

lar mass (Mr) of the purified aureusidin syn- thase was estimated to be 40 kD by gel permeation fast protein liquid chromatogra- phy on a Superdex 200HR column. The sub-

unit Mr of the enzyme was estimated to be 39 kD by SDS–polyacrylamide gel electrophore- sis (SDS-PAGE) (Fig. 2). These results show that the enzyme was monomeric. The aureu- sidin synthase band in the SDS-PAGE gels was positive to sugar staining (10). In addi- tion, the aureusidin synthase was adsorbed on a ConA Sepharose column and was specifi- cally eluted with methyl ␣-glucoside, indicat- ing that the enzyme is a glycoprotein. We then determined the amino acid se- quences of peptide fragments obtained by lysyl-endopeptidase digestion of the purified synthase essentially as described (11). The sequences coincided with the deduced amino acid sequences of a clone that was specifical- ly expressed in the aurone-containing petals, as obtained through a subtractive hybridiza- tion approach (12). Using this clone as a probe, we isolated a cDNA (termed AmAS1) that encodes the precursor of aureusidin syn- thase from a petal cDNA library of Antirrhi-

Fig. 2. SDS-PAGE of purified aureusidin synthase. The purified enzyme (right lane) (8) was simultaneously subjected to electro- phoresis with standard proteins (left lane). Molecular sizes (in kD) Fig. 3. Alignment of the deduced amino acid sequence of AmAS1 protein with those of plant PPOs. are indicated by ar- Amino acid residues identical to that of AmAS1 are shown in red. The dashed underline below the rows. AmAS1 sequence indicates the identification of purified peptides derived from the purified aureusidin synthase. The histidine ligands for the active-site binuclear Cu center, identified in potato PPO by x-ray crystallography (17), are shown with black circles. Putative Cu-binding domains (A and B) of AmAS1 are underlined (shown as a and b, respectively). The NH2-terminal processing sites identified in the PPO sequences of grape and potato and the COOH-terminal processing site identified in that of grape are indicated by vertical lines in the sequence. Sequences of known PPOs representing the “n region” and the “thylakoid transfer domain” are boxed and indicated by I and II, respectively. The nucleotide sequence of AmAS1 has been submitted to the DNA Data Bank of Japan under the accession number AB044884. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

