Plant Molecular Biology 54: 39–54, 2004. 39 © 2004 Kluwer Academic Publishers. Printed in the Netherlands.

A small family of LLS1-related non- in plants with an origin amongst oxygenic photosynthesizers

John Gray1,∗, Ellen Wardzala1,ManliYang1, Steffen Reinbothe2, Steve Haller1 and Florencia Pauli3 1Department of Biological Sciences, University of Toledo, 2801 West Bancroft Street, Toledo, OH 43606, USA (∗author for correspondence; email [email protected]); 2Laboratoire de G´en´etique Mol´eculaire des Plantes, Universit´e Joseph Fourier et CNRS, CERMO, BP 53, 38041 Grenoble, France; 2Department of Genetics, School of Medicine, Stanford University, Stanford, CA 94305, USA

Received 22 July 2003; accepted in revised form 22 December 2003

Key words: Acd1, Cao ( a ), Cmo (choline monooxygenase), dioxygenase, Lls1, non-, Pao (pheophorbide a oxygenase), Ptc52, Tic55

Abstract Conservation of Lethal-leaf spot 1 (Lls1) lesion mimic gene in land plants including moss is consistent with its recently reported function as pheophorbide a oxygenase (Pao) which catalyzes a key step in chlorophyll degrada- tion (Pruzinska et al., 2003). A bioinformatics survey of complete plant genomes reveals that LLS1(PAO) belongs to a small 5-member family of non-heme oxygenases defined by the presence of Rieske and mononuclear iron- binding domains. This gene family includes chlorophyll a oxygenase (Cao), choline monooxygenase (Cmo), the gene for a 55 kDa protein associated with protein transport through the inner chloroplast membrane (Tic55) and a novel 52 kDa protein isolated from chloroplasts (Ptc52). Analysis of gene structure reveals that these genes diverged prior to monocot/dicot divergence. Homologues of LLS1(PAO), CAO, TIC55 and PTC52 but not CMO are found in the genomes of several cyanobacteria. LLS1(PAO), PTC52, TIC55 and a set of related cyanobacterial homologues share an extended carboxyl terminus containing a novel F/Y/W-x2-H-x3-C-x2-C motif not present in CAO. These proteins appear to have evolved during the transition to oxygenic photosynthesis to play various roles in chlorophyll metabolism. In contrast, CMO homologues are found only in plants and are most closely related to aromatic ring-hydroxylating from soil-dwelling bacteria, suggesting a more recent evolution of this , possibly by horizontal gene transfer. Our phylogenetic analysis of 95 extant non-heme dioxygenases provides a useful framework for the classification of LLS1(PAO)-related non-heme oxygenases.

Introduction the recent finding that the Arabidopsis thaliana Lls1 gene encodes pheophorbide a oxygenase (PAO) which The Lls1 gene was originally cloned from maize and catalyzes a key step in chlorophyll degradation (Fig- the absence of this gene function results in a light- ure 1D) (Pruzinska et al., in press). Since the discovery dependent cell death phenotype mediated by chloro- of LLS1 (PAO) in plants a few other genes have been plasts (Gray et al., 1997, 2002). We have found identified in plants that exhibit the same non-heme that this cell protective function is conserved between iron-binding motifs (Caliebe et al., 1997; Tanaka monocots and dicots (Yang et al., submitted). Based et al., 1998). In this study it is established that there are on the presence of two non-heme iron-binding mo- a total of five Lls1(Pao)-related genes in plants. The tifs conserved amongst aromatic ring-hydroxylating phylogenetic relationships between these Lls1(Pao)- enzymes in bacteria it was predicted that the Lls1 related genes and homologous bacterial enzymes were gene encodes an oxygenase function (Gray et al., examined in detail by comparing 95 known and pre- 1997, 2002). This prediction has been confirmed by dicted non-heme oxygenases. 40

Figure 1. Examples of known catalytic functions of non-heme oxygenases from bacteria and plants. A. Naphthalene dioxygenase from Pseudo- monas sp. strain G7 catalyzes the conversion of naphthalene to cis-1,2-dihydroxy-1,2-dihydronaphthalene. B. Choline monooxygenase (CMO) catalyzes the first step in the conversion of choline to the osmoprotectant glycine betaine in plant chloroplasts. C. Chlorophyll a oxygenase (CAO) catalyzes the first step in the conversion of the Chl a to Chl b. D. Pheophorbide a oxygenase (PAO) catalyzes the oxygenolytic opening of pheophorbide a at the α-mesoposition between C4 and C5 to produce a red chlorophyll catabolite (RCC).

