ARTICLE IN PRESS

Journal of Plant Physiology 167 (2010) 561570

Contents lists available at ScienceDirect

Journal of Plant Physiology

journal homepage: www.elsevier.de/jplph

Stress-induced co-expression of two alternative oxidase (VuAox1 and 2b) genes in Vigna unguiculata
Jose Helio Costa a, Erika Freitas Mota a, Mariana Virginia Cambursano b, Martin Alexander Lauxmann b, Luciana Maia Nogueira de Oliveira c, Maria da Guia Silva Lima a, Elena Graciela Orellano b, Dirce Fernandes de Melo a,n
Department of Biochemistry and Molecular Biology, Federal University of Ceara, 60455-760 Fortaleza, Ceara, Brazil Molecular Biology Division, Instituto de Biologa Molecular y Celular de Rosario, Consejo Nacional de Investigaciones Cientcas y Tecnicas, Facultad de Ciencias Bioqumicas y Farmaceuticas, Universidad Nacional de Rosario, Suipacha 531, (S2002LRK) Rosario, Argentina c Academic Unit of Garanhuns, Federal Rural University of Pernambuco, Garanhuns, Pernambuco, Brazil
b a

a r t i c l e in fo
Article history: Received 10 July 2009 Received in revised form 1 November 2009 Accepted 2 November 2009 Keywords: Alternative oxidase Evolution Gene co-expression Intron length Stress

abstract
Cowpea (Vigna unguiculata) alternative oxidase is encoded by a small multigene family (Aox1, 2a and 2b) that is orthologous to the soybean Aox family. Like most of the identied Aox genes in plants, VuAox1 and VuAox2 consist of 4 exons interrupted by 3 introns. Alignment of the orthologous Aox genes revealed high identity of exons and intron variability, which is more prevalent in Aox1. In order to determine Aox gene expression in V. unguiculata, a steady-state analysis of transcripts involved in seed development (owers, pods and dry seeds) and germination (soaked seeds) was performed and systemic co-expression of VuAox1 and VuAox2b was observed during germination. The analysis of Aox transcripts in leaves from seedlings under different stress conditions (cold, PEG, salicylate and H2O2) revealed stress-induced co-expression of both VuAox genes. Transcripts of VuAox2a and 2b were detected in all control seedlings, which was not the case for VuAox1 mRNA. Estimation of the primary transcript lengths of V. unguiculata and soybean Aox genes showed an intron length reduction for VuAox1 and 2b, suggesting that the two genes have converged in transcribed sequence length. Indeed, a bioinformatics analysis of VuAox1 and 2b promoters revealed a conserved region related to a ciselement that is responsive to oxidative stress. Taken together, the data provide evidence for coexpression of Aox1 and Aox2b in response to stress and also during the early phase of seed germination. The dual nature of VuAox2b expression (constitutive and induced) suggests that the constitutive Aox2b gene of V. unguiculata has acquired inducible regulatory elements. & 2009 Elsevier GmbH. All rights reserved.

Introduction The alternative respiratory pathway branches from the mitochondrial electron transport chain at the ubiquinone pool and passes electrons to a terminal (alternative) oxidase (Aox). The alternative pathway is not coupled to ATP synthesis and is not inhibited by cyanide. Alternative oxidases have been found in all plant mitochondria analyzed. Several factors have been identied that are involved in the regulation of its activity such as the amount of Aox protein, the redox-active sulfhydryls of Aox dimers, alpha-keto acids and cellular pH (Lima-Junior et al., 2000; Millenaar and Lambers, 2003). The always renewed interest

Abbreviations: Aox, alternative oxidase; cDNA, DNA complementary to RNA; ROS, reactive oxygen species n Corresponding author. Tel.: + 55 85 3366 9825; fax: + 55 85 3366 9829. E-mail address: dirmelo@ufc.br (D. Fernandes de Melo). 0176-1617/$ - see front matter & 2009 Elsevier GmbH. All rights reserved. doi:10.1016/j.jplph.2009.11.001

in Aox studies is based on its apparent role in protecting plant mitochondria from reactive oxygen species (ROS) due to the ability of this enzyme to catalyze non-coupled respiratory electron transport. Therefore, its activity promotes a more oxidized state of the electron transport chain components, which decreases ROS formation (Purvis, 1997; Maxwell et al., 1999; Robson and Vanlerberghe, 2002). More recently, it has been speculated that Aox may play a crucial role in plant cell programming (Arnholdt-Schmitt et al., 2006; Clifton et al., 2006). Aox is encoded in the nucleus and imported into mitochondria via the general import pathway (Tanudji et al., 1999). It has also been shown that Aox is encoded by a small family of three to ve members in several plant species and appears to undergo regulation during development and in different tissues (Finnegan et al., 1997; McCabe et al., 1998; Considine et al., 2001; ThirkettleWatts et al., 2003). These multigene families can be divided in two subfamilies, Aox1 and Aox2, which are formed by variable gene numbers in several plants. The Aox1 family is present in both

ARTICLE IN PRESS
562 J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570

monocot and eudicot plants and higher protein sequence similarity exists between different species than similarity to Aox2 in the same species (Considine et al., 2002). It is well known that Aox is highly responsive to a variety of treatments that induce oxidative stress (Clifton et al., 2006). Several Arabidopsis studies have described Aox1a as the most stress responsive Aox gene (Saisho et al., 1997; Clifton et al., 2005). More recent research has explored the role of Aoxla to combined abiotic/environmental stresses (Giraud et al., 2008). Aox2 is only found in eudicots and only one Aox2 gene is present in tobacco (Whelan et al., 1996), Arabidopsis (Saisho et al., 1997), mango (Considine et al., 2001) and tomato (Holtzapffel et al., 2003); however, two Aox2-types (Aox2a and Aox2b) are present in soybean (Whelan et al., 1996), cowpea (Costa et al., 2004) and carrot (Costa et al., 2009a). Aox2 is typically constitutive or related to development (Considine et al., 2002). Nevertheless, Arabidopsis Aox2 also plays a role in the stress response related to plastid-dependent signaling (Clifton et al., 2005). V. unguiculata roots also had Aox2b regulation under salt and osmotic stresses (Costa et al., 2007). Approaches involving promoters combined with gene expression of Aox genes indicated that the expression pattern between species is not conserved with gene orthology (Thirkettle-Watts et al., 2003). Recently, promoter mutagenesis revealed common cis-elements in non orthologous Aox genes that may be involved in Aox expression regulation by growth and developmental signals (Ho et al., 2007). cis-acting regulatory elements have also been identied in AtAox1a as part of regulatory pathways controlling gene expression in response to stress (Ho et al., 2008). In spite of the great interest related to structure, function and evolution of Aox multigene family in plants, several aspects are not well understood, creating a challenge in this particular eld. In this paper, Vigna unguiculata Aox genes have been characterized addressing gene structure as a tool to compare evolutionary divergence of Aox introns in the soybean. Furthermore, a clear role for V. unguiculata Aox1 was not found in previous papers of our group (Costa et al., 2004, 2007). Thus, we carried out an expression analysis of Aox transcripts in leaves, under several treatments (cold, PEG, salicylate and H2O2), in seed development and germination to provide insights into the regulation of Aox expression. To support experimental data, a bioinformatics analysis of VuAox1 and VuAox2b promoters was performed in search of cis-elements involved in oxidative stress signaling.

