An ATP synthase harboring an atypical γ–subunit is involved in ATP synthesis in tomato fruit chromoplasts

Authors

  • Irini Pateraki,

    1. Departament de Bioquímica i Biologia Molecular, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    2. Centre de Recerca en Agrigenòmica (CRAG), Consorci CSIC-IRTA-UAB-UB, Campus Universitat Autònoma de Barcelona, Barcelona, Spain
    Current affiliation:
    1. Department of Plant Biology and Biotechnology, Faculty of Life Science, University of Copenhagen, Thorvaldsensvej 40, Frederiksberg C, Copenhagen, Denmark
    Search for more papers by this author
  • Marta Renato,

    1. Centre de Recerca en Agrigenòmica (CRAG), Consorci CSIC-IRTA-UAB-UB, Campus Universitat Autònoma de Barcelona, Barcelona, Spain
    2. Departament de Biologia Vegetal, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    Search for more papers by this author
  • Joaquín Azcón-Bieto,

    1. Departament de Biologia Vegetal, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    Search for more papers by this author
  • Albert Boronat

    Corresponding author
    1. Centre de Recerca en Agrigenòmica (CRAG), Consorci CSIC-IRTA-UAB-UB, Campus Universitat Autònoma de Barcelona, Barcelona, Spain
    • Departament de Bioquímica i Biologia Molecular, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
    Search for more papers by this author

  • The authors declare no conflict of interest.

For correspondence (e-mail aboronat@ub.edu).

Summary

Chromoplasts are non-photosynthetic plastids specialized in the synthesis and accumulation of carotenoids. During fruit ripening, chloroplasts differentiate into photosynthetically inactive chromoplasts in a process characterized by the degradation of the thylakoid membranes, and by the active synthesis and accumulation of carotenoids. This transition renders chromoplasts unable to photochemically synthesize ATP, and therefore these organelles need to obtain the ATP required for anabolic processes through alternative sources. It is widely accepted that the ATP used for biosynthetic processes in non-photosynthetic plastids is imported from the cytosol or is obtained through glycolysis. In this work, however, we show that isolated tomato (Solanum lycopersicum) fruit chromoplasts are able to synthesize ATP de novo through a respiratory pathway using NADPH as an electron donor. We also report the involvement of a plastidial ATP synthase harboring an atypical γ–subunit induced during ripening, which lacks the regulatory dithiol domain present in plant and algae chloroplast γ–subunits. Silencing of this atypical γ–subunit during fruit ripening impairs the capacity of isolated chromoplast to synthesize ATP de novo. We propose that the replacement of the γ–subunit present in tomato leaf and green fruit chloroplasts by the atypical γ–subunit lacking the dithiol domain during fruit ripening reflects evolutionary changes, which allow the operation of chromoplast ATP synthase under the particular physiological conditions found in this organelle.

Introduction

Chromoplasts are plastids specialized in the synthesis and accumulation of carotenoids, which are generally found in fruits and flowers. They are derived from pre-existing plastids such as chloroplasts or amyloplasts (Thomson and Whatley, 1980), and show great morphological diversity (Camara et al., 1995). During ripening, tomato (Solanum lycopersicum) fruit chloroplasts differentiate into photosynthetically inactive chromoplasts (Bathgate et al., 1985) in a process characterized by the breakdown of the photosynthetic apparatus and a massive synthesis and deposition of carotenoids (Cheung et al., 1993). The differentiation of chloroplasts into chromoplasts correlates with a successive loss of autotrophic characteristics and an increasing heterotrophic character of these plastids (Camara et al., 1995). Fruit chloroplasts are able to synthesize ATP and NADPH, which are used for CO2 fixation and anabolic processes (Batz et al., 1995). During chromoplast differentiation chlorophylls are degraded and the plastid becomes unable to photosynthetically produce ATP and NADPH. In addition, there is an increased demand for exogenous precursors needed for anabolic reactions that must be imported from the cytosol (Camara et al., 1995). Similarly, an alternative source for ATP synthesis is required (Camara et al., 1995; Neuhaus and Emes, 2000). It is currently accepted that in non-photosynthetic plastids ATP is imported from the cytosol through the plastidial ATP/ADP translocator, or is obtained from glycolysis using precursors or intermediates also imported by specific membrane transporters (Flügge et al., 2011). We have recently reported, however, that isolated tomato fruit chromoplasts are able to actively synthesize carotenoids and lipids without the exogenous supply of ATP or glycolytic precursors, suggesting that these plastids could synthesize ATP autonomously (Angaman et al., 2012). These results were in agreement with a previous report showing that liposomes containing solubilized daffodil chromoplast membrane proteins were able to synthesize ATP in vitro (Morstadt et al., 2002).