Non-heme iron oxygenases or hydroxylases that incorporation of both atoms of molecular oxygen are incorporate one or two atoms of dioxygen into sub- referred to as dioxygenases. These dioxygenases com- strates are found in many metabolic pathways (Lange prise a large and diverse group of multi-component en- and Que, 1998; Moraswki et al., 2000; Prescott zymes that play important roles in pathways as diverse and Lloyd, 2000; Ryle and Hausinger, 2002). En- as antibiotic synthesis to the degradation of aromatic zymes that incorporate only one atom of dioxygen compounds. The comparison of the deduced amino into substrates are termed monooxygenases (or mixed- acid sequences of numerous oxygenases is permitting function oxygenases). Oxygenases that catalyze the the evolutionary relationships between these enzymes 41 to be determined (Lange and Que, 1998; Prescott and classification scheme that groups these enzymes based Lloyd, 2000; Ryle and Hausinger, 2002). Of particular on the phylogenetic comparison of their terminal oxy- relevance to this study, because of their homology to genase components. This system, which is simple and the LLS1(PAO)-like plant oxygenases, are a group of powerful, is referred to here as the Nam classifica- microbial oxygenases that participate in the aerobic tion system. The Nam system classifies ARH enzymes degradation of aromatic hydrocarbons. These oxy- into four groups and representative examples included genases known as aromatic ring hydroxylases (ARHs) in this study are listed in Supplementary Table 1. In catalyze the hydroxylation of the aromatic ring as a this study we adapt and extend the Nam scheme for first step in the degradation of these compounds by classifying non-heme oxygenases to include plant and soil bacteria (Batie et al., 1991; Harayama et al., cyanobacterial non-heme oxygenases. 1992; Mason and Cammack, 1992; Jiang et al., The evolutionary relationship of plant LLS1-like 1996; Nam et al., 2001). An example is naph- oxygenases and cyanobacterial oxygenases to these thalene dioxygenase (NDO) from the soil bacterium microbial ARH enzymes is examined in this paper. Pseudomonas which oxidizes naphthalene to cis-1,2- Two of the plant oxygenases related to LLS1 have dihydroxy-1,2-dihydronaphthalene (Figure 1A). NDO previously defined biochemical functions and these has been crystallized permitting the structure and reac- are choline monooxygenase (CMO) and chlorophyll tion mechanism to be studied in detail (Kauppi et al., a oxygenase (CAO) (Figure 1B and C). Both oxy- 1998; Karlsson et al., 2003). NDO consists of two genases do not utilize phenolic compounds as sub- subunits (α and β) in a hexameric α3β3 composi- strates. CMO is a ferredoxin-dependent enzyme that tion. The α subunit contains a Rieske[2Fe-2S] center catalyzes the first step in two-step oxidation of choline and mononuclear iron at the . The Rieske to the osmoprotectant glycine betaine (Figure 1B). domain exhibits a conserved iron-binding domain, This enzyme is found as a homodimer in the chloro- C81-x-H83-x17-C101-x2-H104, which contrasts with plast stroma (Rathinasabapathi et al., 1997). CMO is the Rieske domain of chloroplast-type ferredoxins, in unique to plants: in bacteria (including halotolerant which four cysteine residues co-ordinate a [2Fe-2S] cyanobacteria) and mammals that synthesize glycine center. In the NDO Rieske [2Fe-2S] center the Fe1 betaine, this first step is catalyzed by choline dehydro- is coordinated by Cys-81 and Cys-101, while Fe2 is genase (Incharoensakdi and Wutipraditkul, 1999). coordinated by His-83 and His-104. The mononuclear CAO catalyzes the first step in the conversion of iron at the active site is coordinated by the two his- chlorophyll a (Chl a)toChlb (Figure 1c) (Tanaka tidine residues and one carboxylate within the motif et al., 1998). The conservation of this enzyme in the E100-x3-D205-x2-H208x4-H213 and by D362. This prochlorophytes Prochloron didemni and Prochloro- motif is now referred to as a 2-His-1-carboxylate facial thrix hollandica suggests that this enzymatic activity triad (Lange and Que, 1998). Electron transfer from arose once in the common ancestor of oxygenic bac- the Rieske domain to the active site occurs between teria and chloroplasts (Tomitani et al., 1999). The two a subunits which are held in close proximity by absence of a close homologue of CAO from the gen- the hydrogen bonding of D205 to both H104 in the ome of Prochlorococcus which synthesizes divinyl Rieske domain and H208 at the active site (Kauppi Chl b has given rise to the proposition that a Chl b et al., 1998). The conservation of these iron-binding synthase in this organism may have a different phylo- motifs in LLS1-like proteins (Gray et al., 1997, 2002) genetic origin involving convergent evolution (Hess suggests a common reaction mechanism, but not ne- et al., 2001). PTC52 from barley (PORA translocation cessarily common enzyme substrates. Individual mi- complex) is a 52 kDa protein that is associated with the crobial ARH enzymes may operate on several sub- translocation of A strates, for example, vanillate demethylase (VanA) (PORA). The detection of pchlide b within this com- from Acinetobacter, which can act upon the vanil- plex suggests that PTC52 catalyzes the conversion of late analogues m-anisate, m-toluate and 4-hydroxy- pchlide a to pchlide b in a reaction that is analagous to 3,5-dimethylbenzoate (Moraswki et al., 2000). The that of CAO (S. Reinbothe et al., in press). microbial ARH enzymes were originally classified by The recent report that the AtLls1 gene encodes Batie et al. (1991), based on the number of constituent a pheophorbide a oxygenase activity and that pheo- components and the nature of their redox reactions. phorbide a accumulates in maize lls1 plants indicates This system proved difficult with some newly dis- that the LLS1 protein catalyzes a key step in the covered enzymes, and Nam et al. (2001) developed a degradation of chlorophyll (Pruzinska et al., 2003). 42

Thus, three members of the Lls1-related family of non- TBST, blots were incubated for 1 h at room temperat- heme oxygenases have been implicated in either the ure with peroxidase-conjugated goat anti-mouse IgG synthesis or degradation of chlorophyll intermediates. (1:5000, Jackson Immunoresearch Lab., West Grove, The fifth member of this gene family is TIC55 from PA) and the immunoreactive complexes developed for pea (translocon of inner chloroplast membrane) which visualization. encodes a 55 kDa protein that is found associated with the general translocon for proteins through the inner RNA isolation, RT-PCR and sequencing chloroplast membrane (TIC complex) (Caliebe et al., 1997; Küchler et al., 2002). A redox regulatory func- Total RNA was isolated by grinding 1 g frozen Ar- tion has been proposed for TIC55 but its precise role abidopsis ecotype Columbia leaf tissue to a powder in the TIC import complex is yet to be defined. and suspending in 10 ml RNA extraction buffer (REX) The fact that LLS1(PAO)-related proteins exhibit (2.0 M guanidine thiocyanate, 0.6 M ammonium iron-binding motifs that are conserved in bacterial thiocyanate, 0.2 M sodium acetate pH 4.0, 8% gly- ARH enzymes but operate on different substrates cerol, 50% phenol). After vortexing samples for 5 min, leaves open the question of the evolutionary rela- 2 ml of chloroform was added and samples were vor- tionship between these enzymes. In this study we texed for a further 3 min. Phases were separated by hypothesized that all LLS1(PAO)-related oxygenases centrifugation at 12 000 × g for 15 min, the upper share a common origin. We sought support for this phase recovered, added to 5 ml of isopropanol, mixed, incubated at 25 ◦C for 10 min, and centrifuged at hypothesis by examining the phylogenetic relation- ◦ ship between a collection of 95 plant and bacterial 12 000×g for 10 min at 4 C. The RNA precipitate was non-heme oxygenases that share similar iron-binding washed with 70% ethanol and centrifuged again. The motifs. Our results suggest a common origin for LLS1, RNA pellet was allowed to air-dry and re-suspended PTC52, TIC55 and CAO in oxygenic cyanobacteria in RNAase-free water. mRNA was isolated from total but a separate and more recent origin for CMO en- RNA with the MicroPoly(A)pure isolation kit ac- zymes. cording to the manufacturer’s instructions (Ambion, Austin, TX). Reverse transcription was performed with the Retroscript first-strand synthesis kit accord- Materials and methods ing to the manufacturer’s instructions (Ambion). The coding regions of the Arabidopsis Lls1(At3g44880) Plant material and Ptc52(At4g25650) genes were amplified by poly- merase chain reaction (PCR) employing the primer Plants of Arabidopsis ecotype Columbia were used for sets AGSP47/AGSP48 and ALGSP1/ALGSP2 re- mRNA isolation. Plants were grown at 22 ◦C under a spectively (listed below). The annealing temperat- 16 h/8 h light/dark regime at ca. 200 µmol photons ures used were 50 ◦C and 60 ◦C, respectively, −2 −1 m s . Herbaceous specimens were collected from and MgCl2 concentration was 4 mM. The primers the vicinity of the University of Toledo, researchers were ALGSP1 (ATGGAAGCTGCTCTTGCTGCAT- gardens, and the Stranahan Arboretum, Toledo, OH. GCGCTCTTCC), ALGSP2 (CATTTCAAACAAC- AGCATGGTTGTAGTCATGGTAATGG), AGSP47 Protein isolation and immunoblot analysis (ATGTCAGTAGTTTTACTCTCTTCTACTTCTGC), and AGSP48 (CTACTCGATTTCAGAATGTACAT- Total protein extracts were isolated from plant leaves AATCTCTAAAC). as previously described (Cheng et al., 1996). Cel- Amplified PCR products were cloned into the lular lysates were clarified, and protein was quanti- pTOPO cloning vector according to the manufac- fied by Bradford analysis (Bio-Rad, Richmond, CA). turer’s protocol (Invitrogen, Grand Island, NY). The Proteins (20 µg) were separated on a 7.5% SDS- plasmids clones containing Arabidopsis Lls1(Pao) and polyacrylamide gel and blotted onto nitrocellulose. Ptc52 coding regions were named pYM18-1 and Nitrocellulose membranes were blocked for 45 min at pYM20-1 respectively (GenBank accession number room temperature with 5% skim milk in Tris-saline AY344061 and AY344062 respectively). The DNA se- buffer pH 8.0 containing 0.05% Tween-20 (TBST) quence of cloned PCR products was determined with and then incubated in primary antibody, mouse anti- BigDye Terminator Cycle Sequencing and an Applied LLS1:MBP fusion protein for 1 h. After washing in Biosystems 3700 DNA Analyzer at the Plant-Microbe 43

