Sequences Required for the Activity of PTOX (IMMUTANS), a Plastid Terminal Oxidase

The thylakoid membranes of most photosynthetic organisms contain a terminal oxidase (PTOX, the product of the Arabidopsis IMMUTANS gene) that functions in the oxidation of the plastoquinone pool. PTOX and AOX are diiron carboxylate proteins, and based on crystal structures of other members of this protein class, a structural model of PTOX has been proposed in which the ligation sphere of the diiron center is composed of six conserved histidine and glutamate residues. We tested the functional significance of these residues by site-directed mutagenesis of PTOX in vitro and in planta, taking advantage null immutans alleles for the latter studies. These experiments showed that the six iron-binding sites do not tolerate change, even conservative ones. We also examined the significance of a conserved sequence in (or near) the PTOX active site that corresponds precisely to Exon 8 of the IM gene. In vitro and in planta mutagenesis revealed that conserved amino acids within this domain can be altered but that deletion of all or part of the domain abolishes activity. Because protein accumulates normally in the deletion mutants, the data suggest that the conformation of the Exon 8 sequence is important for PTOX activity. An allele of immutans (designated 3639) was identified that lacks the Exon 8 sequence; it does not accumulate PTOX protein. Chloroplast import assays revealed that mutant enzymes lacking Exon 8 have enhanced turnover. We conclude that the Exon 8 domain is required not only for PTOX activity but also for its stability.

The thylakoid membranes of most photosynthetic organisms contain a terminal oxidase (PTOX, the product of the Arabidopsis IMMUTANS gene) that functions in the oxidation of the plastoquinone pool. PTOX and AOX are diiron carboxylate proteins, and based on crystal structures of other members of this protein class, a structural model of PTOX has been proposed in which the ligation sphere of the diiron center is composed of six conserved histidine and glutamate residues. We tested the functional significance of these residues by site-directed mutagenesis of PTOX in vitro and in planta, taking advantage null immutans alleles for the latter studies. These experiments showed that the six iron-binding sites do not tolerate change, even conservative ones. We also examined the significance of a conserved sequence in (or near) the PTOX active site that corresponds precisely to Exon 8 of the IM gene. In vitro and in planta mutagenesis revealed that conserved amino acids within this domain can be altered but that deletion of all or part of the domain abolishes activity. Because protein accumulates normally in the deletion mutants, the data suggest that the conformation of the Exon 8 sequence is important for PTOX activity. An allele of immutans (designated 3639) was identified that lacks the Exon 8 sequence; it does not accumulate PTOX protein. Chloroplast import assays revealed that mutant enzymes lacking Exon 8 have enhanced turnover. We conclude that the Exon 8 domain is required not only for PTOX activity but also for its stability.
The IMMUTANS (IM) 2 gene of Arabidopsis thaliana codes for a plastid membrane protein that is distantly related to the alternative oxidase (AOX) of mitochondrial inner membranes (1,2). AOX is a ubiquinol oxidase that catalyzes the four-electron reduction of oxygen to water and branches from the cytochrome pathway at the quinone pool (3,4). AOX is found in all plants and in some fungi and protozoa, and it is thought to provide an overflow for carbon metabolism (3)(4)(5). Under stress conditions it might also play a role in minimizing the production of reactive oxygen species from electron transport (5)(6)(7)(8)(9).
Plants that lack IMMUTANS are variegated. Whereas cells in the green sectors of immutans (im) plants have normal-appearing chloroplasts, cells in the white sectors have vacuolated plastids that lack pigments and internal membrane structures. The white im sectors accu-mulate phytoene, a colorless C 40 intermediate in carotenoid biosynthesis (10). This indicates that im is impaired in the activity of phytoene desaturase, which converts phytoene to -carotene. Because of its resemblance to AOX, it was early hypothesized that IM is a plastid quinol oxidase that functions as a redox component in phytoene desaturation. According to this hypothesis, electrons are transferred from phytoene to plastoquinone (via phytoene desaturase) and then from plastoquinone to oxygen (via IM) (1,2). It was further suggested that a lack of IM (as in im) results in overreduction of the plastoquinone pool and that this is responsible for the build-up of phytoene in the membranes. As a consequence, the production of downstream, photoprotective (colored) carotenoids would be impaired, and under high light illumination conditions, photooxidized plastids would be generated. These would give rise to white cells and sectors as the leaf develops. In support of this hypothesis, IM has quinol oxidase activity in vitro and in vivo (11)(12)(13)(14).