The recent proteomic analysis of tomato fruit chromoplasts has provided new insights concerning the biochemical and metabolic processes operating in these organelles (Barsan et al., 2010, 2012). The identification of several subunits of the plastid ATP synthase, as well as components of complexes potentially participating in electron transport, like the NADPH dehydrogenase complex (Ndh) or the cytochrome b6f complex (Barsan et al., 2010), suggested the possible operation of processes related to ATP synthesis and respiration in tomato fruit chromoplasts. Moreover, the recent proteomic analysis of the chloroplast-to-chromoplast transition process has revealed that the abundance of proteins involved in energy provision (like ATP synthase complex subunits and metabolic precursor translocators) are maintained during chromoplast differentiation, in contrast to proteins related to photosynthesis, the abundance of which strongly decreases during ripening (Barsan et al., 2012). These findings suggested that energy-production components are preserved in tomato fruit chromoplasts (Barsan et al., 2012). This hypothesis is further supported by the observation that the expression of plastidial genes encoding proteins related to these processes is maintained during the chloroplast-to-chromoplast transition in tomato fruit (Kahlau and Bock, 2008).

The ATP synthase enzyme is responsible for the bulk production of cellular ATP, and its structure and function is highly conserved through all kingdoms of life. In plants there are two different forms of this enzyme complex located in the mitochondria and the chloroplasts. The chloroplast ATP synthase consists of two components, the membrane integral CFO and the hydrophilic CF1, which is associated with the thylakoid membrane through its interaction with CFO (McCarty et al., 2000). Whereas CFO is responsible for proton translocation, and contains four subunits (I1, II1, III14 and IV1), CF1 harbors the catalytic core, and is composed of five subunits (α3, β3, γ1, δ1 and ε1). A key element of the ATP synthase complex is the γ–subunit, which forms the central shaft that connects the CFO with the CF1 catalytic core, and is considered to have a regulatory role in the assembly and function of the whole complex (Bosco et al., 2004). In plants and green algae, the chloroplast γ–subunit has an extra domain, not found in the mitochondrial and bacterial counterparts, which contains two cysteine residues able to form a disulphide bond (Miki et al., 1988). Reduction of this disulphide bond, present in the so-called dithiol domain, is one of the factors that contribute to the activation of ATP synthase (Nalin and McCarty, 1984; Hisabori et al., 2002; Samra et al., 2006). Thus, by sensing the redox status of the chloroplast, the dithiol domain acts as a switch contributing to the interconversion of ATP synthase between its active and inactive forms.

In this work we show that isolated tomato fruit chromoplasts are capable of de novo ATP production through a respiratory pathway using NADPH as an electron donor. We also show that ATP synthesis involves an ATP synthase harboring an atypical γ–subunit lacking the regulatory dithiol domain present in plant and algae chloroplast counterparts. This chromoplast γ–subunit belongs to a distinct family of plastidial γ–subunits that exist in a permanently reduced state, the members of which are present in a variety of plant species.