1164 10 NOVEMBER 2000 VOL 290 SCIENCE www.sciencemag.org R EPORTS The processing sites in the sequence of the product, as follows, with an optimum pH of To confirm the role of the AmAS1 gene AmAS1 protein remain to be determined be- 5.4: in flower coloration, we analyzed the spa- cause the NH -terminus of the purified au- tial and temporal expressions of the AmAS1 2 THC ϩ O 3 aureusidin ϩ H O reusidin synthase was blocked. The deduced 2 2 gene in the snapdragon plants (Fig. 4). amino acid sequence of the AmAS1 protein The rate of aureusidin formation from THC Northern blot analysis revealed that AmAS1 contained two putative Cu-binding domains was greatly enhanced by H2O2, which should was expressed in the aurone-accumulating that are conserved in those sequences of the act as an . This observation petals. More AmAS1 transcripts were found plant PPOs that are binuclear Cu is consistent with the nature of PPO , in the petals of yellow varieties than in the (17). Atomic absorption spectrophotometric where the monophenol monooxygenase ac- petals of pink varieties, which contain analysis of the purified aureusidin synthase tivity of PPO is activated by H2O2 (19). PHC small amounts of aurones; and no tran- allowed us to confirm that it was indeed a served as a much better substrate than THC scripts were detectable in the aurone-lack- binuclear Cu enzyme (18). (relative activity, 2210%) for the purified ing petals of white, pink, and red varieties In accordance with previous observa- aureusidin synthase to produce aureusidin (Fig. 4A). No transcript was observed in tions with crude extracts (7), the purified and bracteatin in a molar ratio of 6:1 in the the stem and leaf (Fig. 4B), both of which aureusidin synthase could catalyze the absence of H2O2 (7). H2O2 instead inhibited have no aureusidin synthase activity (7). 3-hydroxylation and oxidative cyclization the reaction with PHC. The formation of The expression pattern during the develop- of THC to yield aureusidin as a single aureusidin from PHC was also accompanied ment of yellow flowers was also examined. by oxygen uptake and proceeded with the The transcripts were most abundant at stage following stoichiometry and a broad pH op- 5 of flower development (Fig. 4C). Such timum around 5.0 to 7.0: temporal expression was consistent with the trend observed for aureusidin synthase PHC ϩ 1⁄2O 3 aureusidin ϩ H O 2 2 activity, which was highest at stages 5 and The 4Ј-glucosides of THC and PHC were 6(7). Thus, the expression of the AmAS1 also good substrates (relative activities, 220% gene is temporally regulated during flower and 2496%, respectively). These results development, as was the case for the genes prove that the formations of aureusidin and responsible for anthocyanin biosynthesis bracteatin from chalcones are single enzy- (2). These results led us to conclude that matic processes catalyzed by the same en- aureusidin synthase is a plant PPO homolog zyme (Fig. 1C). responsible for the yellow coloration of The present identification of aureusidin snapdragon flowers. synthase as a homolog of PPO prompted us to examine whether the PPOs generally show References and Notes aureusidin synthase activity. When the Neu- 1. R. Brouillard, O. Dangles, in The Flavonoids: Advances rospora crassa , a binuclear Cu in Research Since 1986 (Chapman & Hall, London, PPO distantly related to plant PPOs, was 1993), pp. 565–588. 2. Y. Tanaka, S. Tsuda, T. Kusumi, Plant Cell Physiol. 39, reacted with THC (20), it also yielded aureu- 1119 (1998). sidin as a single product (specific activity, 59 3. E. C. Bate-Smith, T. A. Geissman, Nature 167, 688 U/mg), establishing the ability of PPO to (1951). 4. M. Shimokoriyama, S. Hattori, J. Am. Chem. Soc. 75, catalyze aurone synthesis from chalcones. 2277 (1953). However, aureusidin synthase showed virtu- 5. W. G. Rathmell, D. S. Bendall, Biochem. J. 127, 125 ally no 3Ј,4Ј-dehydrogenation activity toward (1972). aureusidin. 6. K. C. Vaughn, A. R. Lax, S. O. Duke, Physiol. Plant. 72, 659 (1988). The deduced NH2-terminal amino acid 7. T. Sato et al., Plant Sci., in press. sequence of AmAS1, as well as the properties Fig. 4. Northern blot analysis of AmAS1 gene in 8. Purification was completed at 0° to 5°C, unless oth- erwise stated, as follows. Enzyme preparation B (7) snapdragon plants. Ten (A and C) and 20 ␮g (B) of aureusidin synthase elucidated in this study, provides important information for in- prepared with buffer A [0.01 M sodium acetate (pH of total RNA were used for each analysis (11). 5.0) containing 1 ␮M THC] from 32 kg of buds [stages (A) The expression of AmAS1 in petals (at stage sights into the subcellular localization of en- 2to5(7)], taken from 16,000 stems of yellow 5) of different floral color was analyzed. Total zyme. Plant PPOs are localized in plastids snapdragon plant, was dialyzed against 5 mM potas- RNA was prepared from petals of the following (chloroplasts) (6); the NH -terminal amino sium phosphate (pH 5.0) containing 1 ␮M THC and varieties: YB, Yellow Butterfly; BLP, Butterfly 2 acid sequences of the plant PPOs have fea- 0.3 mM CaCl2 (buffer B). The enzyme solution was Light Pink; OW, Oakland White; ML, Maryland then applied to a column of Gigapite hydroxyapatite ( W, white; Y, yellow; R, red; P, pink). Plus and tures that are common to transit peptides (Seikagaku, Tokyo) equilibrated with buffer B. The minus signs indicate the accumulation and ab- known to be targeted to the internal lumen of enzyme activity was eluted witha5to500mMlinear sence, respectively, of aurones in the petals (7). thylakoid membranes in plastids (15) (Fig. 3). gradient of potassium phosphate (without CaCl2). (B) Spatial expression of the AmAS1 gene in However, the localization of aureusidin syn- After concentration by ultrafiltration, the concen- snapdragon plants with yellow flowers. L, 3- to trate was applied to a HiLoad16/60 Superdex 75 HR thase in plastids is less likely because the column equilibrated with buffer A, containing 0.15 M 4-cm-long leaves; S, stems; P1, petals at stage deduced NH -terminal sequence of the NaCl and 0.07% CHAPS, and eluted. The active frac- 1; and P6, petals at stage 6. (C) The expressions 2 AmAS1 protein does not share such features tions were combined and dialyzed against buffer A of the AmAS1 (upper) and the isomer- containing 0.07% CHAPS. The enzyme solution was ase genes (CHI, lower) in the petals at five (21). Moreover, are not found applied to a column of SP-Sepharose FF equilibrated different stages of flower development (7). To localized in plastids. More likely, the enzyme with buffer A containing 0.07% CHAPS. The enzyme analyze the expressions of CHI gene, we ampli- occurs in , because 4Ј-glucosides of was eluted with a linear gradient of NaCl (0.04 to 0.32 M) in the same buffer. The active fractions were fied partial CHI cDNAs by polymerase chain THC and PHC, which are very good sub- reaction on the basis of the submitted se- combined, dialyzed against buffer B containing 0.07% strates for the enzyme even without added CHAPS, and rechromatographed on a Gigapite col- quence of snapdragon CHI cDNA (GenBank ac- H O , are found in vacuoles, and the enzyme umn essentially as described above, except that buff- cession number M68326). The amplified frag- 2 2 er B contained 0.07% CHAPS. To the active fractions, 32 ment was identified by DNA sequencing, P- activity shows optimum pH’s at the pH typ- ammonium sulfate was added to 20% saturation. The labeled, and used as a probe. ical of vacuoles. resultant enzyme solution was applied to a Phenyl-