IM is expressed in nearly all Arabidopsis tissues and organs (15). Consistent with the idea that IM plays an important role in plastid metabolism, the differentiation of many plastid and tissue types is impaired in im (15), as well as in the ghost variegation mutant of tomato (16). GHOST is the tomato ortholog of IM (12,16). The ubiquitous expression of IM suggests that its function might not be limited to carotenogenesis. In support of this notion, evidence has accumulated that IM is the elusive terminal oxidase of chlororespiration (and hence is frequently designated "PTOX," for plastid terminal oxidase) (11,13). In addition, IM has been implicated as an important component in the arsenal of plastid responses to oxidative stress, likely as a "safety valve" for the dissipation of excess electron flow (17)(18)(19)(20). 3 Our current working hypothesis is that IM is a versatile alternative electron sink in plastid membranes and that it lies at the intersection of many redox pathways.
Sequence comparisons have revealed that AOX and PTOX are nonheme diiron carboxylate proteins (21)(22)(23)(24)(25). By analogy to crystal structure determinations of non-plant members of this protein class, it has been proposed that the diiron centers of AOX and PTOX are coordinated by four carboxylate and two histidine residues on a four helix bundle (Fig. 1, A and 1B) (23)(24)(25). Support for this model has come from EPR spectroscopy of AOX, as well as from mutagenesis of five of the six proposed iron ligands of AOX using prokaryotic model systems to test function (such as a heme-deficient Escherichia coli strain) (3, 26 -30). In addition to the four alpha helices that bear the active site, AOX and PTOX have a fifth, smaller ␣-helix. This helix and part of helix 1 are thought to be embedded in the membrane but not span it, i.e. AOX and PTOX are modeled as interfacial membrane proteins (Fig. 1B) (23)(24)(25).
The proposition that the structure of PTOX is similar to that of AOX is based on a single sequence, IM from A. thaliana (1,2). The first purpose of this paper was to test the universality of the Berthold/ Andersson/Nordlund model of PTOX (24), given the diversity of PTOX sequences that have become available since the model was proposed. These studies revealed that 1) the six iron ligands are perfectly conserved in PTOX from diverse sources; and 2) nearly all PTOX enzymes have a conserved 16 amino acid sequence that is not present in AOX, and which corresponds precisely to Exon 8 of the higher plant gene. Thus, the second purpose of this paper was to examine the functional significance of the putative iron ligands and Exon 8 sequences both in vitro and in planta. The in planta experiments were made possible by the availability of null im alleles of Arabidopsis and by the ability to obtain fully viable transgenic im plants that bear constructs containing mutated IM genes. Finally, we identified an Arabidopsis im allele that lacks Exon 8 and used this mutant as a tool to further assess the importance of the Exon 8 sequence for PTOX structure and function.

MATERIALS AND METHODS
Plant Strains-Strain 3639 is an uncharacterized A. thaliana variegation mutant from the Arabidopis Biological Resource Center (Ohio State University). It is light-sensitive and was maintained under condi-tions previously described for other light-sensitive Arabidopsis variegations (10). The spotty allele of im (10) and 3639 were used in this report; both are in the Columbia background. Wild type Columbia seedlings served as "wild type" controls.