Materials and methods Cloning and sequence analysis of Aox genes

in the exon 3 region. To obtain the complete Aox1 gene, 50 and 30 primers were designed from an alignment with soybean Aox1 and Arabidopsis Aox1a, 1b and 1c. Aox1 was amplied by PCR using the following primer pairs: P3 [50 ATGATGATGAGTCGCAGC 30 (sense)] and P4 [50 TTGTCCAATTCCTTGAGGA 30 (antisense)] to amplify the 50 end and P5 [50 TCCTCAAGGAATTGGACAA 30 (sense)] and P6 [50 AGTGATAACCAATWTGGAGC 30 (antisense)] to amplify the 30 end. The Aox1 fragments were also cloned into the pGEM T-Easy vector and sequenced. Aox2a and 2b of V. unguiculata were amplied by PCR using specic primers spanning the DNA complementary to RNA (cDNA) sequences (AJ319899 and AJ421015), respectively. After PCR amplication, the genes were cloned into the pGEM T-Easy vector and sequenced (University of Maine Service). Sequence data were analyzed using Clustal X algorithm software (Thompson et al., 1997). The Aox1, 2a and 2b gene sequences from V. unguiculata are available in the GenBank database under the access numbers DQ100440, EF187463 and DQ100439, respectively. The phylogenetic tree was constructed by the neighbor-joining method (Saitou and Nei, 1987) of the CLUSTAL X program (Thompson et al., 1997). Accession numbers of Aox sequences used in phylogenetic analyses published in the GenBank or TIGR databases are as follows: common dracunculus (Dracunculus vulgaris) Aox1 (AB189673); cowpea (Vigna unguiculata) Aox1 (DQ100440), Aox2a (AJ319899) and Aox2b (AJ421015); green algae (Chlamydomonas reinhardtii) Aox0A (AF047832) and Aox0B (AF285187); maize (Zea mays) Aox1a (AY059647), Aox1b (AY059648) and Aox1c (AY059646); mango (Mangifera indica) Aox2 (X79329); Novosphingobium aromaticivorans Aox0 (CP000248); Neurospora crassa Aox0 (L46869); pacic oyster (Crassostrea gigas) Aox0 (BQ426710); poplar (Populus tremula) Aox1a (AJ251511) and Aox1b (AJ271889); rice (Oryza sativa) Aox1a (AB004864), Aox1b (AB004865) and Aox1c (AB074005); root-knot nematode (Meloidogyne hapla) Aox0 (BM901810); sea vase (Ciona intestinalis) Aox0 (TC17302); soybean (Glycine max) Aox1 (X68702), Aox2a (U87906) and Aox2b (U87907); thale cress (Arabidopsis thaliana) Aox1a (NM_113135), Aox1b (NM_113134), Aox1c (NM_113678) and Aox2 (NM_125817); Trypanosoma brucei Aox0 (AB070617); tobacco (Nicotiana tabacum) Aox1a (S71335); tomato (Lycopersicon esculentum) Aox1a (AY034148) and Aox1b (AY034149); voodoo lily (Sauromatum guttatum) Aox1 (M60330); wine grape (Vitis vinifera) Aox2 (TC46683); wheat (Triticum aestivum) Aox1a (AB078882) and Aox1c (AB078883); yeast (Candida albicans) Aox0A (AF031229) and Aox0B (AF116872).

Plant material and growth conditions Genomic DNA was extracted from dark-grown seedlings of 7-day-old cowpea [Vigna unguiculata (L.) Walp] hypocotyls via nuclei isolation (Henfrey and Slater, 1988). In order to remove polysaccharides and polyphenols, an additional purication step was conducted using a cetyltrimethylammonium bromide (CTAB) protocol (Murray and Thompson, 1980). The Aox1 gene of V. unguiculata was amplied by PCR in two steps. First, an Aox gene fragment was obtained by PCR using degenerate primers, P1 (50 CTGTAGCAGCAGTVCCTGGVATGGT 30 ) and P2 (50 GGTTTACATCYCGYTGYTGWGCCTC 30 ) deduced of exon 3 conserved regions (Saisho et al., 1997). The expected PCR product, a single band of 444 bp, was cloned into the pGEM T-Easy vector (Promega, Madison, WI), and the clones that showed different RFLP patterns were puried and sequenced (University of Maine Service). Second, the identied Aox1 fragment nucleotide sequence was aligned with the soybean and V. unguiculata Aox genes (accession numbers X68702, U87906, U87907, AJ319899, AJ421015) to design specic primers Cowpea [Vigna unguiculata (L.) Walp], var. Vita 5 seeds were obtained from the seed bank of the Departamento de Fitotecnia, Universidade Federal do Ceara, Fortaleza, Ceara, Brasil. Seeds were surface sterilized for 5 min in 0.5% (w/v) CaOCl, exhaustively rinsed with water, and sowed in different conditions. In order to evaluate the steady-state mRNA levels in seed development, plants were cultivated in the environmental eld (Universidade Federal do Ceara) for 4050 days under irrigated control conditions to obtain young or old owers, 3- or 10-day-old pods, and dry seeds (from 20-day-old pods). Flowers were differentiated through the coloration changes of green (young) to yellow (old). The seed germination experiment was carried out with seeds sowed in lter paper imbibed with distilled water for 24 h in the dark. Embryo and cotyledons from dry and soaked seeds (24 h) were also obtained for this study. All tissues (from at least 3 plants or several seeds) were immediately frozen in liquid nitrogen and stored at 80 1C before total RNA was extracted.