Genomics Facility (Ohio State University, Columbus, sativa Tic55 gene; Scaffold AAAA01023723, trans- OH). lation of 718–2337 and supported by ESTs D47027, BU667200, BM422257, BM422260, J010H03, Retrieval of sequences from databases and BI808977, BI808984. A. thaliana Lls1(Pao) (Acd1) determination of gene structure gene Atg344880 (NC−003074); join bases 16392818– 16393231, 16393535–16393770, 16393865– The neighborhood search algorithm BLAST (Ba- 16393991, 16394080–16394216, 16394300– sic Local Alignment Search Tool; Altschul et al., 16394459, 16394532–16394816, 16394910– 1997) was employed for database searches through 16395164, and supported by full-length cDNA clone the National Center for Biotechnology Informa- RAFL07-16-F05 and pYM18-1 (AY344061), and tion (NCBI), The Arabidopsis Information Resource ESTs 175M23T7, BE526342, 84E8T7, BE844958, (TAIR), Cyanobase, and DOE Joint Genome Insti- pAZNII0788R, APZ16c10R, RAFL09-43-F19, tute BLAST WWW servers. In addition, rice gen- APZ74b02F. O. sativa Lls1(Pao) gene (scaffold omic sequences were retrieved from the Rice Genome AAAA01000724); join complement bases 16095– Database (http://210.83.138.53/rice/). The amino acid 15707, 15609–15372, 15284–15158, 14631–14495, sequences of 95 non-heme oxygenases were down- 14354–14195, 13993–13717, 13065–12811, and loaded or conceptually translated from genomic DNA supported by ESTs AA753785 and BI801593. and cDNA sequences available from GenBank files Z. mays Lls1(Pao) gene (U77346); join bases which are listed in Supplementary Tables 1 and 2. 3115–3764, 3854–4089, 4178–4304, 5480–5616, The exon/intron structure for Arabidopsis,riceand 5729–5888, 6119–6397, 6923–>7129, and suppor- Chlamydomonas oxygenase genes was determined by ted by cDNA AAC49676, and ESTs AI979621, pairwise comparison of genomic and EST sequence AW065214, AW289039, BG265514. A. thaliana information. The gene structure predicted in Genbank Ptc52 gene At4g25650 (AL050400); join comple- files did not always match that determined by our ana- ment bases 13885–13370, 13246–13059, 12977– lysis. The actual coding regions used in Figure 3 and 12892, 12822–12638, 12555–12367, 12273–12083, the accession numbers of ESTs used in this analysis 12008–11753, and supported by RT-PCR are: A thaliana Cao gene At1g4446 (AF17720050); pYM20-1(AY344062). O. sativa Ptc52 gene (scaf- join bases 1–587, 715–951, 1222–1326, 1405– fold AAAA01007078); join complement bases 5663– 1680, 1763–1910, 1986–2170, 2259–2371, 2448– 5151, 5072–4885, 4668–4566, 4449–4295, 4191– 2742, 2826–2990, and supported by EST AB021316. 3391, 3870–3683, 3514–3256, and supported by O. sativa Cao1 gene (scaffold AAAA01000620); ESTs AU181024, AU057990, BM421452. A. thali- join complement bases 27477–27403, 26860–26618, ana Cmo gene At4g29890 (BAC clone F27B13) 26443–26339, 26257–25979, 25883–25736, 25595– join bases 33446–33743, 33822–33867, 33945– 25411, 25298–25186, 25065–24771, 24618–24436, 34080, 34174–34248, 34327–34542, 34627–34722, and supported by ESTs AF284781, AB021310, 34834–34955, 35036–35129, 35125–35265, 35349– D46313, D48707, AU095684, C24864, AU225916, 35483, and supported by ESTs BE526553, AI996605, and BU672866 (Unigene Os.22029). O. sativa Cao2 AV815748, AY090377. O. sativa Cmo gene (scaffold gene (scaffold AAAA01000620); join complement AAAA01000729); join complement bases 36454– bases 18299–18206, 17823–17578, 17214–17110, 36329, 35650–35605, 35390–35255, 34793–34719, 17034–16756, 16672–16525, 16111–159227, 15812– 34633–34426, 34317–34222, 33942–33824, 33726– 15700, 15612–15318, 15167–14985, and suppor- 33633, 33171–33122, 32673–32529. ted partially by EST BI805076. C. reinhardtii Cao gene (scaffold 8); join complement bases 3804– Multiple sequence alignment and phylogenetic tree 33745, 33529–33200, 33003–32506, 32343–31840, analysis and supported by cDNA AB015139 and ESTs AB015139, AV629131, BG 846842. A thaliana Protein sequences were aligned with Clustal V within Tic55 gene At2g24280 (AC006585); join comple- the Megalign program (DNAStar, Madison, WI) and ment bases 90163–91284, 91366–91468, 91560– with PAM250 residue weights. The PAUP 4.0b10 91954, and supported by ESTs NM 128041, program (Swofford, 2002) was employed for the gen- AY089083, BE529873, AV442063, BE038210, eration of phylogenetic trees and consensus clado- AV439633, AI995341, BE528579, AV531633. O. grams with distance and parsimony optimality cri- 44