www.sciencemag.org SCIENCE VOL 290 10 NOVEMBER 2000 1165 R EPORTS

Sepharose HR5/5 column and eluted by washing the oxygen electrode system (Hansatech Instruments, 18. Purified aureusidin synthase was dialyzed in a poly- column with 0.01 M potassium phosphate (pH 5.0) Norfolk, UK) (7). styrene flask at 4°C for 48 hours against metal-free containing 0.1% CHAPS and ammonium sulfate at 10. J. T. Clarke, Ann. N.Y. Acad. Sci. 121, 428 (1964). Milli-Q water. The inner and outer solutions for the 20% saturation. Active fractions were immediately 11. H. Fujiwara et al., Plant J. 16, 421 (1998). dialysis were analyzed for Cu content by atomic dialyzed against 0.01 M potassium phosphate (pH 12. K. Yonekura-Sakakibara, unpublished results. absorption spectrometry with a model AA-6700F 5.0) containing 0.1% CHAPS. The purified enzyme 13. References and GenBank accession numbers for PPO apparatus (Shimadzu). gave a single protein band as judged by silver staining sequences are as follows: grape [I. B. Dry, S. P. Rob- 19. A. Sanchez-Ferrer, J. N. Rodriguez-Lopez, F. Garcia- after SDS-PAGE [U. K. Laemmli, Nature 227, 680 inson, Plant Mol. Biol. 26, 495 (1994); S52629]; and Canovas, F. Garcia-Carmona, Biochim. Biophys. Acta (1970)]. potato [P. W. Thygesen, S. P. Robinson, Plant Physiol. 1247, 1 (1995). 109, 525 (1995); AAA85122]. 9. Aureusidin synthase activity was assayed by re- 20. The N. crassa PPO (Sigma) was reacted with THC at 14. Supplemental Web material is available at Science versed-phase high-performance liquid chromatogra- its optimum pH (6.5) in the absence of H O as Online at www.sciencemag.org/feature/data/ 2 2 phy as described [Method I in (7)]. One unit of described (9). enzyme was defined as the amount that catalyzed 1053401.shl. 21. PSORT analyses (http://psort.nibb.ac.jp/) did not pre- the formation of 1 nmol of aureusidin per minute. 15. R. W. Joy IV, M. Sugiyama, H. Fukuda, A. Komamine, dict the localization of aureusidin synthase in the The specific activity of aureusidin synthase was ex- Plant Physiol. 107, 1083 (1995). plastids . pressed as units/mg of protein. The rate of oxygen 16. S. P. Robinson, I. B. Dry, Plant Physiol. 99, 317 (1992). consumption during aureusidin synthase reaction 17. T. Klabunde, C. Eicken, J. C. Sacchettini, B. Krebs, was also monitored with the Hansatech DW1/CB1D Nature Struct. Biol. 5, 1084 (1998). 22 June 2000; accepted 26 September 2000