ARTICLE IN PRESS
J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570 563

To analyze the effect of several stresses on Aox expression, a hydroponic system was performed. Seeds were sowed in the dark, on lter paper imbibed with distilled water. After 3 days, the seedlings were transferred to hydroponics systems and transported to a greenhouse or growth chambers (cold stress) with a light intensity of 200 mE m 2 s 1 at leaf level for a 12 h photoperiod at 70% relative humidity. The seedlings were grown in Knop medium (1.44 g L 1 Ca(NO3)2, 0.25 g L 1 KNO3, 0.25 g L 1 KH2PO4 and 0.246 g L 1 MgSO4 7H2O) with micronutrients (65.7 mg L 1 iron-ethylene-diamine-tetra-acetic acid (FeEDTA), 2.86 mg L 1 H3BO3, 2.84 mg L 1 MnCl2 4H2O, 0.286 mg L 1 Na2O4Mo 2H2O, 0.22 mg L 1 ZnSO4 7H2O, 0.079 mg L 1 CuSO4 5H2O and 0.0476 mg L 1 CoSO4 7H2O) and the stresses were induced after 6 days of germination by adding 200.67 g L 1 polyethylene glycol (PEG) and 10 mM H2O2 to the nutrient medium or by spraying the leaves with 0.5 mM sodium salicylate. For cold treatment, seedlings were maintained in growth chambers at 25 1C (control) or 4 1C (cold). Leaves from 3 plants of each condition were harvested at 0, 6, 12 and 24 h, rinsed in distilled water at 4 1C, immediately frozen in liquid nitrogen and stored at 80 1C before extraction of total RNA. Extraction of total RNA and semi-quantitative RT-PCR Seeds, owers, pods and leaves were pulverized with liquid nitrogen using a mortar and pestle. Total RNA was then extracted using the RNeasy plant mini kit (Qiagen, Hilden, Germany). Total RNA (1 mg) was heated to 70 1C in a water bath for 10 min and then cooled on ice for 2 min. RT-PCR was performed using the Ready To Go RT-PCR beads kit (Pharmacia) with specic spanning introns primers for each Aox gene. The reactions were individually controlled by amplication of V. unguiculata actin cDNA. The VuAox1 primers were designed from the sequenced gene while the VuAox2a, VuAox2b and actin primers were designed according to Costa et al. (2004). The amplied cDNA fragments were 530, 692, 1031 and 840 bp in length for actin, Aox1, Aox2a and Aox2b, respectively. PCR assays were carried out to establish the optimal annealing temperature of primers at 55, 55, 58 and 57 1C for actin, Aox1, Aox2a and Aox2b, respectively. Preliminary experiments with various PCR cycle numbers indicated that in the RT-PCR conditions of this study, amplications were not in the plateau phase, and therefore allowed for semi-quantitative estimations of transcript levels. The cycle number used for each gene was 25, 27, 28 and 35 for Actin, VuAox2b, VuAox2a and VuAox1, respectively. The primer sequences were: VuAox1 sense: 50 GGTTTAAGCGGTGAAGTTG 30 and antisense: 50 TTGTCCAATTCCTTGAGGA 30 ; VuAox2a sense: 50 GCATTGAGTTGTACGGTTCG 30 and antisense: 50 TGGTAAAGGACTGTACTAAGC 30 ; VuAox2b sense: 50 GGATGTCCACTCTTCCAGAC 30 and antisense: 50 GCTCAATGGTAACCAATAGG 30 ; Actin sense: 50 GCGTGATCTCACTGATGCC 30 and antisense: 50 TCGCAATCCACATCTGTTGG 30 . The RT-PCR products were analyzed by electrophoresis in a 1.5% (w/v) agarose gel, stained with ethidium bromide and photographed with a gel imaging system (itf labortechnikGermany). To compare the level of expression of Aox mRNA against the actin reference gene mRNA data, image analysis was employed. The band densities were analyzed with Scion Image Release beta 3b software (Scion Corporation USA). Database search of VuAox1 and VuAox2b upstream promoter regions The VuAox1 and 2b promoter regions were identied by BLAST (Altschul et al., 1997) search homology with 50 ends of the respective cDNAs against GenBank (NCBI) and CGKB (Cowpea Genomics Knowledge Base) (Chen et al., 2007). Nucleotide

sequences sharing the 325 (GenBank access number: EI910014) and 556 bp (CGKB access number: 962_91_ 14540664_16654_45477_013.ab1) upstream regions of the transcriptional start site (TSS) of VuAox1 and 2b, respectively, were used in bioinformatics search for regulatory cis-elements.