In Vitro Site-directed Mutagenesis-A full-length IM cDNA has previously been isolated in our laboratory (1). A truncated version of this cDNA that lacks the chloroplast transit peptide sequence was generated by PCR amplification, and the resulting sequence was confirmed by DNA sequencing. The truncated cDNA was cloned into the BamHI and NheI sites of pET-11a (Novagen, Madison, WI), and the QuikChange TM Site-Directed Mutagenesis protocol was used to generate mutations (Stratagene, La Jolla, CA). Instructions provided by the manufacturer were followed, except for deletion mutations, where a 48°C annealing temperature was used instead of 55°C. All the mutated sequences were confirmed by sequencing. Following standard notation, the mutant clones are designated by the amino acid that was altered, its location in the A. thaliana IM sequence, and the amino acid to which it was changed (e.g. H299A: His 299 was changed to Ala).
The primers listed in TABLES ONE and TWO were used to produce mutations in the iron-binding sites of IM and in the Exon 8 sequence.

Mutation
Primer sequences

Mutation Primer sequences
DEX8 is a deletion of the entire Exon 8 sequence, while D1-8 and D9 -16 are deletions of the first eight or last eight amino acids of Exon 8.
Site-directed Mutagensis in Planta-For in planta mutagenesis experiments, mutations were generated using the same procedures and primers described above for the in vitro experiments, except that the full-length IM cDNA was used (to assure proper chloroplast targeting) (1). The mutant cDNAs were cloned behind the cauliflower mosaic virus (CaMV) 35S promoter in the binary vector pBI121, and the constructs were transferred into Agrobacterium tumefaciens; the floral dip method (31) was used to transform im plants (spotty allele). Kanamycinresistant seedlings were selected at the T 1 generation on plates containing 1ϫ Murashige-Skoog salts (pH 5.7), 1% sucrose and 50 g ml Ϫ1 kanamycin. PCR and Southern Blotting methods were used to verify that the plants were transformed. Phenotypic analyses were performed on T 2 generation plants.
In Vitro PTOX Activity Assays-PTOX activity assays were conducted in vitro using procedures described by Josse et al. (12,14). In brief, the mutant constructs were transformed into E. coli strain BL21-DE3, and the cells were grown in LB medium supplemented with 100 g/ml ampicillin until they reached an A 600 of 0.6. Protein expression was induced by the addition of 1.6 mM isopropyl 1-thio-␤-D-galactopyranoside (final concentration). After 2 h, the cells were harvested by centrifugation (3000 ϫ g for 10 min at 4°C), and the pellet was resuspended on ice in a lysis buffer containing 20 mM Tris-HCl (pH 8.0), 1 mM phenylmethylsulfonyl fluoride, and 0.3% protease inhibitor mixture (Sigma). The resuspension was sonicated briefly and unbroken cells were removed by centrifugation (3,000 ϫ g for 10 min at 4°C). The supernatant was then centrifuged in a Beckman L8 -70W ultracentrifuge (200,000 ϫ g for 2 h at 4°C), and the membrane pellet was resuspended in a solution containing 0.75 M sucrose, 0.2 M Tris-HCl (pH 7.5). The concentration of proteins in the pellet was measured using the Bio-Rad protocol with bovine serum albumin as a standard.
Detection of RNA and Protein-Procedures for total cell RNA isolation and Northern blotting have been described (32). The nitrocellulose filters were probed with labeled IM sequences, and RNAs were visualized by phosphorimaging analysis.
E. coli proteins were isolated as described above. To isolate partially purified chloroplasts, green leaf tissues were collected from Arabidopsis seedlings (3-4 weeks old) and homogenized in a Waring blender in the cold in a solution containing 0.33 M sorbitol, 10 mM EDTA, and 50 mM HEPES (pH 8.0). The homogenate was filtered through two layers of Miracloth (Calbiochem, La Jolla, CA) and centrifuged at 2,600 ϫ g for 5 min. The pellet was washed with 10 mM MOPS (pH 8.0). Following centrifugation (10,000 ϫ g for 10 min), the pellet was resuspended in 0.33 M sorbitol, 5 mM MgCl 2 , 50 mM HEPES (pH 8.0). Chlorophyll concentrations were measured on the resuspended membranes as outlined previously (32).