ulations, most workers are infected by some Survival for Immunity: The Price parasite but nevertheless show normal behav- iors and activities (5). Bumblebee workers of Immune System Activation usually do not reproduce themselves. Hence, worker (inclusive) fitness is determined by their survival, and therefore any cost of im- for Bumblebee Workers munity that reduces survival also reduces fit- Yannick Moret* and Paul Schmid-Hempel ness (6). As in other insects, immunity in bumblebees is innate and based on both cel- Parasites do not always harm their hosts because the immune system keeps an lular (7) and humoral mechanisms (8). An infection at bay. Ironically, the cost of using immune defenses could itself immune response starts with the recognition reduce host fitness. This indirect cost of parasitism is often not visible because of immunogens released by or present on the of compensatory resource intake. Here, workers of the bumblebee, Bombus surface of parasites entering the host hemo- terrestris, were challenged with lipopolysaccharides and micro–latex beads to coel. Various pathways of the immune sys- induce their immune system under starvation (i.e., not allowing compensatory tem then become activated (9), leading to the intake). Compared with controls, survival of induced workers was significantly destruction of the parasite and its removal by reduced (by 50 to 70%). cellular reactions such as phagocytosis or encapsulation. Parasitic infections are pervasive, but hosts indirectly, for example, by forcing the indi- To measure the survival cost of the im- often show no obvious effects. Alas, this does vidual to increase its parental effort and mea- mune reaction, we experimentally activated not mean that parasites impose no fitness suring the corresponding decrease in the im- the worker’s immune system with two kinds costs on their host, because the immune sys- mune response (2). of established immune elicitors. (i) Lipopoly- tem is often able to keep the infection within Here, the survival cost for the activation saccharides (LPS; Sigma L-2755), i.e., sur- bounds. Recent discussions in the field of of the immune system was analyzed when the face molecules extracted from Escherichia evolutionary ecology have concentrated on host was denied compensation for increased coli. This nonpathogenic and nonliving elic- the idea that the evolution of the immune demand. In particular, the host’s condition itor is specifically recognized by pattern rec- system is traded off against other fitness com- was experimentally “frozen” by adopting a ognition proteins of the invertebrate immune ponents (1). In addition, the activation and starvation protocol at the point when an “in- system (9). LPS induces several pathways of use of the immune system are thought to be fection”—a standardized immunogenic chal- the immune response (8, 10) that persist over costly and therefore cannot be sustained si- lenge—occurred (3). When an individual is many hours (11). LPS is cleared from insect multaneously with other demanding activities starved, any future allocation to defense re- hemolymph by lipophorin, a transport protein (2). With such costs, the main effect of infec- duces the resources available for other needs that shuffles LPS to the fat body (11, 12). tion is not the direct damage by the parasite and thus eventually for maintenance and sur- Hence, the clearance of LPS should not in- itself but the cost imposed when the host vival. The starvation paradigm also mimics volve processes that are responsible for clear- immune system is activated. Why, if these some important ecological conditions, such ing bacteria from the hemolymph such as costs exist, are they not more often evident? as the natural occurrence of adverse weather phagocytosis. (ii) Sterile micro–latex beads One reason is that hosts may compensate for and limited food availability, that are typical (Polysciences; diameter 4.5 ␮m). These increased demand by increased resource in- for most animal populations. beads are a similar size to bacteria and are take. Costs are thus masked and no outward In this study, workers of the bumblebee, cleared from the hemolymph by a combina- signs of a parasitic infection are observed, Bombus terrestris L., were used as hosts. tion of processes (8, 10), including phagocy- although the host pays a cost to prevent the Bumblebees are primitively eusocial insects tosis. In both cases, the immune system is establishment and spread of the parasite. To inhabiting temperate habitats where weather activated, but the artificial “parasite” is un- date, such fitness costs have only been shown conditions often vary over short time periods. able to generate any pathogenic effect. Foraging activity is often interrupted by A first experiment tested whether de- spells of rain and cold weather, leading to the creased survival might result from a toxic Eidgeno¨ssische Technische Hochschule (ETH) Zu¨rich, Experimental Ecology, ETH-Zentrum, NW, CH-8092 starvation and demise of the colony if work- side effect of the immune elicitors, assuming Zu¨rich, Switzerland. ers fail to collect sufficient amounts of pollen that such effects would also decrease the *To whom correspondence should be addressed. E- and nectar (4). Starved workers cannot sur- survival of nonstarved animals. In addition, mail: [email protected] vive for long (20 to 30 hours). In field pop- survival should then correlate with dose (13).

1166 10 NOVEMBER 2000 VOL 290 SCIENCE www.sciencemag.org