Results Identication of the V. unguiculata Aox1 gene A 1543 bp genomic sequence (accession number DQ100440) was obtained by PCR and revealed a precursor protein of 316 amino acid residues homologous to alternative oxidase (Aox) proteins. A phylogenetic tree based on the amino acid sequences of various Aox proteins showed that this deduced protein was very similar to soybean Aox1 and was very distinct from V. unguiculata Aox2a and 2b (Fig. 1). Therefore, the novel protein encoded by the Aox1 gene was called V. unguiculata Aox1 (VuAox1). An alignment of the deduced amino acid sequences of V. unguiculata and soybean Aox1 reveals an identity of 89% (Fig. 2). These proteins presented the VRSEST conserved motif, which is the predicted site of cleavage of the putative mitochondrial-targeting presequence. The soybean Aox1 motif was previously deduced by Whelan et al. (1995) while the V. unguiculata Aox1 motif was deduced by PSORT software (Nakai and Kanehisa, 1992). The deduced presequence and the putative mature sequence of VuAox1 showed 84% and 90% of identity to soybean Aox1, respectively. In this approach, VuAox1 may be considered orthologous to Aox1 from soybean. Intron/exon structure and primary transcripts of Aox genes Fig. 3 shows alignments of intron/exon structures (A) as well as the estimated primary transcript lengths (B) of Aox genes between Glycine max (Gm) and V. unguiculata (Vu). The VuAox1and VuAox2-type genes consist of 4 exons interrupted by 3 introns. In comparing all the genes, a length similarity was revealed between VuAox2a and GmAox2a. VuAox1 and VuAox2b were smaller than respective orthologous Aox genes in G. max. The bars between exons of Aox2-type genes indicate intron regions with high identities ranging from 70% to 95% (Fig. 3A). The primary transcript lengths of Aox of V. unguiculata and G. max estimated from gene, cDNA or EST data (Fig. 3B) revealed a distinct prole between both species. VuAox1 and VuAox2b were similar lengths, and they were smaller than VuAox2a; however, the G. max Aox primary transcript lengths were variable compared to each other. Steady-state mRNA levels of Aox genes The expression of VuAox1, 2a and 2b was studied by RT-PCR as shown in Fig. 4. The transcripts were evaluated in seed development [young and old owers, 3- and 10-day-old pods (with seeds) and dry seeds] and germination (24 h soaked seeds) (Fig. 4A). The mRNA levels were also analyzed in embryos and cotyledons from dry and soaked seeds (Fig. 4B). The specicity of Aox2-type gene primers was tested according to Costa et al. (2004). The Aox1 primers specicity was certied after sequencing of RT-PCR fragment from soaked seeds (access number DQ100441). These analyses demonstrated that VuAox1 and VuAox2b were expressed in all tissues. In dry seeds, the transcripts were expressed at very low levels, and in soaked seeds, VuAox1 and 2b were expressed at high levels. Concerning VuAox2a, it is regulated in an inconsistent manner in young to old

ARTICLE IN PRESS
564 J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570

Fig. 1. Phylogenetic tree of alternative oxidase from several plants, fungi (N. crassa and C. albicans), protists (T. brucei), green algae (C. reinhardtii), animalia (C. gigas, C. intestinalis and M. hapla) and eubacteria (N. aromaticivorans) showing the orthologous pairs soybean and V. unguiculata Aox1 and Aox2 genes. Classication of Aox proteins is according to Considine et al. (2002). Branches are drawn in proportion to genetic distance. The tree was constructed according to sequence data indicated in the Material and Methods section.

ARTICLE IN PRESS
J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570 565

Fig. 2. Alignment of the deduced amino acid sequences of orthologous Aox1 proteins of Vigna unguiculata (Vu) and Soybean (Gm). Identical amino acid residues are shown on a black background. The predicted site of cleavage of the putative mitochondrial-targeting presequence is indicated by one lled, underlined triangle and grayish background. Conserved Cys residues are shown on a gray background. Helical regions that are assumed to be involved in the formation of a hydroxy-bridged binuclear iron center (Andersson and Nordlund, 1999; Berthold et al., 2000) are shown with lines above the amino acid sequences. E (glutamate) and H (histidine) amino acid residues involved in the iron-binding are indicated by lled circles. Possible membrane-binding domains (Andersson and Nordlund, 1999; Berthold et al., 2000) are shown by twoheaded arrows above the amino acid sequences.

Fig. 4. Transcript steady-state level of the V. unguiculata Aox multigene family. (A) Transcript level of Aox genes in seed development [young (Y) and old (O) owers, 3 (3 d) and 10-day-old (10 d) pods and dry (D) seeds] and seed germination [24 h soaked seeds (Sk)]. (B) Transcript level of Aox genes in embryos and cotyledons of dry and soaked seeds. Actin cDNA was amplied as an RT-PCR control and the gel proles are representative of three independent RT and PCR reactions from one set of total RNA.

tissues (owers and pods) and an invariable transcript prole was observed between dry and soaked seeds (Fig. 4A). In order to explore the Aox transcript levels in dry and soaked seeds, similar analyses were carried out in embryos and cotyledons separately (Fig. 4B). The VuAox2a mRNA levels were constant in dry and soaked tissues; however, the VuAox1 and 2b transcripts were drastically enhanced in soaked tissues.

Effect of different treatments in Aox gene expression

Fig. 3. Comparison of intron/exon structure and primary transcript lengths of soybean and V. unguiculata Aox genes. (A) Scale diagram of the intron/exon structure of the Aox1- and Aox2-type genes of soybean (GmAox) and V. unguiculata (VuAox). Filled boxes represent exons and lines represent introns. Bars over introns represent regions with high identity (7095%). The scale bar corresponds to 500 bp of chromosomal DNA. (B) Estimated primary transcripts lengths of the Aox1- and Aox2-type genes of soybean (G. max) and V. unguiculata (VuAox). 50 and 30 UTRs (untranslated regions) of primary transcript for VuAox2a and VuAox2b were estimated by full-length cDNA (AJ319899; AJ421015) while the UTRs of VuAox1 were estimated by V. unguiculata EST data (FG920773; FG820441).

The expression of the Aox multigene family was also studied in leaves of V. unguiculata at 0, 6, 12 and 24 h under the following treatments: cold (4 1C), osmotic stress (200.67 g L 1 PEG), salicylate (0.5 mM) and H2O2 (10 mM) (Fig. 5). To evaluate the amount of RNA template in each RT-PCR reaction, Actin was amplied in parallel from the leaf samples. The amount of Aox transcripts was normalized through a ratio of integrated densities of Aox and Actin cDNA bands (Figs. 5B and D). VuAox2a and VuAox2b transcripts were detected in all

ARTICLE IN PRESS
566 J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570

Fig. 5. Transcript level of the Aox multigene family in leaves of V. unguiculata under control conditions or the following stresses: cold (4 1C), 200.67 g/L PEG, 0.5 mM salicylic acid and 10 mM H2O2. (A and C): RT-PCR products of Aox and Actin on 1.5% agarose gel stained with ethidium bromide. (B and D) Normalization of the quantity of Aox transcripts through a ratio of integrated densities of the Aox cDNA and Actin cDNA bands. Data (B and D) are the average values 7SD of three independent RT and PCR reactions/gels from one set of total RNA. Statistical analysis (t-test, p o0.05) was applied to each gene separately. Different letters indicate signicant transcript differences.