and the Arabidopsis and rice genomes. A survey of these databases revealed that no more than five genes in both Arabidopsis and rice contain the signature Rieske (Motif A) and mononuclear (Motif B) iron- binding motifs exhibited by LLS1(PAO) (Gray et al., 2002). The Arabidopsis Acd1 (At3g44880) gene is an orthologue of the maize Lls1(Pao) gene (Yang et al., 2004) which, in turn, is 30% identical to the product of the Arabidopsis At4g25650 gene. The At4g25650 Figure 2. Detection of an LLS1(PAO)-like protein in vascular and gene is highly homologous to the 52 kDa PTC52 gene non-vascular plants. Western blot of protein samples isolated from leaf tissue of the indicated species. A 20 µg portion of protein was from barley which is proposed to encode a pchlide b loaded per lane. Blots were probed with an anti-LLS1:MBP fusion synthase activity in the chloroplast membrane (Rein- protein. Arrow indicates the location of the 52 kDa LLS1(PAO) pro- bothe et al., in press). Two other genes are Cao and tein from maize and similarly sized cross-reactive proteins in other Cmo, which have known functions in catalyzing Chl species. b production and choline biosynthesis, respectively, within the chloroplast compartment (Figure 1) (Bur- teria respectively. Node support for these trees was net et al., 1995; Rathinasabapathi et al., 1997; Tanaka evaluated with the bootstrap method and was per- et al., 1998; Espineda et al., 1999). There is a direct formed for 500 replicates. For parsimony trees all duplication of the Cao gene in rice (Cao2) but it is characters were weighted equally. Starting trees were not clear if this is a functional gene or a pseudogene obtained by random stepwise addition, and the tree- (predicted coding region in Materials and methods). bisection-reconnection algorithm was used for branch ThefifthgeneisTic55, which encodes an unknown swapping. For distance trees the distance measure function in the inner chloroplast membrane but has was mean character difference; starting trees were been found to be loosely associated with protein trans- obtained by neighbor joining and the tree-bisection- port complex proteins (Caliebe et al., 1997). Pairwise reconnection algorithm. Bootstrap analysis with the comparison of the coding regions of these genes shows PAUP 4.0b10 program was performed on a Macintosh that each of the family members is conserved between G3 workstation or at the Ohio Supercomputer Cluster monocots and dicots, and that the five members are (www.osc.edu). distinct from one another (Figure 3A). The completed Arabidopsis and rice plant genomic sequences were used to compare the intron/exon struc- Results and discussion ture of these genes so as to discern if any of these genes are the result of a recent duplication event. The ex- Lls1(Pao) belongs to a small family of non-heme act positions of introns were determined by alignment oxygenases in plants. of genomic DNA sequences with extant ESTs from various databases (Figure 3B). In the case of A. thali- The relationship between the maize Lls1(Pao) genes in ana Lls1(Pao) and Ptc52 genes, a full-length cDNA different species was examined. An anti-LLS1(PAO) was isolated and sequenced by RT-PCR. For three of monoclonal antibody was used to detect a unique the five pairs of homologous genes (Lls1(Pao), Cao related protein not only in a selection of dicotyle- and Cmo) the gene structure was conserved between donous woody and herbaceous plants but also in the rice and Arabidopsis. In the case of the Chlamydo- more primitive vascular plant Equisetum and the non- monas Cao gene on scaffold 8 (predicted gene 8.14), vascular moss Polytrichum (Figure 2). These results for which a cDNA clone is known, the intron/exon indicate that an LLS1(PAO)-like protein is probably positions are also not shared with land plants (Fig- present universally in land plants which is consistent ure 3B) as anticipated given the ancient origin of this with its reported role in the opening of the enzyme function (Tomitani et al., 1999). In the case ring during chlorophyll detoxification (Hortensteiner, of Tic55,theArabidopsis gene exhibits just two in- 1999; Hortensteiner et al., 2000). trons and there are none in rice. The Arabidopsis and The phylogenetic relationship of the Lls1(Pao) rice Ptc52 genes differ slightly in the size of exons gene to other plant genes was then examined by per- three and four but the positions of the introns are forming homology searches with plant EST databases conserved. However pairwise alignments of the exon 45

Figure 3. A. Pairwise sequence identity values of Arabidopsis and rice non-heme oxygenases. The coding regions of the plant genes depicted in A (minus predicted transit peptides) were aligned with Clustal W (default parameters) and pairwise distance values tabulated as shown. B. Gene structure of plant and algal non-heme oxygenases. Gene structure is shown as exons (blocks) joined by introns (lines). The conserved Rieske and mononuclear iron-binding motifs are shown as hatched and shaded boxes. A third conserved motif of unknown function is shown as a dotted box. Scale as shown. For intron/exon boundary and sequence information, see Materials and methods. sequences did not reveal any conservation of intron A preliminary examination of the Chlamydomonas re- position between the five pairs of genes. This observa- inhardtii draft genome reveals at least six non-heme tion is particularly obvious in the vicinity of conserved oxygenases other than Cao (predicted genes 37.24, motifs (Figure 3B). The conservation of intron posi- 37.26, 41.13, 327.2, 2597.4, 282.0) with signific- tions between the same gene family members but not ant homology to Lls1(Pao), Tic55 and Ptc52 (but not between different gene family members indicates that Cmo). Two of these genes appear to be the result all of these five genes had existed prior to monocot and of a recent duplication event (predicted genes 37.24 dicot divergence about 200 million years ago and thus and 37.26 on Scaffold 37) like the Cao1andCao2 are not the result of a recent duplication event in land genes in rice (data not shown). The absence of cD- plants. It is known that (LLS1)PAO and CAO activ- NAs to confirm gene structure precluded the analysis ities are present in the algae Chlorella protothecoides of these genes in this study. The detection of these indicating a more ancient origin for these gene func- genes in a eukaryotic alga, however, suggests that the tions (Hortensteiner, 1999; Hortensteiner et al., 2000). 46 roles of these genes evolved prior to the transition to a In addition, the plant proteins exhibited transit peptide terrestrial habitat. sequences that were not present in bacterial proteins. In a region of each protein comprising the two highly Conservation of iron-binding centers in conserved iron binding motifs and the variable inter- LLS1(PAO)-related non-heme oxygenases vening region, it was feasible to align all 95 proteins with the ClustalW algorithm program (Figure 4A–B Homologues of LLS1(PAO)-like plant non-heme oxy- and Supplementary Figure 1). An examination of the genases are also found in cyanobacteria indicating that pairwise distance values indicated that it was possible these gene functions may have existed prior to the to place the bacterial dioxygenases into the same four evolution of algae (Gray et al., 1997, 2002). This groupings as Nam although different percent similar- is clearly true for the Cao gene, which has been ities were used as cutoff points (data not shown). In demonstrated to be present in the Chl b-containing agreement with the Nam classification system all pro- cyanobacteria Prochloron didemniii and Prochloron teins exhibited a Rieske-type [2Fe-2S] cluster binding hollandica (Moreira et al., 2000). However, the hy- site that could be expressed as Cys-X1-His-X16−18- pothesis that the Cao gene has a monophyletic origin Cys-X2-His (Figure 4A). In addition, an arginine for all chlorophyll b-producing organisms (Tomitani immediately adjacent to the first histidine, and two et al., 1999) has been questioned by the absence of aromatic residues in the vicinity of the second his- a convincing Cao homologue in the genomes of two tidine, were nearly universally conserved within this Prochlorococcus strains (Hess et al., 2001). A single motif (Figure 4A). The mononuclear Fe2+-binding non-heme oxygenase with Rieske and mononuclear site in ARH enzymes can be expressed as Glu-X3−4- motifs is encoded by these genomes but its relation- Asp-X2-His-X3−5-His (Jiang et al., 1996; Nam et al., ship to Cao or other non-heme oxygenases has not 2001). According to the Nam classification system, the been defined (Hess et al., 2001). plant oxygenases CAO and TIC55 exhibit a spacing of To define the evolutionary and possible functional residues in both motifs that matches that of vanillate relationships of plant LLS1(PAO)-like oxygenases to demethylases such as VanA(19151) in Nam group 1 other non-heme oxygenases, a phylogenetic analysis (Figure 4B and Supplementary Figure 1). The spacing of 95 related proteins from a wide spectrum of spe- of residues within the mononuclear Fe2+-binding mo- cies was performed. Reiterative BLAST searches re- tif for LLS1(PAO) and PTC52 is also similar to Nam vealed that the LLS1(PAO)-like oxygenases share the group 1 proteins, except that there is a 17 residue iron-binding motifs exhibited by the oxygenase com- spacing at the center of the Rieske motif which is ponents of multicomponent bacterial aromatic ring- seen in Nam groups 2, 3 and 4 and gene 7970 from hydroxylating (ARH) enzymes (Pfam00484) (Neidle Burkholderia fungorum. et al., 1991). The Nam classification system categor- In contrast, the residue spacings and alignments ized 54 different ARH enzymes into 4 groups based within these motifs for CMO oxygenases do not match on pairwise alignment scores and the separation of any of the Nam groupings precisely. In particular the key amino acid residues in the consensus Rieske type CMO group is unique amongst all the oxygenases ex- and mononuclear-type iron-binding sites (Nam et al., amined in having three residues between the aspartate 2001). Plant and cyanobacterial LLS1(PAO)-like oxy- and the first histidine residue of the mononuclear iron- genases were not included in the development of that binding motif (Figure 4B). In addition, plant CMO Nam classification system, but this sequence-based proteins are distinguished from other plant oxygenases classification system easily facilitates the phylogenetic in that they each exhibit the NWK triplet that is al- comparison with novel related proteins (where inform- ways conserved 7 residues prior to motif B in Nam ation on subunit composition is not yet known). At groups 2, 3 and 4 (Figure 4B and Supplementary least four members from each of the ARH groups were Figure 1). An examination of the crystal structure selected for inclusion in our phylogenetic analysis and of NDO indicates that the NWK triplet is between designated NG1 (Nam Group 1) to NG4 following the I191 and P198 residues that line the pocket be- their names in Figures 4–7. low the active site (Kauppi et al., 1998). A number In order to meaningfully align a larger set of 95 of other predicted oxygenases from bacteria also do oxygenases from a broader variety of species than has not fit easily into four groupings of the Nam classi- been previously considered, an operational taxonomic fication system including gene 1092 from Ralstonia unit that was less than the entire protein was chosen. metalidurans, gene 2224 from Ralstonia solancearum 47