Procedures for Western blotting have been described (32). In brief, equal amounts of chlorophyll (5 g) were electrophoresed through 12.5% SDS-polyacrylamide gels, and the proteins were transferred to a nitrocellulose filter. The filter was incubated with a polyclonal antibody generated to the Arabidopsis IM protein (1:3,000 dilution) (17), and the proteins were visualized using the ECL immunodetection procedure (Pierce, Erembodegum, Belgium).
Protein Import Assays-An IM cDNA containing the entire coding sequence and an IM cDNA in which Exon 8 was deleted (DEX8) (see above) were cloned into the BamHI and NheI sites of pET-11a. These two constructs were used as DNA templates to carry out in vitro transcription and translation with the TNT Quick Coupled Transcription/ Translation system (Promega, Madison, WI). A standard reaction mixture of 50 l contained 40 l of TNT Quick Master, 20 Ci of [ 35 S]methionine, and 1 g of template DNA.
The import assays were conducted with intact pea chloroplasts using established methods (32,33,34). In brief, leaf tissues from pea seedlings (10 -15 days-old) were homogenized in a blender (in the cold) in GR buffer: 0.33 M sorbitol, 2 mM EDTA, 1 mM MgCl 2 , 1 mM MnCl 2 , 0.5g/ liter bovine serum albumin, 5 mM ascorbate, and 50 mM HEPES-KOH (pH 8.0). The homogenate was filtered through two layers of Miracloth (Calbiochem) and centrifuged 5 min at 2,600 ϫ g. The resulting crude chloroplast pellet was resuspended in GR buffer, then loaded onto a Percoll gradient. After centrifuging 10 min at 10,500 ϫ g, intact chloroplasts were removed and washed twice in a buffer containing 0.33 M sorbitol and 50 mM HEPES-KOH (pH 8.0).
A standard import reaction of 500 l contained 330 mM sorbitol, 50 mM HEPES-KOH (pH 8.0), 10 mM NaHCO 3 , 8 mM MgCl 2 , 0.1% bovine serum albumin, 1.5 mM dithiothreitol, 1.2 mg antipain (Sigma), 5 mM ATP and 40 l of 35 S-labeled IM transcription/translation products. The import assay was initiated by adding chloroplasts (corresponding to 200 g of chlorophyll). The preparations were maintained at 25°C in the light. After 30 min, most of the label had been incorporated into the chloroplasts. The fate of the labeled IM protein was monitored for up to 3 h.

RESULTS AND DISCUSSION
Mutagenesis of Putative Iron-binding Ligands in Vitro-Berthold and colleagues have proposed that the structure of PTOX is similar to that of AOX ( Fig. 1) (24). This hypothesis is based on several dozen AOX sequences, but a single PTOX sequence, IM from Arabidopsis (1, 2). As a first approach to test the validity of this model, we compared PTOX sequences that have become available since the model was proposed. Fig. 1C shows the phylogenetic relatedness of five higher plant, two green algal and five cyanobacterial sequences, using Arabidopsis AOX1a as an outgroup. The long branch length indicates that AOX is distantly related to PTOX, as reported previously (1). The higher plant PTOX enzymes form a distinct clade that can be separated into monocot (rice and wheat) and dicot (Arabidopsis, pepper, and tomato) proteins. The green algal sequences (Bigelowiella and Chlamydomonas) also cluster together. Sequence analyses of completed genomes and/or molecular hybridization experiments have revealed that PTOX is a single copy gene in Arabidopsis, tomato, and rice (1,12,16,35). Fig. 1A shows that PTOX proteins, regardless of source, resemble Arabidopsis IM in that their active sites are contained within a four helix bundle. They also have a fifth, shorter ␣-helix that likely anchors PTOX in the plastid membrane (24). We also found that all PTOX enzymes have six perfectly conserved amino acids that could potentially function as iron-binding ligands. These residues are glutamate and histidine; in both PTOX and AOX there are two perfectly conserved EXXH sequences and two perfectly conserved glutamates in the active site region (Fig. 1A). Taken together, these data support the Berthold/ Andersson/Nordlund model of PTOX (24).