control conditions, unlike VuAox1. VuAox2a was constitutively expressed in all the leaves, whereas, VuAox1 and 2b were induced by cold, PEG, salicylate and H2O2. The expression of VuAox1 and 2b was visibly induced by cold at 24 h (Figs. 5A and B). The induction of VuAox1 and 2b was time-dependent in response to PEG, salicylate and H2O2. The highest mRNA levels in response to PEG were attained at 24 h for both genes. In response to salicylate, VuAox1 appears to be uniformly induced while in response to H2O2, the transcript was detected by 12 h and disappeared at 24 h. The highest mRNA levels of

VuAox2b were found at 6 h in response to salicylate, and it appears to be induced by H2O2 only at 6 h (Figs. 5C and D). Analyses of VuAox1 and VuAox2b upstream promoter regions In accordance with the results shown in Fig. 5, which depicts VuAox1 and 2b co-expression in response to several treatments, a comparative analysis of VuAox1 and 2b upstream promoter regions was carried out studying the cis-elements involved in the

ARTICLE IN PRESS
J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570 567

Table 1 Comparative analysis of Aox cis-elements involved in response to treatments with H2O2, rotenone or both (Ho et al., 2008) with upstream regions of the transcriptional start site (TSS) of VuAox1 and VuAox2b. Sequences in AtAox1a (Ho et al., 2008) Possible sequences in VuAox1 Possible sequences in VuAox2b

TGAAGC

cis-elements playing a role in response to treatment with H2O2, rotenone, or both 175 to 180 TGAACC (rev comp) 117 to 122 AGTGAT (rev comp).

454 to 459 TGAAGA 97 to 102 CGTGGT or 49 to 54 CCTGAT. (rev comp) 65 to 69 ACCCG (rev comp) 36 to 41 CACAAA No signicant similarity 361 to 367 TTGTAAA (rev comp) 302 to 306 TCTCC 52 to 57 ATCATC 441 to 445 ATGTG 155 to 162 CTCAATCA

CGTGAT

ATCCG

283 to 287 ATCCA 289 to 294 CACATA (rev comp) No signicant similarity 48 to 54. TCTTAAA 24 to 28 TTTCT (rev comp) No signicant Similarity 106 to 110 Identical No signicant similarity

CACACA

E F

CGGCTTT TCGTAAA

TCTCT

H I J

GTCATC ACGTG TTCGATCA

The cis-elements in the AtAox1a promoter are designated by a letter according Ho et al. (2008). Degenerated nucleotides are represented by bold and underlined letters. Numbers with each sequence refer to the position at which the motifs are present within each upstream region.

AtAox1a response to H2O2, rotenone, or both (Ho et al., 2008). The results summarized in Table 1 revealed several possible cis-elements in V. unguiculata Aox promoters. Ten cis-elements (A to J) of AtAox1a were analyzed and the majority presented at least one degeneration in V. unguiculata Aox promoters except the I cis-element in VuAox1 that was identical. Indeed, VuAox1 and 2b upstream promoter regions were run through the PLACE database (http://www.dna.affrc.go.jp/PLACE) using the Signal Scan program or through PlantCARE (http://bioinfor matics.psb.ugent.be/webtools/plantcare/html) using the search for CARE, and a total of 5 common cis-acting elements were found: classical TATA and CAAT boxes, the AAGAA-motif (unknown function), G-boxes (light regulatory elements) and a CORE (Coordinate regulatory element for antioxidant defense). Among these common cis-elements the CORE appeared to be related to stress-induced coexpression of VuAox genes. This CORE, a cis-element responsive to oxidative stress, is found as a 28 bp conserved motif on the promoter of 3 antioxidant defense genes in rice (Tsukamoto et al., 2005) and reveals high identity with regions (35 bp) in the promoters of both V. unguiculata genes. The VuAox1 and VuAox2b 35 bp regions aligned with the rice CORE showed identities of 64% and 85%, respectively, while VuAox1 and VuAox2b 35 bp regions presented 68% identity when compared with each other (Fig. 6).

Discussion Our results revealed that V. unguiculata Aox is encoded by a small gene family that has at least three genes: Aox1, 2a and 2b. These three genes present a similar prole to the Aox1 and Aox2

genes of the well-studied soybean. The phylogenetic tree (Fig. 1) shows that a clear division of homology exists between Aox1-type (monocots and dicots) and Aox2-type (dicots) proteins. The high degree of identity between soybean and V. unguiculata Aox1 proteins indicate that VuAox1, identied here, is orthologous to the soybean Aox1 as it was the case of both V. unguiculata and soybean Aox2-type genes described by Costa et al. (2004). The alignment between the orthologous Aox1 proteins of soybean and V. unguiculata (Fig. 2) conrms the high degree of identity indicated in the phylogenetic tree (Fig. 1). The amino acid sequences of these proteins are conserved, including the mitochondrial-targeting signal, which is present in orthologous Aox proteins of proximal plant species (Costa et al., 2004). In the process of identifying Aox1 in V. unguiculata, several clones were obtained from PCR products and three distinct RFLP patterns were found corresponding to Aox2-type genes and one Aox1 gene. Indeed, a BLAST search against a cowpea gene-rich space database (Chen et al., 2007) with more than 95% open reading frames revealed several Aox sequence fragments corresponding only to allelic variations of the 3 genes in the diploid genome of the cowpea (data not shown). The structure of VuAox1- and VuAox2-type genes consists of 4 exons interrupted by 3 introns like the majority of plant Aox genes (Considine et al., 2002). Conserved intron regions in V. unguiculata and soybean orthologous Aox2-type genes in contrast to Aox1 intron variability (Fig. 3A) could be representative of differences in Aox chromosomal allocation between both species. In Arabidopsis, Aox1 and Aox2 genes are located on different chromosomes (Clifton et al., 2006). The chromosomal position of Aox genes in V. unguiculata is unknown. A BLASTn search in the preliminary

ARTICLE IN PRESS
568 J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570

Fig. 6. Alignment of the Rice CORE (Coordinate regulatory element for antioxidant defense) with possible homologous cis-elements found in VuAox1 and VuAox2b promoters. In black, conserved nucleotides on the promoter regions of three antioxidant genes in rice: SodCc1 (cytosolic CuZn-SOD1); trxh (cytosolic thioredoxin) and grx (glutaredoxin) according to Tsukamoto et al. (2005). In gray, conserved nucleotides on the VuAox1 and VuAox2b promoters presenting divergence with conserved nucleotides of Rice CORE.