Results

De novo ATP synthesis in isolated tomato fruit chromoplasts is dependent on redox co–factors

To confirm that tomato fruit chromoplasts have the capacity to synthesize ATP and also to investigate the mechanisms underlying this process, we performed in vitro assays using isolated organelles. Chromoplasts were incubated in buffer A (100 mm Hepes, pH 7.4, 10 mm MgCl2, 2 mm MnCl2, 10 mm KH2PO4, 1 mm NADPH, 1 mm NADP+, 20 μm FAD, 2 mm pyruvate and 330 mm sorbitol; Angaman et al., 2012), and ATP levels were measured after different periods of time. The ATP content increased up to fourfold after 10 min of incubation, and was maintained at this level during the first 30 min. After 60 min, the ATP levels decreased to basal levels (Figure 1a). ATP production was dependent on the presence of NADPH and NADP+ in the incubation buffer (Figure 1b, I and II). At the same concentration, NADH/NAD+ could not sustain the ATP synthesis observed in the presence of NADPH/NADP+ (Figure 1b, III). Although FAD could not support ATP synthesis in the absence of NADP+/NADPH (Figure 1b, I and II), this co–factor was required for the efficient production of ATP in the presence of NADPH/NADP+ (Figure 1b, IV). These results revealed a synergistic effect of FAD in the synthesis of ATP driven by NADPH/NADP+. NADPH, NADP+ and FAD are co–factors involved in redox processes, and their requirement pointed to the operation of an electron transport system linked to the observed ATP synthesis. This hypothesis was reinforced with the observation that ATP was not produced when NADPH/NADP+ was replaced by glycolytic substrates, like glucose or glucose-6-phosphate (Figure 1b, V and VI). Furthermore, the addition of extra NADPH/NADP+ into the incubated chromoplast samples at regular time intervals (15, 30 and 45 min) resulted in an increase of ATP accumulation compared with the controls, which was maintained at high levels, even after 60 min of incubation (Figure 1c), indicating that the availability of these redox co–factors was limiting for ATP synthesis under the experimental conditions used.

Figure 1.

Relative ATP levels of isolated tomato chromoplasts.(a) Relative ATP levels of samples incubated in buffer A for 0, 5, 10, 15, 30 and 60 min. (b) Relative ATP levels of samples incubated for 0 and 30 min in: buffer A (I); buffer A devoid of NADPH and NADP+ (II); buffer A in which NADPH and NADP+ were replaced by NADH and NAD+ at the same concentrations (III); buffer A devoid of FAD (IV); buffer A in which NADPH and NADP+ were replaced by 10 mm glucose (V); and buffer A in which NADPH and NADP+ were replaced by 10 mm glucose-6-phosphate (VI). (c) Relative ATP levels of samples incubated for 0, 30 and 60 min in buffer A (CON), and in buffer A in which NADPH and NADP+ (1 mm each) were added at regular time intervals (15, 30 and 45 min). (d) Relative ATP levels of samples incubated for 0 and 30 min in buffer A (CON0 and CON30, respectively), and for 30 min in buffer A containing the following inhibitors: 0.1 mm oligomycin (OLIGO), 5 mm N,N'–dicyclohexylcarbodiimide (DCCD), 10 μm carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP), 0.1 mm carbonyl cyanide m–chloro phenyl hydrazon (CCCP) and 0.1 mm 2,4–dinitrophenol (DNP). The ATP determination experiments in (A), (B) and (C) were performed independently at least four times.

ATP synthesis in isolated tomato fruit chromoplasts is blocked by ATP synthase inhibitors and proton gradient uncouplers

Based on the above findings and along with the identification of several ATP synthase complex subunits in the chromoplast proteome (Barsan et al., 2010), we hypothesized that chromoplasts could have retained the ATP synthase complex present in the chloroplasts of green fruits, but driven by alternative electron donors not dependent on photosynthetic activity. To investigate whether ATP production in the isolated tomato chromoplasts was linked to the operation of an active ATP synthase complex, we first evaluated the effect of two well-known ATP synthase inhibitors, oligomycin and dicyclohexylcarbodiimide (DCCD). Although oligomycin is mainly known for the inhibition of the mitochondrial ATP synthase, it is also capable of inhibiting the plastidial ATP synthase if higher concentrations are used (Wu and Berkowitz, 1992). The results shown in Figure 1d revealed that ATP synthesis was inhibited by both inhibitors under the specific experimental conditions used.