Figure 4. Multiple sequence alignment of conserved iron-binding sites of selected representative plant and bacterial non-heme oxygenases (an alignment of all 95 oxygenases used in this study is provided in Supplementary Figure 1). The alignment was performed with a region of each + protein spanning both Rieske [2Fe-2S] and mononuclear Fe2 iron-binding motifs as the operational taxonomic unit (start and end residues are shown and the distance between motifs is shown in the central column). Sequences were aligned with the CLUSTAL W program and a PAM250 weight matrix, a gap penalty of 8, and a gap length penalty of 10. Other settings were default. Strictly conserved residues are shaded gray and other highly conserved residues are shown in reverse type. Proteins are listed according to their gene name or gene number (Supplementary Tables 1 and 2), followed by the abbreviated species and strain number in parenthesis. Plant oxygenases are shaded black and cyanobacterial oxygenases are shaded gray. 48

Figure 5. Evolutionary relationships among plant, cyanobacterial and bacterial proteins containing Rieske and mononuclear iron-binding motifs. Partial phylogenetic tree of 95 non-heme oxygenases estimated with the weighted neighbor joining distance method. The excluded tree branches are shown in Figure 6 and are connected to this figure by a dotted line. Phylogenetic tree was estimated from the alignment represented in Supplementary Figure 1. Branch lengths are proportional to the expected number of amino acid substitutions per site (values shown above center of each branch; for parameters, see Materials and methods). The reliability of each bifurcation was estimated using bootstrap analysis (percentage values over 50% are shown encircled next to nodes, values less than 50% are not shown), and the support for each of the branches is indicated by line thickness. The tree is unrooted with OXOO and CARAa as a monophyletic outgroup. Plant oxygenases are shaded black and cyanobacterial oxygenases are shaded gray.

GMI1000, gene 4529 from Burkholderia fungorum trees were estimated using both distance and parsi- and gene 1360 from Novosphingobium europea. mony criteria (Figures 5 and 6 and Supplementary Figure 1). A comparison between trees obtained using Lls1(Pao)-like genes are related to non-heme both of these methods is useful in discerning reliable oxygenases in aerobic photosynthesizers phylogenetic relationships (Hall, 2001). A compar- ison of the two trees obtained in this study shows In order to characterize the evolutionary relationships correlative branching, and most unreliable bifurca- between the aforementioned proteins, phylogenetic 49

Figure 6. Evolutionary relationships among cyanobacterial and bacterial proteins containing Rieske and mononuclear iron-binding motifs. Partial phylogenetic tree of 96 non-heme oxygenases estimated with the weighted neighbor joining distance method. Tree branches shown are connected to those in figure 5 by a dotted line. Phylogenetic trees were estimated from the alignment represented in Supplementary Figure 1. Plant oxygenases are shaded black and cyanobacterial oxygenases are shaded gray. tions (bootstrap values <50%) seen in the distance teins. Likewise, TIC55 and CAO homologues each tree (Figures 5 and 6) are reduced to polytomies in form clades that include two cyanobacterial proteins. the parsimony tree (Supplementary Figure 1). Us- The failure of the single dioxygenase from Prochloro- ing a distance-based method for tree construction re- coccus species strains CCMP1378 and MIT9313 to capitulated the groupings determined by Nam et al. cluster near CAO is in agreement with the obser- (2001) for bacterial ARH enzymes (Figures 5 and vations of Hess et al. (2001), who propose that 6). Nam groups 2, 3 and 4 form distinct clades sup- theseenzymesmayhavearisenbyconvergentevol- ported by reliable forks with subclades that include ution. Enzymes related to vanillate demethylase form carbazole (CarAa), dioxin (DxnA1), and aniline di- a clade that does not include any plant or cyanobac- oxygenase (Tdn1) (Figure 6). The inclusion of more, terial enzymes. The remainder of this major group recently discovered ARH-type enzymes reveals that include four cyanobacterial genes that may encode a most cluster within a major division of all the en- chlorobenzoate dioxygenase-like function and a fifth zymes that includes Nam group I ARHs as evidenced cyanobacterial enzyme related to a dioxygenase in- by a basal polytomy for this major clade (Supplement- volved in the synthesis of the aromatic ring-containing ary Figure 2). This group includes a diverse set of antibiotic pyrrolnitrin encoded by Pseudomonas and enzymes to which the majority of plant and cyanobac- Myxococcus species. terial oxygenases are more closely related. Within this larger group, LLS1(PAO) and PTC52 form a distinct grouping that includes 11 cyanobacterial pro- 50

Figure 7. Evolutionary relationships among LLS1(PAO)-like plant and cyanobacterial non-heme oxygenases. A. Unrooted phylogram of concensus tree of 25 LLS1(PAO)-related oxygenases estimated using neighbor joining distance method and bootstrap analysis. A total of 25 non-heme oxygenases containing the motif in described in B were aligned with the latter 75% of each protein length as the operational taxonomic unit (the Rieske motif to the carboxyl terminus). Branch lengths are proportional to the expected number of amino acid substitutions per site (values shown next to each branch). The reliability of each bifurcation was estimated using bootstrap analysis (percentage values over 50% are shown encircled next to nodes, values less than 50% are not shown). B. A novel conserved motif defines a set of closely related LLS1(PAO)-like proteins in cyanobacteria and plants. Examination of phylogenetic trees (Figure 3) revealed a clade of proteins related to LLS1(PAO) that share a concensus motif that can be defined as D/E/N-x-F/Y/W-x2-H-x3-C-x2-C. This motif is found at a common distance of 82–84 amino acids from the carboxyl terminus of these proteins. The last sequence is from a 110 amino acid ORF from Nostoc (all0986), in which the motif is found near the center of the predicted protein. Plant oxygenases are shown in white text against a black background and cyanobacterial oxygenases are shown in black text against a gray background.