As a first approach to test the functional significance of the six putative iron ligands of PTOX, we conducted site-directed mutagenesis experiments using an in vitro activity assay developed by Josse et al. (12,14). In this assay, O 2 consumption is measured in membranes isolated from E. coli that have been transformed with various mutant IM sequences. Addition of NADH as an electron donor results in the formation of reduced quinone (by membrane-bound NADH dehyrdrogenase), and electrons are then transferred to molecular oxygen via PTOX or the cytochrome pathway. PTOX activity is inhibited by pyrogallol analogues, such as propyl gallate and octyl gallate, but is insensitive to cyanide (12,14). Thus, O 2 consumption occurs by the cytochrome pathway in the absence of KCN but by PTOX activity in the presence of KCN. KCN and n-propyl gallate (n-PG) together abolish O 2 consumption. PTOX becomes engaged in this system only when the cytochrome pathway is blocked (12,14).
For our experiments, the conserved iron-binding ligands of Arabidopsis PTOX (IM gene product) were changed to conservative or non-conservative amino acids; the Glu residues were changed to ala (nonconservative) or Asp (conservative), and the His residues were changed to Ala (non-conservative), or Asn (conservative with respect to the space occupied, but non-conservative with respect to charge). Glu was also changed to His (or vice versa) to test whether one type of iron ligand can substitute for another. As controls, His or Glu residues that reside near the putative iron-binding sites were mutated to Ala. The constructs were transformed into E. coli, membranes were isolated, and respiratory measurements were performed in the presence of KCN and n-PG. Fig. 2 shows the O 2 consumption traces of four representative experiments to test whether His 178 (in helix 2) is essential for PTOX activity. In contrast to membranes from non-transformed E. coli ( Fig. 2A and Fig. 3A, lane 1), membranes from cells that contain wild type PTOX display cyanide-resistant O 2 consumption ( Fig. 2B and Fig. 3, lane 2). This activity is inhibited by n-PG, suggesting that it arose from PTOX. Fig. 2D (and Fig. 3A, lane 10) reveals that mutation of His 178 to Ala (H178A) nearly abolishes cyanide-resistant O 2 consumption but that mutation of the adjoining His residue to Ala (H177A) has no affect on consumption ( Fig. 2C and Fig. 3A, lane 9). n-PG sensitive O 2 consumption is also blocked by a partially conservative substitution of His 178 to Asn (H178N) (Fig. 3, lane 12), as well as by alteration of His to Glu (H178E) (Fig. 3, lane 11). Taken together, these data indicate that His 178 is required for PTOX activity and/or stability and that His at this site is essential.
3) The iron ligands do not appear to be interconvertible, i.e. O 2 consumption is inhibited when Glu residues are converted to His or vice versa (E136H, E175H, E227H, E296H, and H299E) (lanes 5, 8, 15, 18, and 21). 4) O 2 consumption is blocked by a   (E298A, lane 19). This suggests that the introduced ironbinding site mutations do not affect the conformation of the active site.
In summary, the in vitro mutagenesis studies suggest that the six putative iron-binding sites of PTOX do not tolerate change. In fact, the only changes that do not impact activity are those that involve nearby carboxylate residues (H177A and E298A, lanes 9 and 19).