Chloroplast ATP synthase activity depends on a ΔpH across thylakoid membranes generated by redox reactions during photosynthesis. To study whether the ATP synthesis observed in chromoplasts was dependent on similar processes, we tested the effect of the uncouplers carbonylcyanide-p-trifluoromethoxyphenylhydrazone (FCCP), carbonylcyanide-m-chlorophenylhydrazone (CCCP) and dinitrophenol (DNP). Uncouplers are hydrophobic weak acids with protonophoric activities that lead to the collapse of transmembrane proton gradients. Given that ATP synthesis is dependent on such gradients, the presence of uncouplers should result in the inhibition of ATP synthase, whereas electron transport should not be affected (Terada, 1990). Indeed, FCCP, CCCP and DNP all resulted in the blocking of ATP synthesis in isolated chromoplasts (Figure 1d).

The respiratory activity of tomato fruit chromoplasts is induced by NADPH and proton gradient uncouplers

The results reported above suggested that ATP synthesis in tomato fruit chromoplasts is dependent on the operation of respiratory activities in these organelles. To test this hypothesis, the respiratory rate of isolated chromoplasts was measured. The results shown in Figure 2 revealed that tomato fruit chromoplasts have a basal respiratory rate that was highly stimulated upon the addition of NADPH and the uncoupler CCCP. The increase of O2 consumption observed upon the addition of CCCP results from the loss of respiratory control after the dissipation of the proton gradient caused by the uncoupler (Palet et al., 1991; Azcon-Bieto et al., 1994). These results reinforce the notion that NADPH acts as the primary electron donor of a redox pathway that uses O2 as the final acceptor in a process coupled to the generation of the membrane ΔpH needed to drive ATP synthesis.

Figure 2.

Rate of oxygen consumption of chromoplasts isolated from MicroTom fruits. The addition of chromoplasts, NADPH (1 mm final concentration) and CCCP (1 μm final concentration) is indicated by arrows. Numbers under the trace line indicate the rate of oxygen consumption expressed in nmol O2 min−1 mg−1 protein. The trace was selected from at least three independent chromoplast preparations with similar results.

An atypical plastid γ–ATP synthase subunit lacking the regulatory dithiol domain is induced during tomato fruit ripening

Chloroplast ATP synthase is a light-regulated enzyme able to synthesize ATP in the presence of light using the proton gradient generated across thylakoid membranes during photosynthesis. It is also well established that the γ–subunit plays a key role in controlling the light-dependent activity of chloroplast ATP synthase. To examine whether chromoplast ATP synthase holds a distinctive regulatory feature to allow ATP synthesis in this particular non-photosynthetic plastid, we first made a search to identify the gene(s) encoding the γ–subunit in tomato. Two unigenes coding for plastid ATP synthase γ–subunits (SGN-U581255 and SGN-U575748) were identified in the tomato genome (SOL Genomics Network database, http://solgenomics.net). Whereas SGN-U581255 encoded a typical chloroplast γ–subunit containing the dithiol domain (hereafter referred to as SlgATPp1), SGN-U575748 encoded a predicted plastid γ–subunit lacking the dithiol domain (hereafter referred to as SlgATPp2). The alignment of the amino acid sequence of SlgATPp1 and SlgATPp2 (79% identity) is shown in Figure 3, together with the sequence of the predicted tomato mitochondrial ATP synthase γ–subunit (SlgATPm) encoded by the unigene SGN-U565849. The dithiol domain is not present in either SlgATPm or SlgATPp2 (Figure 3). The coding region of the cDNAs corresponding to SlgATPp1, SlgATPp2 and SlgATPm were amplified by RT-PCR, and their nucleotide sequence was validated by DNA sequencing. Localization studies using constructs encoding the N–terminal region of SlgATPp1, SlgATPp2 and SlgATPm (including the predicted transit peptides; Figure 3) fused to GFP indicated that SlgATPp1 and SlgATPp2 were targeted to plastids, whereas the chimeric SlgATPm-GFP showed the typical mitochondrial targeting pattern (Figure 4).