assembly and annotation of the soybean genome, available by the US Department of Energy Joint Genome Institute (DOE JGI, 2008), indicates that Aox1- and Aox2-type genes are on different chromosomes. Soybean Aox1 is on chromosome 4 and soybean Aox2a and 2b are in a tandem arrangement on chromosome 8. For Medicago truncatula, a close related species to soybean and V. unguiculata, a different scenario of chromosomal Aox organization is found. Aox1 (CU914135) and Aox2b (FP102223) appear in two different fragments of chromosome 5. The determination of the chromosomal sequence of Vigna unguiculata will be of great aid to discover if the co-expressed VuAox1 and VuAox2b are proximal. On average, genes in close proximity in the genome show co-expression, even if they are not immediate neighbors, perhaps this phenomenon is due to transcriptional control similarity (Hershberg et al., 2005) and/or due to chromatin remodeling (Batada et al., 2007). The alignment of the intron/exon structure of V. unguiculata and soybean Aox genes (Fig. 3A) revealed intron positioning conservation and intron length reduction (except VuAox2a). These ndings may be reected in reduced primary transcript lengths of V. unguiculata Aox genes (Fig. 3B) compared to the transcript lengths soybean Aox genes. Considering that transcription is a slow and expensive process: in eukaryotes, approximately 20 nucleotides can be transcribed per second at the expense of at least two ATP molecules per nucleotide (Castillo-Davis et al., 2002), several different models have been proposed to explain small introns within genes such as economy selection and time-economy selection (Chen et al., 2005). In this study, we found that VuAox1 and VuAox2b, both genes that are stress-induced, have small introns (Fig. 3A) in agreement with economy and time-economy selections. In fact, both VuAox genes have been grouped among the smallest primary transcripts in comparing several dicots and monocots species (Costa et al., 2009b). Interestingly, the small introns are accompanied by similar primary transcript lengths (Fig. 3B) supporting the idea that the two genes have converged in transcribed sequence length establishing a time synchronism of mRNA synthesis for co-expression. In previous work, a clear role for Aox1 in V. unguiculata was not achieved (Costa et al., 2004, 2007). Thus, the steady-state transcripts of Aox genes were investigated in seed development (young and old owers, 3- and 10-day-old pods and dry seeds) and seed germination (24 h soaked seeds) (Fig. 4). It is well known that Aox2a expression is related to green (photosynthetic) tissues (Finnegan et al., 1997; Considine et al., 2002; Costa et al., 2004). VuAox2a is cyclically regulated from young (green) to old tissues (not green) in accordance with the fact that Aox2a is expressed in green tissues. Alternatively, for the rst time, high levels of VuAox2a mRNA in dry and soaked seeds (24 h) were shown. The transcript level remained constant, suggesting that VuAox2a mRNA had already accumulated in dry seeds and was preserved at the early stages of germination (24 h). Unlike VuAox2a, VuAox1 and 2b were co-expressed during seed development and germination. Furthermore, their transcript levels were in dry seeds and were drastically enhanced in soaked seeds (24 h). Regardless of divergences of Aox gene responsiveness with orthologous genes

(Thirkettle-Watts et al., 2003), these data are in accordance with the data found in early stages of Arabidopsis germination. For example, only AtAox2 mRNA accumulated in dry seeds with similar levels at 24 h of imbibition while AtAox1a, a typical Aox gene responsive to stress (Saisho et al., 1997; Clifton et al., 2005, 2006; Giraud et al., 2008; Ho et al., 2008), was induced after imbibition (Saisho et al., 2001). Therefore, considering that increased cellular levels of ROS occur during seed germination (Job et al., 2005), the transcript level enhancement of VuAox1 as well as VuAox2b in soaked seeds may be related to a cooperative ROS-induced response. Indeed, Aox gene expression in V. unguiculata seeds was similar in both embryos and cotyledons, suggesting a systemic response occurred (Fig. 4B). In order to elucidate stress-induced co-expression of Aox genes, transcript analysis in V. unguiculata leaves was conducted under cold (4 1C), PEG, salicylate (salicylic acid) and H2O2 treatments (Fig. 5). VuAox1 and VuAox2b clustered in response to stress signaling. The patterns of transcript abundance over time revealed a further prevalent response of VuAox2b, which was also expressed in control conditions (Fig. 5). Despite the prevalent VuAox1 response to cold stress, VuAox1 mRNA appeared to always be at a lower level than VuAox2b mRNA because more PCR cycles (35) were needed for amplication. Here, it was found that VuAox2b is both constitutive and stress inducible, and a similar situation occurs with AtAox1a. In Arabidopsis, Aox1a is very stress inducible, and it is also by far, the highest expressed member of the Aox gene family under normal growth conditions (Clifton et al., 2006). Recently, Campos et al. (2009) observed a similar coexpression pattern of Daucus carota Aox1a and Aox2a during growth and development. In a previous study, PEG treatments failed to induce VuAox1 and induced only VuAox2b in the roots of Vita 5 cultivar (Costa et al., 2007). However, these results were obtained under different conditions than the treatments reported here (Vita 5 leaves); therefore, it impossible to suggest that VuAox1 and 2b coexpression is tissue-specic. Furthermore, considering the differential Aox expression between cultivars (Vita 3 and Vita 5) (Costa et al., 2007), it will be important to examine if co-expression is a general feature of these V. unguiculata genes. Taking into account our results compared to the soybean, a crucial difference exists because the transcript patterns of the GmAox genes revealed only GmAox1 to be stress responsive (Millar et al., 1997; Djajanegara et al., 2002; Thirkettle-Watts et al., 2003). In fact, differences in Aox regulation (protein and activity levels) in response to cold treatment were also previously found in the soybean and the mung bean (Vigna radiata), a species from  the Vigna genus (Gonzalez-Meler et al., 1999). In spite of the paradigm statement that Aox1 is induced by a range of stress stimuli and Aox2 is constitutively expressed, admitted as playing a housekeeping role in respiratory metabolism (Considine et al., 2002), Clifton et al. (2005) indicated a role for Arabidopsis Aox2 in stress responses, reinforcing our present results and the results also obtained with V. unguiculata roots (Costa et al., 2007). Furthermore, several EST data from subtracted libraries of cold aerial tissues (FE898183; FD792528) or infected leaves with rust pathogen (FE700224; FE700225;