A separate origin for CMO amongst non-heme closely related to Nam group1 oxygenases whereas oxygenases of soil-dwelling bacteria CMO is more closely related to oxygenases from Nam Groups 2 to 4. Within the second large clade CMO The distance tree structure indicates that plant CAO, forms a small clade that does not include any strong LLS1(PAO), TIC55 and PTC52 proteins are more bacterial homologues which reflects the fact that the 51 use of a dioxygenase in the first step of glycine betaine Rieske motif to the end of each protein as the opera- synthesis is provided by alternative enzymes in bac- tional taxonomic unit (amino terminus sequences that teria (Rathinasabapathi et al., 1997; Incharoensakdi contain choroplast signal peptides were not present in and Wutipraditkul, 1999). None of the six Lls1(Pao)- the cyanobacterial homologues). An unrooted phylo- related genes from C. reinhardtii genes show signi- genetic tree estimated with a distance method reveals ficant homology to Cmo (data not shown) which was a largely equidistant relationship between these cy- anticipated because this algae synthesizes glycerol anobacterial and plant proteins (Figure 7A). Plant and not glycine betaine in response to hyperosmotic LLS1(PAO) and PTC52 proteins are slightly more re- shock (Rosa and Galvan, 1995). Strong homologues lated to gene 1779 from the marine cyanobacterium of CMO were not identified in any cyanobacterial spe- Trichodesmium erythraeum and gene 4354 from the cies, and gene 1261 from Synechococcus sp WH8102 freshwater cyanobacterium Nostoc sp. PCC7120 than is the only cyanobacterial protein that falls within this they are to the clade that includes gene 1747 from Syn- second major clade. Gene 1261 is the only cyanobac- echocystis sp. PCC6803. The plant TIC55 proteins are terial oxygenase that exhibits the NWK tripeptide more related to gene 5007 from Nostoc sp. PCC7120 immediately prior to motif B (Figure 4), but other- and gene 2180 from Nostoc punctiforme.Theremain- wise it does not show strong homology to CMO. It ing five proteins from cyanobacteria form two clades appears that CMO, whose is not aromatic, each containing a member from Trichodesmium eryth- has an evolutionary origin distinct from all other plant raeum and Nostoc sp. PCC7120. The short basal oxygenases and its closer relatedness to bacterial ARH branches at the center of this tree indicate that these enzymes is suggestive of the recruitment of an oxy- 25 proteins are closely related and likely to be derived genase upon horizontal transfer from a soil-dwelling from a common ancestor. In support of this interpret- bacterium to a plant. The existence of the unique Syn- ation is the existence within the extended terminus of echococcus sp. WH8102 gene 1261 and the fact that these proteins of a third motif that is strictly conserved most plants are not known to accumulate significant (Figure 7B). This motif of unknown function contains amounts of glycine betaine despite the presence of a two aromatic amino acids and two cysteine residues Cmo homologue is enigmatic. The reported inability and can be summarized by the consensus sequence of the Arabidopsis Cmo homologue to exhibit CMO F/Y/W-x2-H-x3-C-x2-C (Figure 7B). A search for this activity in Escherichia coli (Hibino et al., 2002) is motif within all non-redundant proteins in Genbank most likely the result of a mutated PCR product used was negative. A survey of the Nostoc genome, how- in that study because the mutated residue I371V is ever, revealed a small ORF (all0956) that is predicted conserved across species and is not observed in extant to contain a 110 amino acid protein that also contains Arabidopsis ESTs. It is possible that glycine betaine this motif (Figure 7b). Closer examination reveals that accumulates at very low levels in many plants, or that this gene may be slightly longer and have further ho- it accumulates only under unusual osmotic stress con- mology to LLS1(PAO) but does not contain the two ditions. Alternatively, it is possible that these plants motifs shared by this group of proteins. Further ana- have alternative mechanisms to resist osmotic stress lysis is required to determine if all0956 is expressed or that these genes encode a related function that has or if it is a relic gene fragment. not yet been discovered. Neither homologues to the Lls1(Pao) gene or pro- teins containing this third motif were identified in a Lls1(Pao), Tic55 and Ptc52 are closely related to survey of the complete genome of the photosynthetic genes in oxygenic cyanobacteria bacterium Chlorobium tepidum TLS. C. tepidum is a thermophilic green sulfur bacterium and is strictly an- Phylogenetic analysis revealed a tight clustering of the aerobic and obligatorily autotrophic. This bacterium plant LLS1(PAO), TIC55 and PTC52 proteins with a synthesizes bacteriochlorophyll c and carotenoids for set of predicted proteins from marine and freshwa- light harvesting but utilizes hydrogen and various sul- ter cyanobacteria (Figure 5). A closer examination of fur compounds as terminal electron acceptors and these proteins revealed that they share an extended not dioxygen. An anaerobic environment negates carboxyl terminus that is ca. 120 amino acids longer the possibility of using dioxygen-requiring enzymes than that in CAO and CMO proteins. The stronger in metabolism but not necessarily the requirement sequence homology between these proteins permitted for redox-sensing proteins. Homologues of Lls1(Pao) the alignment of all 25 sequences in a region from the were found in the genomes of all oxygenic cyanobac- 52 teria examined with the exception of both strains of and their shared motifs are currently being sought Prochlorococcus marinus. Thus, although the pres- by studying knockouts of the homologous proteins in ence of LLS1(PAO)-like proteins correlates strongly cyanobacteria that were identified in this study. with the emergence of oxygenic photosynthesis they are not requisite for photosynthesis itself. These pro- Classification of LLS1(PAO)-related non-heme teins may have evolved to allow aerobic photosynthes- oxygenases based on conserved iron-bonding motifs izers organisms to adapt and become fine-tuned to the wide variety of oxygen and light levels that exist in the The phylogenetic analysis by distance and parsi- water and on land (Ting et al., 2002). All chlorophyll mony methods that we performed supports the re- b-containing organisms are aerobic and the utilization classification of dioxygenases by Nam et al. (2001) of dioxygen for the synthesis of Chl b and pchlide b by and the use of a smaller, more widely conserved CAO and PTC52 were significant adaptations for light operational taxonomic unit was successful in recapit- harvesting by aerobic photosynthesizers (Tanaka et al., ulating the major groupings that they reported. The 2001). Since most of the cyanobacteria examined in analysis of a much larger group of genes indicates this study do not synthesize Chl b, the function of however that further sub-classification may be re- Lls1(Pao)-related genes in these species may indicate quired in future categorization of these enzymes. In a role in the removal of Chl which is a potent photo- particular, it is clear that Nam group I dioxygenases toxin in aerobic environments. It is known that algae include diverse enzymes with non-phenolic substrates exhibit an LLS1/PAO activity but this has not yet been as exemplified by LLS1(PAO), CAO and PTC52. The reported in cyanobacteria which may use excretion as phylogenetic tree that was estimated using parsimony the mechanism of chlorophyll detoxification (Miller (Supplementary Figure 1) shows 15 clades branching and Holt, 1977; Richaud et al., 2001). Finally, the from a common polytomy that includes all group I fact that 3 out of 4 Lls1(Pao)-related genes appear enzymes (including CarAa and OxoO). Newly dis- to have a Chl related function is suggestive that the covered members of this large group may be assigned TIC55 protein also plays a