Figure 3.

Alignment of the tomato γ–ATP subunits SlgATPp1, SlgATPp2 and SlgATPm performed using clustalw and boxshade. The dithiol domain of the SlATPp1 subunit is boxed. Underlined with a dotted line are the N–terminal regions used for the localization studies. Black boxes indicate identical residues in at least two proteins. Gray boxes indicate conserved residues.

Figure 4.

Subcellular localization of the tomato γATP-subunits. Confocal images showing the localization of chimeric proteins containing the N–terminal region of SlgATPp1, SlgATPp2 and SlgATPm fused to GFP after transient expression in Arabidopsis leaves. Scale bars: 10 μm.

The expression of the genes encoding SlgATPp1 and SlgATPp2 (SlgATPp1 and SlgATPp2) was analyzed in tomato leaves and fruits of different developmental and ripening stages. RNA blot analysis performed with gene-specific probes showed that SlgATPp1 and SlgATPp2 were differentially expressed (Figure 5). Whereas SlgATPp1 is mainly expressed in leaves and green fruits, and its expression decreases during ripening, the opposite applies for SlgATPp2, as its transcript was detected only in fruit, and increased during ripening. The highest levels were observed in ripe red fruit. Additional experiments showed that the expression of SlgATPp2 was induced during fruit maturation in tomato mutants unable to ripen, like rin (ethylene deficient; Giovannoni, 2004) and nr (ethylene insensitive; Giovannoni, 2004), or in the r mutant that does not produce carotenoids during ripening (Fray and Grierson, 1993) (Figure S1).

Figure 5.

Expression analysis of the genes encoding SlgATPp1 and SlgATPp2. Transcript levels were determined by northern blot using gene-specific probes. Equal quantities of total RNAs (15 μg) were analyzed in every lane. Samples of Solanum lycopersicum cv. Ailsa Craig leaves (Lf) and fruit pericarp of different developmental stages (G, green; MG, mature green; OR, orange; and R, ripe red) were used. Ribosomal RNAs (rRNAs) were stained with methylene blue as a loading control. The blots presented here are representative of three independent experiments.

Silencing of SlgATPp2 impairs ATP synthesis in isolated chromoplasts

To confirm that SlgATPp2 actually plays a role in the synthesis of ATP in tomato fruit chromoplasts, its expression was suppressed during fruit ripening using the virus-induced gene silencing (VIGS) method described by Orzaez et al. (2009). This method takes advantage of transgenic tomato plants expressing the Antirrhinum majus Delila and Rosea1 genes (encoding transcription factors responsible for anthocyanin synthesis) specifically in the fruit, to easily visualize the silenced regions (Orzaez et al., 2009). Fruits were agroinfiltrated with a modified VIGS vector incorporating partial sequences from Rosea1 and Delila genes (pTRV2_RD vector, used as control), and with the pTRV2_RD vector containing a short DNA sequence specific for the SlgATPp2 gene (referred to as pTRV2_RD_gATPp2). Silenced regions showed a significant reduction in SlgATPp2 transcript levels in comparison with regions transformed with the control pTRV2_RD vector (Figure 6a). Chromoplasts isolated from the SlgATPp2 silenced and control regions were analyzed for their capacity to produce ATP. As shown in Figure 6b, the reduction of SlgATPp2 transcript levels severely affected ATP synthesis. In chromoplasts isolated from the control regions, ATP synthesis was similar to that observed in chromoplasts isolated from wild-type fruits (Figure 1a), indicating that the silencing process did not interfere with plastid ATP synthesis. The specific role of SlgATPp2 in ATP synthesis in tomato fruit chromoplast is further supported by the observation that the silencing of SlgATPp1 had no effect on ATP synthesis (Figure 6).

Figure 6.