ARTICLE IN PRESS
J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570 569

FE700226; FE700227) of the common bean (Phaseolus vulgaris) Aox2b, indicate that in a plant species that is very closely related to V. unguiculata, Aox2b also appears to be stress inducible. The stress-induced co-expression of VuAox1 and VuAox2b also suggests regulation occurs at the post-translational level. The presence of multiple Aox subunits in the same tissue/conditions raises the possibility that subunit heterodimerism occurs (Finnegan et al., 1997). Both VuAox1 and 2b isoenzymes may be needed for Aox activity adjustment in stress conditions. All VuAox proteins have both conserved cysteine residues, CysI and CysII, which are assumed to be involved in dimer formation and in the differential regulation of plant Aox proteins (Crichton et al., 2005; Umbach et al., 2006). Recently, conserved amino acids residues, on the vicinity of both sides of the rst cysteine residue (CysI), clearly distinguish Aox1 from Aox2 and have been speculated to present structural signicance to regulatory function (Costa et al., 2009a). In order to gain insight into V. unguiculata Aox regulatory cis-elements, a comparative analysis with established Arabidopsis Aox1a cis-elements involved in stress (Ho et al., 2008) was performed using 325 and 556 bp upstream regions of the transcriptional start site (TSS) of VuAox1 and VuAox2b, respectively. Despite the partial sequences of V. unguiculata Aox promoter regions, the mutagenesis analyses with Arabidopsis Aox1a performed by Ho et al. (2008) showed that the majority of cis regulatory elements for Aox genes are located until 300 bp upstream of the TSS. It was revealed that both V. unguiculata Aox genes have degenerate sequences (except one in VuAox1), suggesting that the stress-induced co-expression of VuAox1 and 2b may be regulated by different pathways than Arabidopsis Aox1a. In this context, the comparison of upstream promoter regions of VuAox1 and 2b revealed a common region presenting high identity with a 28 bp conserved motif on the promoter of 3 antioxidant defense co-expressed genes in rice, designated as CORE (Coordinate regulatory element for antioxidant defense) (Tsukamoto et al., 2005). The presence of a cis-element possibly responsive to oxidative stress in VuAox1 and 2b corroborate the data demonstrating stress-induced co-expression (Fig. 6). In conclusion, our results support, for the rst time, that Aox1 and Aox2b are co-expressed in response to stress, which reveals coordination in gene length as well as the possibility that a cis-element is involved in this regulation. Mutagenesis studies will be of great aid in establishing the actual role of this common cis-element. Indeed, the constitutive and induced nature of VuAox2b expression suggests that initially it was a constitutive gene that has acquired inducible regulatory elements.

Acknowledgements We are grateful to Ivan de Godoy Maia from the Genetic ~ Department IB UNESP, Botucatu, Sao Paulo, Brazil, for discussions and comments on the manuscript. We also want to express ~ our gratitude to Joao Henrique Frota Cavalcanti from the Federal University of Ceara for his helpful contribution on experimental essays. This research was supported by CAPES/SeCyT, CNPq and FUNCAP. References
Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W, et al. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 1997;25:3389402. Andersson ME, Nordlund P. A revised model of the active site of the alternative oxidase. FEBS Lett 1999;449:1722. Arnholdt-Schmitt B, Costa JH, Fernandes de Melo D. AOX a functional marker for efcient cell reprogramming under stress?. Trends Plant Sci 2006;11:2817.