One possibility to explain the lack of PTOX activity is that the mutant proteins are not expressed in E. coli or that they are expressed but degraded. To examine this question, we performed Western immunoblot analyses on membrane proteins from the various transformants using a polyclonal IM antibody (17). PTOX was detected in the membrane but not soluble protein fractions, indicating that the various mutations do not likely compromise the incorporation of PTOX into E. coli membranes (data not shown). Fig. 3B reveals that the transformed cells accumulate the mutant proteins (lanes 3-24) at levels found in cells that express the wild type enzyme (lane 2); control transformants with the empty vector do not accumulate PTOX (lane 1). These data indicate that the mutant proteins are expressed and stable in E. coli. Yet, it is still possible that they fail to fold properly in the membrane. We cannot rule this out. However, several lines of evidence make this possibility unlikely. First, it has been reported that the three-dimensional structures of diiron carboxylate proteins are quite stable and that the metal ligands contribute little to the conformation of the active site (36,37). The overall conformation of these proteins also appears to be resistant to alterations in charge and mobility of side chains in the vicinity of the diiron center. Our data support this idea inasmuch as PTOX enzymes with mutations at sites near the putative iron-binding ligands have wild type levels of cyanide-resistant O 2 consumption.
Mutagenesis of Putative Iron-binding Ligands in Planta-To examine the impact of each of the mutations described above on PTOX activity in planta, the mutant constructs were introduced into a null im allele (spotty) (1). Full-length cDNAs were used for proper chloroplast targeting, and the cDNAs were cloned behind the CaMV 35S promoter; phenotypes were examined in the T 2 plants. At least eight independent transformants were examined per construct (a minimum of 168 different transformation events for the 21 mutations in Fig. 3A). The data from 10 representative transformants (10 different constructs) are shown in Fig. 3C. These data show that: 1) transformation of im with a wild type IM cDNA ("WT" lane) abolishes the variegation phenotype, generating normal appearing plants. This demonstrates that the IM cDNA is able to complement the im defect. 2) Control transformations of im with the empty binary vector ("pBI121" lane) remain variegated. 3) T 2 plants from the H177A and E298A transformations (i.e. mutant DNAs with changes at nearby carboxylate residues) resemble wild type, suggesting that these two mutations do not impact PTOX activity in planta. 4) T 2 plants from the E136A, E175A, H178A, E227A, E296A, and H299A transformations remain variegated, suggesting that these mutations affect PTOX activity in planta. Similar results were obtained for the 12 other mutations: E136D, E136H, E175D, E175H, H178N, H178E, E227D, E227H, E296D, E296H, H299E, and H299N (data not shown). Fig. 3D shows that the mutants in Fig. 3C have similar levels of IM protein accumulation. This suggests that a failure to complement im is not due to a lack of IM expression and/or protein abundance. Therefore, we hypothesize that conservative and non-conservative mutations in any of the six putative iron-binding sites are deleterious to PTOX activity in planta. The only two mutations that do not seem to affect activity are those in carboxylate residues that reside near the iron-binding ligands.
In summary, we conclude that the results from the in vitro and in planta experiments are in striking agreement with one another and show that the six putative iron ligands of PTOX are required for enzyme activity. In addition, these residues do not tolerate change, even con-  H299N (lane 22). B, total membrane proteins were isolated from transformed E. coli, and equal amounts were electrophoresed through 12.5% SDS-PAGE gels, then transferred to nitrocellulose membranes. The membranes were treated with an antibody to PTOX and visualized by the ECL system. Lanes are as in A. C, the same mutations as in the in vitro mutagenesis experiments were introduced into im plants. Kanamycin-resistant seedlings were scored by PCR and Southern blotting for the presence of the kanamycin gene (NPTII). Over 168 bone fide transformation events were examined (at least eight per construct); the figure shows 10 representative T 2 generation seedlings (10 different constructs). All plants were grown for 3-4 weeks under continuous light (60 -90 mol m Ϫ2 s Ϫ1 ). D, Western immunoblot analyses were conducted using 5 g of chlorophyll per lane from chloroplast membranes of partially purified plastids using an antibody to Arabidopsis PTOX.  DECEMBER 30, 2005 • VOLUME 280 • NUMBER 52 servative ones. This might explain why these residues are perfectly conserved in all AOX and PTOX proteins examined to date.