Fruit-specific silencing of the SlgATPp1 and SlgATPp2 gene. (a) Relative transcript levels of SlgATPp1 and SlgATPp2 genes in the silenced sectors of fruits agroinfiltrated with the pTRV2_RD control vector (Con1 and Con2), and with pTRV2_RD_gATPp1 (Sil1) or with pTRV2_RD_gATPp2 (Sil2). Data are means ± SDs of silenced regions from three agroinfiltrated fruits. (b) Relative ATP levels after 30 min of incubation in chromoplasts isolated from silenced sectors of fruits agroinfiltrated with the pTRV2_RD (Con1 and Con2), and with pTRV2_RD_gATPp1 (Sil1) and pTRV2_RD_gATPp2 (Sil2). Data are means ± SDs of a pool of silenced regions of more than 10 fruits from a representative experiment. Experiments have been performed independently four times.

Two different classes of γ–ATP synthase subunits are present in plants

Several lineages of dicot plants contain two or more γ–ATP synthase subunits (Kohzuma et al., 2012). In most species where sequence data are available, both subunits belong to the typical plastidial group containing the regular cysteine motif. However, in specific species, like tomato, two different classes of plastidial γ–ATP synthase subunits co-exist: the typical chloroplast form containing the dithiol domain (type I), and the novel plastidial form with an altered dithiol domain (type II). Figure 7a shows an alignment of the region around the dithiol domain of type-I and type-II higher plant γ–subunits, together with the equivalent region of γ–ATP synthase subunits of algae chloroplasts and cyanobacteria. It can be observed that the majority of higher plant type–II subunits contain a version of the dithiol domain in which at least one of the two regulatory cysteines has been replaced by a different amino acid, in contrast to the tomato type-II subunit where the entire domain is absent. Algae plastid subunits are similar to higher plant type–I subunits as they contain the dithiol domain, whereas the cyanobacterial homologs lack this domain.

Figure 7.

Sequence comparisons and classification of γ–ATP-subunits. (a) Alignment of plastid and cyanobacterial γ–ATP synthase subunits. The dithiol domain and the corresponding region of type–II proteins are boxed. (b) Phylogenetic tree of γ–ATP synthase subunits. Land plant and algal plastid subunits have been analyzed together with cyanobacterial counterparts. Protein prefixes: Am, Acaryochloris marina; Av, Anabaena variabilis; Cr, Chlamydomonas reinhardtii; Ct, Cyanothece sp.; Ds, Dunaliella salina; Gh, Gossypium sp.; Ha, Helianthus annuus; Lj, Lotus japonicus; Ma, Microcystis aeruginosa; Mp, Micromonas pusilla; Mt, Medicago truncatula; Na, Nostoc azollae; Nt, Nicotiana tobacum; Ol, Ostreococcus lucimarinus; Pm, Prochlorococcus marinus; Pt, Populus trichocarpa; Sc, Synechococcus sp.; Sl, Solanum lycopersicum; Te, Thermosynechococcus elongates; Vc, Volvox carteri; Vv, Vitis vinifera. The accession numbers of the proteins used in the alignment and the phylogenetic tree are listed in Table S1.

A phylogenetic analysis has been performed using the higher plant plastidial type–I and type–II γ–subunits and the γ–subunits from cyanobacteria and algae plastids (species listed in Table S1). Three main clades were formed containing the higher plant plastidial subunits, the algae plastidial isoforms and the cyanobacteria homologs (Figure 7b). Plant type–I and type–II γ–subunits are grouped in two paralogous subclades, suggesting that they originated from a common ancestor that was first duplicated and then diverged during the evolution of certain plant lineages.

Discussion

The results presented here clearly show that isolated tomato fruit chromoplasts have the capacity to synthesize ATP through the operation of an ATP synthase complex, driven by a membrane proton gradient generated by an electron transport process in which NADPH acts as an electron donor. At present it is not known whether this feature is specific to tomato fruit chromoplasts or if it is also found in other non-photosynthetic plastids. The report that liposomes containing daffodil chromoplast membrane proteins were able to synthesize ATP in a process also linked to a redox pathway requiring NADPH (Morstadt et al., 2002) leaves open the possibility that autonomous ATP synthesis could be a general feature of chromoplasts.