hitherto unsuspected role a category based on clustering nearest to a group of in Chl metabolism which is not mutually exclusive proteins with a known substrate type. Thus, class from a proposed redox regulatory role in the chloro- IA would include enzymes clustering near genes en- plast (Küchler et al., 2002). One possibility is that coding vanillate demethylase (VanA), chlorobenzoate TIC55 catalyzes Chl b to Chl a conversion – an en- dioxygenase(CbaA) or phthalate dioxygenase (Pht3), zyme activity that has been reported in plants but for which act on phenolic compounds. This class would which the corresponding gene has not been isolated include enzymes such as aminopyrrolnitrin D oxi- (Rudoi and Shcherbakov, 1998; Beale, 1999). Chl dase (PrnD), which catalyzes oxidation of the amino degradation products may also serve as intracellular side group of the aromatic ring of aminopyrrolnitrin signals during development and stress responses and to a nitro group in the biosynthesis of the antibi- thus indirectly influence the import of proteins into otic pyrrolnitrin. Indeed, many of the dioxygenases in the chloroplast (Küchler et al., 2002). The finding in this group have been shown to have broad specificity, this study that TIC55, LLS1(PAO) and PTC52 share but are unified in that they act on phenolic com- a common C-terminus motif with conserved cysteine pounds. Subclass IA would include carbazole 1,9a- residues may lend further support to the hypothesis of dioxygenase (from Pseudomonas strain CA10) and Küchler et al. (2002). Redox-regulated proteins ex- 2-oxo-1,2-dihydroquinoline 8-monooxygenase (from hibit conserved cysteine residues (Kobayashi and Ito, Pseudomonas strain 86). It was found that these two 1999; Kuge et al., 2001), and a recent proteomics enzymes grouped separately from other group I en- study identified many previously unsuspected targets zymes when using the shorter amino acid sequence of thioredoxin regulation in the chloroplast stroma as operational taxonomic subunit, but they were more (Yano et al., 2002; Balmer et al., 2003). Although clearly classified when the full-length protein was used the C-terminus motifs of LLS1(PAO), TIC55 and in the analysis. Subclass 1B would include CAO en- PTC52 exhibit a CxxC motif conserved in the cata- zymes from various sources which use chlorophyll a lytic site of thiol:disulfide (Chivers (a cyclic tetrapyrrole) as a substrate. The unique oxy- et al., 1997), the surrounding amino acids do not genases from Prochlorococcus strains CCMP1378 and suggest a thioredoxin-like fold. Further insights into MIT9313, are tentatively assigned in this subclass if the functions of LLS1(PAO)-related proteins in plants they can be shown to act on divinyl Chl a as a sub- 53 strate. Prochlorococcus, like other prochlorophytes, is extensive phylogenetic analysis of LLS1(PAO) related defined by its production of Chl b, and it has been non-heme oxygenases provides a useful reference for proposed that there is a single origin of CAO (Tom- future classification of these enzymes. itani et al., 1999). Our alignments suggest that these oxygenases may have originated from a Nam group I- type enzyme, but convergent evolution from a Rieske Acknowledgements domain protein is a plausible alternative explanation. Prochlorococcus may have evolved in iron-limited We thank Scott Leisner for helpful discussion and ad- oceans and diversified from other cyanobacteria from vice on this manuscript. Funding for this research was a common phycobilisome-containing ancestor (Ting provided by the U.S. Department of Agriculture (grant et al., 2002). The fact that these species produce un- 2000–01465 to J.G.) and by the University of Toledo usual divinyl Chl derivatives in order to occupy a dif- (laboratory startup funds to J.G.). We thank Stefan ferent ecological niche could account for the observed Hörtensteiner (University of Bern, Switzerland) for sequence differences in a CAO-like enzyme. sharing results prior to publication. Subclass IC could include all LLS1(PAO) related enzymes (including PTC52 and TIC55) that exhibit the third motif identified in Figure 7B until definitive References substrates are defined for all these enzymes. If these enzymes all share a chlorophyll intermediate as sub- Altschul, S.F., Madden, T.L., Schäffer, A.A., Zhang, J.H., Zhang, Z., Miller, W. and Lipman, D.J. 1997. Gapped BLAST and PSI- strate then it may be appropriate to combine them with BLAST: a new generation of protein database search programs. CAO in subclass 1B. It is clear that CMO-like proteins Nucl. Acids Res. 25: 3389–3402. form a distinct clade separate from Nam Groups I to Balmer, Y., Koller, A., del Val, G., Manieri, W., Schurmann, P. and IV and we propose a new group V (Figure 6), whose Buchanan, B.B. 2003. Proteomics gives insight into the regulat- ory function of chloroplast thioredoxins. Proc. Natl. Acad. Sci. members are closely related to the monooxygenase USA 100: 370–375. from spinach, which uses choline as substrate. Finally, Batie, C.J., Ballou, D.P. and Correll, C.C. 1991. Phthalate di- a clade of 9 oxygenases with unknown substrates that oxygenase reductase and related flavin-iron sulfur containing also group separately from Nam groups I to IV may electron . Chem. Biochem. Flavoenzymes 3: 543– 556. be designated as class VI (Figure 6). These sugges- Beale, S.I. 1999. Enzymes of chlorophyll biosynthesis. Photosyn. ted classifications build on the groupings defined by Res. 60: 43–73. Nam et al. (2001) using phylogenetic analysis, which Burnet, M., Lafontaine, P.J. and Hanson, A.D. 1995. Assay, puri- continues to be a useful classification system in the fication and partial characterization of choline mononxygenase from spinach. Plant Physiol. 108: 581–588. absence of subunit or substrate information. Caliebe, A., Grimm, R., Kaiser, G., Lubeck, J., Soll, J. and Heins, L. 1997. The chloroplastic protein import machinery contains a Conclusions Rieske-type iron-sulfur cluster and a mononuclear iron-binding protein. EMBO J. 16: 7342–7350. Cheng, W.-H., Im, K.H. and Chourey, P.S. 1996. Sucrose phos- In conclusion, our analysis indicates that LLS1(PAO) phate synthase expression at the cell and tissue level is co- belongs to a small but diverse group of non-heme ordinated with sucrose sink-to-source transitions in maize leaf. oxygenases in plants. Of these, CMO homologues Plant Physiol. 111: 1021–1029. have evolved recently in higher plants although their Chivers, P.T., Prehoda, K.E. and Raines, R.T. 1997. The CXXC motif: a rheostat in the active site. Biochemistry 36: 4061–4066. conservation in plants such as Arabidopsis,which Espineda, C.E., Linford, A.S., Devine, D. and Brusslan, J.A. 1999. is not known to accumulate glycine betaine, sug- The AtCAO gene, encoding chlorophyll a oxygenase, is required gests another alternative function in most plants. In for chlorophyllb synthesis in Arabidopsis thaliana. Proc. Natl. contrast, LLS1(PAO)-related non-heme oxygenases Acad. Sci. USA 96: 10507–10511. Gray, J., Close, P.S., Briggs, S.P. and Johal, G.S. 1997. A novel including PTC52, TIC55 and CAO appear to have suppressor of cell death in plants encoded by the Lls1 gene of evolved amongst oxygenic cyanobacteria that synthes- maize. Cell 89: 25–31. ize chlorophyll but are not required in all such species Gray, J., Janick-Buckner, D., Buckner, B., Close, P.S. and Johal, such as Prochlorococcus marinus. The presence of G.S. 2002. Light-dependent death of maize lls1 cells is mediated by mature chloroplasts. Plant Physiol. 130: 1–14. a novel C-terminus motif with conserved cysteine Hall, B.G., 2001. Phylogenetic Trees Made Easy: A How-To residues suggests that LLS1(PAO), TIC55 and PTC52 Manual for Molecular Biologists. Sinauer, Sunderland, MA. may share similar redox-regulation connected with Harayama, S., Kok, M. and Neidle, E.L. 1992. Functional and evol- chlorophyll metabolism in the chloroplast. Finally, our utionary relationships among diverse oxygenases. Annu. Rev. Microbiol. 46: 565–601. 54