Sequences Required for PTOX Activity
The present findings are in agreement with previous mutagenesis studies on AOX, in which the functional significance of five of the six iron-binding sites has been examined (the helix 2, 3, and 4 ligands) (26 -30). Mutagenesis of the sixth ligand (Glu 183 on helix 1) has not been reported. Some of the mutagenesis data are equivocal. For example, Berthold et al. (26) found that mutations in four of the Arabidopsis AOX1a iron-binding sites (E222A, H225A, E273A, and H327A) do not restore aerobic respiration to a heme-deficient strain of E. coli. However, the lack of complementation was due to protein instability in the H225A and H327A mutants, rather than to an inactive enzyme, as in the other two cases (monitored by lack of detectable mixed valent EPR signals characteristic of a binuclear iron center). The identification of Glu 222 as important for activity has been confirmed by Albury et al. (30), who reported that an E217A mutant of the Sauromatum guttatum AOX (analogous to Glu 222 of Arabidopsis AOX1a) is inactive in an S. pombe expression system; this mutation does not affect protein stability. Of the six sites, Berthold et al. (26) demonstrated directly that two of them bind iron (Glu 222 and Glu 273 , corresponding to Glu 175 and Glu 227 of PTOX). Although physical evidence that PTOX binds iron is lacking, Josse et al. (14) found that PTOX requires iron for activity.
Structural Domains of PTOX Essential for Activity: Exon 8-Umbach and Siedow (38) hypothesized that there are two general types of AOX, higher plant and fungal, distinguished by their subunit structures and regulators: ␣-keto acid or succinate stimulation of activity corresponds with a dimeric structure, while purine nucleotide stimulation corresponds with a monomeric structure. The N terminus of AOX contains a conserved cysteine (Cys 127 of Arabidopsis AOX1A, see Fig. 1A) that is responsible for homodimer formation via the formation of a disulfide bridge (3)(4)(5)38). Activation of the enzyme occurs by reduction of this linkage and the binding of pyruvate (and other organic acids) to the free Cys residue (39,40). This Cys is embedded within a conserved ϳ40amino acid domain that might mediate protein/protein interactions ("dimerization domain, " Fig. 1A). Whereas some species (e.g. rice and tomato) lack the regulatory Cys and instead have a serine at this site (and are activated by succinate), these enzymes nonetheless form dimers, perhaps via their dimerization domains (38). In contrast to the higher plant enzymes, fungal AOX proteins lack the regulatory Cys, as well as the dimerization domain (38). Yet, they contain a conserved ϳ20 -25 amino acid sequence in their N terminus that is not present in the higher plant enzymes; the function of this domain is obscure. Fig. 1A shows that PTOX proteins lack the AOX-specific regulatory regions, i.e. the regulatory Cys (or ser), the dimerization domain, and the fungus-specific N-terminal domain (Fig. 1A). This confirms earlier observations (1,2,38). However, our comparisons further revealed that PTOX contains a conserved 16-amino acid insertion in (or near) the active site that is not found in AOX from any source. This sequence is present in all the species in Fig. 1C with the exception of Synechococcus (where it is truncated) and P. marinus (where it is totally absent). Alignments show that 7 of the 16 residues are identical in this sequence across species (Fig. 4). Other amino acids are also highly conserved (e.g. Glu 268 is found in all species except Chlamydomonas). Interestingly, the 16-amino acid sequence corresponds precisely to Exon 8 of the higher plant genomic sequence and is located between helix 3 and helix 4 (Fig. 1, A  and B). Data base searches revealed that this sequence is not found in other proteins and that it does not contain or comprise a known protein motif.