Batada NN, Urrutia AO, Hurst LD. Chromatin remodelling is a major source of coexpression of linked genes in yeast. Trends Genet 2007;23:4804. Berthold DA, Andersson ME, Nordlund P. New insight into the structure and function of the alternative oxidase. Biochim Biophys Acta 2000;1460: 24154. Campos MD, Cardoso HG, Linke B, Costa JH, Fernandes de Melo D, Justo L, et al. Differential expression and co-regulation of carrot Aox genes (Daucus carota). Physiol Plant 2009; doi:10.1111/j.1399-3054.2009.01282.x. Castillo-Davis CI, Mekhedov SL, Hartl DL, Koonin EV, Kondrashov FA. Selection for short introns in highly expressed genes. Nat Genet 2002;31:4158. Chen J, Sun M, Hurst LD, Carmichael GG, Rowley JD. Human antisense genes have unusually short introns: evidence for selection for rapid transcription. Trends Genet 2005;21:2037. Chen X, Laudeman WT, Rushton PJ, Spraggins TA, Timko MP. CGKB: an annotation knowledge base for cowpea (Vigna unguiculata L.) methylation ltered genomic genespace sequences. BMC Bioinformatics 2007;8:129. Clifton R, Lister R, Parker KL, Sappl PG, Elhafez D, Millar AH, et al. Stress-induced co-expression of alternative respiratory chain components in Arabidopsis thaliana. Plant Mol Biol 2005;58:193212. Clifton R, Millar AH, Whelan J. Alternative oxidases in Arabidopsis: a comparative analysis of differential expression in the gene family provides new insights into function of non-phosphosrylating bypasses. Biochim Biophys Acta 2006;1757:73041. Considine MJ, Daley DO, Whelan J. The expression of alternative oxidase and uncoupling protein during fruit ripening in mango. Plant Physiol 2001;126: 16191629. Considine MJ, Holtzapffel RC, Day DA, Whelan J, Millar AH. Molecular distinction between alternative oxidase from monocots and dicots. Plant Physiol 2002;129:94953. Costa JH, Hasenfratz-Sauder MP, Pham-Thi AT, Silva Lima MG, Dizengremel P, Jolivet Y, et al. Identication in Vigna unguiculata (L.) Walp. of two cDNAs encoding mitochondrial alternative oxidase orthologous to soybean alternative oxidase genes 2a and 2b. Plant Sci 2004;167:2339. Costa JH, Jolivet Y, Hasenfratz-Sauder MP, Orellano EG, Silva Lima MG, Dizengremel P, et al. Alternative oxidase regulation in roots of Vigna unguiculata cultivars differing in drought/salt tolerance. J Plant Physiol 2007;164:71827. Costa JH, Cardoso HC, Campos MD, Zavattieri A, Frederico AM, Fernandes de Melo D, et al. Daucus carota L. an old model for cell reprogramming gains new importance through a novel expansion pattern of alternative oxidase (Aox) genes. Plant Physiol Biochem 2009a;47:7539. Costa JH, Fernandes de Melo D, Gouveia Z, Cardoso HG, Peixe A, Arnholdt-Schmitt B. The alternative oxidase family of Vitis vinifera reveals an attractive model to study the importance of genomic design. Physiol Plant 2009b; doi:10.1111/ j.1399-3054.2009.01267.x. Crichton PG, Affourtit C, Albury MS, Carre J, Moore AL. Constitutive activity of Sauromatum guttatum alternative oxidase in Schizosaccharomyces pombe implicates residues in addition to conserved cysteines in a-keto acid activation. FEBS Lett 2005;579:3316. Djajanegara I, Finnegan PM, Mathieu C, McCabe T, Whelan J, Day DA. Regulation of alternative oxidase gene expression in soybean. Plant Mol Biol 2002 73542. Finnegan PM, Whelan J, Millar AH, Zhang Q, Smith MK, Wiskich JT, et al. Differential expression of the multigene family encoding the soybean mitochondrial alternative oxidase. Plant Physiol 1997;114:45566. Giraud E, Ho LHM, Clifton R, Carroll A, Estavillo G, Tan Y-F, et al. The absence of alternative oxidase 1a in arabidopsis results in acute sensitivity to combined light and drought stress. Plant Physiol 2008;147:595610.  Gonzalez-Meler MA, Ribas-Carbo M, Giles L, Siedow JN. The effect of growth and measurement temperature on the activity of the alternative respiratory pathway. Plant Physiol 1999;120:76572. Henfrey RD, Slater RJ. Isolation of plant nuclei. In: Walker JM, editor. Methods in molecular biology. Totowa, New Jersey: Humana Press; 1988. p. 44752. Hershberg R, Yeger-Lotem E, Margalit H. Chromosomal organization is shaped by the transcriptional regulatory network. Trends Genet 2005;21:13842. Ho LHM, Giraud E, Lister R, Thirkettle-Watts D, Low J, Clifton R, et al. Characterization of the regulatory and expression context of an alternative oxidase gene provides insights into cyanide-insensitive respiration during growth and development. Plant Physiol 2007;143:151933. Ho LHM, Giraud E, Uggalla V, Lister R, Clifton R, Glen A, et al. Identication of regulatory pathways controlling gene expression of stress-responsive mitochondrial proteins in arabidopsis. Plant Physiol 2008;147:185873. Holtzapffel RC, Casteli J, Finnegan PM, Millar AH, Whelan J, Day DA. A tomato alternative oxidase protein with altered regulatory properties. Biochim Biophys Acta 2003;1606:15362. Job C, Rajjou L, Lovigny Y, Belghazi M, Job D. Patterns of protein oxidation in arabidopsis seeds and during germination. Plant Physiol 2005;138:790802. Lima-Junior A, Fernandes de Melo D, Costa JH, Orellano EG, Jolivet Y, Jarmuszkiewicz W, et al. Effect of pH on CN-resistant respiratory activity and regulation on Vigna unguiculata mitochondria. Plant Physiol Biochem 2000;38:76571. Maxwell DP, Wang Y, McIntosh L. The alternative oxidase lowers mitochondrial reactive oxygen production in plant cells. Proc Natl Acad Sci USA 1999;96: 82718276.

ARTICLE IN PRESS
570 J.H. Costa et al. / Journal of Plant Physiology 167 (2010) 561570

McCabe PC, Finnegan PM, Millar H, Day DA, Whelan J. Differential expression of alternative oxidase genes in soybean cotyledons during postgerminative development. Plant Physiol 1998;118:67582. Millar AH, Finnegan PM, Whelan J, Drevon JJ, Day DA. Expression and kinetics of the mitochondrial alternative oxidase in nitrogen-xing nodules of soybean roots. Plant Cell Environ 1997;20:127382. Millenaar FF, Lambers H. The alternative oxidase: in vivo regulation and function. Plant Biol 2003;5:215. Murray MG, Thompson WF. Rapid isolation of high molecular weight plant DNA. Nucleic Acids Res 1980;8:43215. Nakai K, Kanehisa M. A knowledge base for predicting protein localization sites in eucaryotic cells. Genomics 1992;14:897911. Purvis AC. Role of the alternative oxidase in limiting superoxide production by plant mitochondria. Physiol Plant 1997;100:16570. Robson CA, Vanlerberghe GC. Transgenic plant cells lacking mitochondrial alternative oxidase have increased susceptibility to mitochondria-dependent and independent pathways of programmed cell death. Plant Physiol 2002;129:190820. Saisho D, Nambara E, Naito S, Tsutsumi N, Hirai A, Nakazono M. Characterization of the gene family for alternative oxidase from Arabidopsis thaliana. Plant Mol Biol 1997;35:58596. Saisho D, Nakazono M, Lee KH, Tsutsumi N, Akita S, Hirai A. The gene for alternative oxidase-2 (Aox2) from Arabidopsis thaliana consists of ve exons

unlike other AOX genes and is transcribed at an early stage during germination. Genes Genet Syst 2001;76:8997. Saitou N, Nei M. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 1987;4:40625. Tanudji M, Sjoling S, Glaser E, Whelan J. Signals required for the import and processing of the alternative oxidase into mitochondria. J Biol Chem 1999;274: 12861293. Thirkettle-Watts D, McCabe TC, Clifton R, Moore C, Finnegan PM, Day DA, et al. Analysis of the alternative oxidase promoters from soybean. Plant Physiol 2003;133:115869. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG. The Clustal X. windows interface: exible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res 1997;25:487682. Tsukamoto S, Morita S, Hirano E, Yokoi H, Masumura T, Tanaka K. A novel ciselement that is responsive to oxidative stress regulates three antioxidant defense genes in rice. Plant Physiol 2005;137:31727. Umbach AL, Ng VS, Siedow JN. Regulation of plant alternative oxidase activity: a tale of two cysteines. Biochim Biophys Acta 2006;1757:13542. Whelan J, Hugosson M, Glaser E, Day DA. Studies on the import and processing of the alternative oxidase precursor by isolated soybean mitochondria. Plant Mol Biol 1995;27:76978. Whelan J, Millar AH, Day DA. The alternative oxidase is encoded in a multigene family in soybean. Planta 1996;198:197201.