Hess, W.R., Rocap, G., Ting, C.S., Larimer, F., Stilwagen, S., Leam- Nam, J.-W., Nojiri, H., Yoshida,T.,Habe,H.,Yamane,H.and erdin, J. and Chisholm, S.W. 2001. The photosynthetic apparatus Omori, T. 2001. New classification system for oxygenase com- of Prochlorococcus: insights through comparative genomics. ponents involved in ring-hydroxylating oxygenations. Biosci. Photosyn. Res. 70: 53–71. Biotechnol. Biochem. 65: 254–263. Hibino, T., Waditee, R., Araki, E., Ishiwaka, H., Kenji, A., Tanaka, Neidle, E.L., Hartnett, C., Ornston, L.N., Bairoch, A., Rekik, M. Y. and Takebe, T., 2002. Functional characterization of choline and Harayama, S. 1991. Nucleotide sequences of the Acinetobac- monooxygenase, an enzyme for betaine synthesis in plants. J. ter calcoaceticus benABC genes for benzoate 1,2-dioxygenase Biol. Chem. 277: 41352–41360. reveal evolutionary relationships among multicomponent oxy- Hortensteiner, S., 1999. Chlorophyll breakdown in higher plants and genases. J. Bact. 173: 5385–5395. algae. Cell. Mol. Life Sci. 56: 330–347. Prescott, A.G. and Lloyd, M.D. 2000. The iron(II) and 2-oxoacid- Hortensteiner, S., Chinner, J., Matile, P., Thomas, H. and Donnison, dependent dioxygenases and their role in metabolism. Nat. Prod. I.S. 2000. Chlorophyll breakdown in Chlorella protothecoides: Rep. 17: 367–383. characterization of degreening and cloning of degreening-related Rathinasabapathi, B., Burnet, M., Russell, B.L., Gage, D.A., Liao, genes. Plant Mol. Biol. 42: 439–450. P.-C., Nye, G.J., Scott, P., Golbeck, J.H. and Hanson, A.D. 1997. Incharoensakdi, A. and Wutipraditkul, N. 1999. Accumulation of Choline monooxygenase, an unusual iron-sulfur enzyme catalyz- glycinebetaine and its synthesis from radioactive precursors un- ing the first step of glycine betaine synthesis in plants: prosthetic der salt-stress in the cyanobacterium Aphanothece halophytica. group characterization and cDNA cloning. Proc. Natl. Acad. Sci. J. Appl. Phycol. 11: 515–523. USA 94: 3454–3458. Jiang, H., Parales, R.E., Lynch, N.A. and Gibson, D.T., 1996. Site- Richaud, C., Zabulon, G., Joder, A. and Thomas, J.-C., 2001. Ni- directed mutagenesis of conserved amino acids in the alpha sub- trogen or sulfur starvation differentially affects phycobilisome unit of toluene dioxygenase: potential mononuclear non-heme degradation and expression of the nblA gene in Synechocystis iron coordination sites. J. Bact. 178: 3133–3139. strain PCC 6803. J. Bact. 183: 2989–2994. Karlsson, A., Parales, J.V., Parales, R.E., Gibbon, D.T., Eklund, Rosa, L. and Galvan, F. 1995. Metabolic pathway for glycerol H. and Ramaswamy, S. 2003. Crystal structure of naphthalene synthesis under osmotic stress in the freshwater green alga dioxygenase: side-on binding of dioxygen to iron. Science 299: Chlamydomonas reinhardtii. Plant Physiol. Biochem. 33: 213– 1039–1042. 218. Kauppi, B., Lee, K., Carredano, E., Parales, R., Gibson, D.T., Rudoi, A.B. and Shcherbakov, R.A., 1998. Analysis of the chloro- Eklund, H. and Ramaswamy, S. 1998. Structure of an aromatic- phyll biosynthetic system in a chlorophyll b-less barley mutant. ring-hydroxylating dioxygenase – naphthalene 1,2 dioxygenase. Photosyn. Res. 58: 71–80. Structure 6: 571–586. Ryle, M.J. and Hausinger, R.P. 2002. Non-heme iron oxygenases. Kobayashi, T. and Ito, K., 1999. Respiratory chain strongly oxidizes Curr. Opin. Chem. Biol. 6: 193–201. the CXXC motif of DsbB in the Escherichia coli disulfide bond Tanaka, A., Ito, H., Tanaka, R., Tanaka, N., Yoshida, K. and Okada, formation pathway. EMBO J. 18: 1192–1198. K. 1998. Chlorophyll a oxygenase is involved in chlorophyll b Küchler, M., Decker, S., Hörmann, F., Soll, J. and Heins, L. formation from chlorophyll a. Proc. Natl. Acad. Sci. USA 95: 2002. Portein import into chloroplasts involves redox-regulated 12719–12723. proteins. EMBO J. 21: 6136–6145. Tanaka, R., Koshino, Y., Sawa, S., Ishiguro, S., Okada, K. Kuge, S., Arita, M., Murayama, A., Maeta, K., Izawa, S., Inoue, and Tanaka, A. 2001. Overexpression of a oxy- Y. and Nomoto, A. 2001. Regulation of the yeast Yap1p nuc- genase (CAO) enlarges the antenna size of photosystem II in lear export signal is mediated by redox signal-induced reversible Arabidopsis thaliana. Plant J. 26: 365–373. disulfide bond formation. Mol. Cell. Biol. 21: 6139–6150. Ting, C.S., Rocap, G., King, J. and Chisholm, S.W. 2002. Lange, S. J. and Que, L. Jr. 1998. Oxygen activating nonheme iron Cyanobacterial photosynthesis in the oceans: the origins and enzymes. Curr. Opin. Chem. Biol. 2: 159–172. significance of divergent light-harvesting strategies. Trends Mi- Mason, J.R. and Cammack, R. 1992. The electron-transport proteins crobiol. 10: 134–142. of hydroxylating bacterial dioxygenases. Annu. Rev. Microbiol. Tomitani, A., Okada, K., Miyashita, H., Matthijs, H.C. and Ohno, 46: 277–305. T. 1999. Chlorophyll b and in the common ancestor Miller, L.S. and Holt, S.C. 1977. Effect of carbon dioxide on pig- of cyanobacteria and chloroplasts. Nature 400: 159–162. ment and membrane content in Synechococcus lividus.Arch. Yang, M., Wardzala, E., Johal, G.S. and Gray, J. 2004. The Microbiol. 115: 185–198. wound-inducible Lls1 gene from maize is an ortholog of the Moraswki, B., Segura, A. and Ornston, L.N. 2000. Substrate range Arabidopsis Acd1 gene, and the LLS1 protein is present in and genetic analysis of Acinetobacter vanillate demethylase. J. non-photosynthetic tissues. Plant Mol Biol. 54: 175–191. Bact. 182: 1383–1389. Yano, H., Kuroda, S. and Buchanan, B.B. 2002. Disulfide proteome Moreira, D., Guyader, H.L. and Phillipe, H. 2000. The origin of red in the analysis of protein function and structure. Proteomics 2: algae and the evolution of chloroplasts. Nature 405: 69–72. 1090–1096.