To test whether the Exon 8 domain affects PTOX activity, three deletion constructs were generated: DEX8, which lacks the entire Exon 8 sequence; D1-8, in which the first eight amino acids of Exon 8 were deleted, and D9 -16, in which the last eight amino acids of Exon 8 were deleted. Point mutations were also generated to the seven perfectly conserved amino acids in Exon 8 (D267A, E268A, F269A, Q270A, T271A, R278A R279A, and P280A), as well as to the highly conserved E268 (E268A). The mutants were then tested in E. coli for their cyanideresistant, n-PG-sensitive O 2 consumption activity (Figs. 2 and 3A). They were also introduced into im to test their in planta activity (Fig. 3C). Fig. 5A shows that deletion of all or part of Exon 8 nearly abolishes in vitro PTOX activity, whereas activity is not appreciably affected by mutations in the eight conserved amino acids. Protein expression in E. coli is similar for all the mutant proteins (Fig. 5B). As expected, the DEX8, D1-8, and D9 -16 proteins migrate at a lower molecular weight than the normal sized protein.    5C shows that mutations in the eight conserved amino acids do not impact PTOX activity in planta inasmuch as the T 2 plants resemble wild type. In contrast, DEX8, D1-8, and D9 -16 T 2 plants are variegated. This suggests that the deleted sequences are essential for activity. The various transformants in Fig. 5C have wild type levels of IM protein, and the deletion mutant proteins are smaller than normal, again as anticipated (Fig. 5D). Considered together, the in vitro and in planta data are in accord with one another.
In an ongoing survey of variegation mutants of Arabidopsis, we discovered one that turned out to be allelic to im (strain 3639, obtained from the Ohio State Arabidopis Biological Resource Center) and that lacks Exon 8. This allele was generated by ethyl methane sulfonate and Exon 8 is perfectly deleted from this strain (data not shown). 3639 and wild type Columbia are shown in Fig. 6A. Northern blot analyses revealed that IM mRNAs are expressed at normal levels in 3639 (Fig.  6B); however, the protein product is undetectable by Western blot analysis (Fig. 6C). The phenotype of 3639 resembles that of other alleles of im that are predicted to be null (1,10).
In contrast to the normal accumulation of protein that occurs when the DEX8 construct is overexpressed in im (Fig. 5D, above), the lack of IM protein in 3639 (i.e. which is also predicted to contain an Exon 8 deletion) suggests that accumulation of the mutant protein is regulated post-translationally in 3639 but not in the overexpressor. To test this hypothesis, we examined the stability of the mutant protein after import into isolated chloroplasts. The DEX8 and wild type IM constructs were transcribed and translated in vitro using [ 35 S]methionine, and the labeled proteins were then imported into purified pea chloroplasts. Other investigators have shown that import is nearly complete 30 min after mixing the plastids and labeled precursor proteins together (33,34). This is illustrated in Fig.  7A, which shows substantial import of the wild type and mutant proteins after 30-min incubation (time "0"): the upper band is the precursor protein, and the lower band is the imported, mature protein. As in Fig. 5D, the DEX8 proteins are smaller than normal because of the Exon 8 deletion. 30 min after incubation, the plastids were washed, and the fate of the labeled proteins was followed during a 3-h chase. Fig. 7A reveals that both proteins decrease in amount during the chase, and that after 3 h, the levels of the wild type protein decrease about 20% versus 60% for the mutant protein (Fig. 7B). This suggests that the DEX8 protein is less stable than the wild type protein. We conclude that deletion of the Exon 8 domain causes PTOX to be turned over more rapidly than normal in strain 3639. This suggests that in addition to being important for activity, the Exon 8 domain is essential for PTOX stability, e.g. the protein might be turned over if this domain is necessary for assembly with other proteins in the membrane.
Why does the mutant PTOX behave differently in 3639 versus the DEX8 overexpressor? One difference between the two strains is the promoter used to drive expression of the Exon 8 deletion: in 3639, transcription is driven by the relatively weak IM promoter, while in the DEX8 plants, transcription is driven by the high level expression CaMV 35S promoter. This promoter difference is reflected in the abundance of IM mRNAs, which are significantly higher in the DEX8 plants (data not shown). Therefore, we speculate that a post-transcriptional system is limiting in the DEX8 plants and, consequently, that this causes DEX8 proteins to accumulate. One possibility is that the capacity of the proteolytic system(s) of the plastid responsible for degrading defective or excess PTOX is limiting and not sufficient to handle the large amounts of protein that are produced.