Although the components of the electron transport system operating in tomato fruit chromoplast to generate the membrane proton gradient required for ATP production are currently unknown, it is likely that they could be related to those proposed to be involved in chlororespiration (Peltier and Cournac, 2002), a process described as a respiratory electron transport chain independent of photochemical reactions that take place in thylakoid membranes. In spite that some aspects related to chlororespiration are still controversial, it has been proposed to be involved in the bioenergetic metabolism of non-green plastids by supplying ATP or by reoxidizing metabolic compounds (Peltier and Cournac, 2002). Chlororespiration involves two main electron transport components, the plastid terminal plastoquinone oxidase (PTOX) and the plastid NAD(P)H:plastoquinone dehydrogenase (Ndh) complex (Peltier and Cournac, 2002). The fact that tomato fruit chromoplasts are derived from chloroplasts present in the green fruit opens the possibility that the components of the electron transport systems involved in chlororespiration could remain active in differentiated chromoplasts, as suggested after the proteomic and translatomic analysis of tomato fruit chromoplasts (Kahlau and Bock, 2008; Barsan et al., 2010).

Plastid terminal plastoquinone oxidase (PTOX) was initially characterized as an enzyme involved in the desaturation of phytoene during carotenoid biosynthesis (Carol and Kuntz, 2001). In agreement with this role, its expression is induced during tomato fruit ripening (Josse et al., 2000). In addition, PTOX has also been related to the synthesis of ATP in daffodil chromoplasts. In this respect, an NADPH-dependent respiratory activity associated with the chemiosmotic synthesis of ATP (referred to as chromorespiration) has been reported in liposomes containing solubilized daffodil chromoplast membrane proteins (Morstadt et al., 2002), and in daffodil chromoplast homogenates and chromoplast membrane preparations (Nievelstein et al., 1995). The results shown in Figure 2 indicated that tomato fruit chromoplasts exhibit active O2 consumption rates that were highly stimulated upon the addition of NADPH and the uncoupler CCCP. Taken together, these results indicate that chromorespiration seems to play an essential role not only in carotenoid biosynthesis but also in the synthesis of ATP in chromoplasts.

The increase of O2 consumption rates in response to NADPH and uncouplers also suggests a major role of the Ndh complex (Figure 2). It could not only act as an electron component, but also as a proton pump generating the membrane ΔpH required for the operation of ATP synthase (Peng et al., 2011). A relevant role of the plastid Ndh complex in tomato fruit chromoplasts has recently been proposed after the observation that tomato mutants lacking a functional Ndh complex do not ripen normally, accumulate significantly less carotenoids and are affected in other chromoplast metabolic processes (Nashilevitz et al., 2010).

An alternative acceptor of electrons from NADPH could be the plastid type–II NAD(P)H dehydrogenase (Ndh2), an FAD-dependent enzyme composed of a single subunit. Ndh2 has a function similar to that of the Ndh complex, although it cannot generate a membrane proton gradient (Geisler et al., 2007). In contrast to higher plants, most green algae do not contain the genes coding for the Ndh complex subunits in their genome, and instead contain genes encoding the Ndh2 enzyme (Desplats et al., 2009). In higher plants, Ndh2 is encoded by a small multigene family (Geisler et al., 2007). Although most of the Ndh2 isoforms are localized into the mitochondria, there are reports describing plastid-targeted isoforms (Eugeni Piller et al., 2011), but their role is not clear at present. According to the sequence data available in public databases (http://solgenomics.net and http://www.kazusa.or.jp/e/index.html), the tomato genome contains at least seven candidate genes encoding Ndh2. Although the localization of the encoded proteins cannot be easily predicted, as some of them carry dual targeting peptides (Carrie et al., 2008), it is probable that at least one of them could operate in plastids. The involvement of Ndh2 as an electron acceptor in tomato fruit chromoplasts may be supported by the observation that the rate of ATP synthesis was significantly reduced when FAD, a co–factor of this enzyme, was not included in the incubation buffer (Figure 1b).

An alternative electron transport system component operating as a proton pump in the chromoplast could be the cytochrome b6f complex, which could accept electrons from the Ndh complex or Ndh2. In this respect, it is worth noting that several subunits of the cytochrome b6f complex have been identified in the tomato chromoplasts proteome (Barsan et al., 2010), and that the presence of this complex in other non-photosynthetic plastids is also well documented (Plöscher et al., 2011). Although cytochrome b6f could represent an alternative candidate for the generation of the membrane proton gradient in the chromoplasts, further studies would be needed to identify the components acting downstream.

The origin of the NADPH required for ATP synthesis in fruit chromoplasts is currently unknown. However, two main processes related to the formation of NADPH can be considered: the oxidative pentose-phosphate pathway and the metabolism of malate by the malic enzyme. It is well known that in non-photosynthetic cells, the oxidative pentose-phosphate pathway is a major source of the NADPH needed for biosynthetic processes (Kruger and von Schaewen, 2003). In agreement with the presence of the full set of enzymes involved in oxidative pentose-phosphate pathway in buttercup flower and pepper fruit chromoplasts (Thom et al., 1998; Tetlow et al., 2003), it is worth noting that almost all enzymes of the oxidative pentose-phosphate pathway have been identified in the tomato chromoplast proteome (Barsan et al., 2010, 2012). The glucose-6-phosphate required to drive this pathway could be imported from the cytosol or obtained from glucose metabolism. The presence of the glucose-6-phosphate transporter in tomato fruit chromoplast has been proposed from proteomic data (Barsan et al., 2010). An alternative source of NADPH in the chromoplast may be via the plastidial metabolism of malate. In developing tomato fruits citric and malic acids accumulate in nearly equal quantities; however, when the fruit begins to ripen, malic acid levels decline whereas citric acid levels increase (Guillet et al., 2002; Roessner-Tunali et al., 2003), indicating that malate metabolism is an active process during tomato fruit ripening. Malic enzyme catalyzes the oxidative decarboxylation of malate to form pyruvate, CO2 and NAD(P)H (Maier et al., 2011). Using isolated tomato fruit chromoplasts we have recently shown that [14C]malate is efficiently taken up and metabolized to pyruvate, thus confirming the presence of malic enzyme in these organelles (Angaman et al., 2012). As the plastidial malic enzyme is NADP+ dependent, the synthesis of pyruvate, a required precursor for the synthesis of carotenoids and lipids in the chromoplast, is linked to the production of NADPH. It can thus be proposed that the malic enzyme could efficiently contribute to the generation of reducing power in the chromoplasts. The NADPH derived from the operation of malic enzyme has also been proposed to be involved in some biosynthetic processes in other non-green plastids (Smith et al., 1992; Kang and Rawsthorne, 1994; Pleite et al., 2005).

Our results show that ATP synthesis in tomato fruit chromoplasts is dependent on the operation of an ATP synthase complex containing an atypical γ–subunit (SlgATPp2) lacking the regulatory dithiol domain present in the chloroplast isoform (SlgATPp1). Although the role of this atypical γ–subunit in the regulation of the ATP synthase complex is currently unknown, it is likely that it can provide novel features to meet the particular metabolic needs of this organelle. It is known that the active form of the chloroplast ATP synthase can catalyze both the synthesis and the hydrolysis of ATP. As the balance of ATP synthesis and hydrolysis is of high importance for the cell energetics, these processes are regulated tightly through several mechanisms. In the presence of light the chloroplast enzyme is fully active, synthesizing ATP, whereas in the dark it is converted to a catalytically inactive form to prevent ATP hydrolysis (Wu et al., 2007). It has been reported that the ΔpH across thylakoid membranes generated through photosynthesis results in conformational changes of the ε–subunit that promotes the exposure of the dithiol domain of the γ–subunit to reducing agents, and thus to the activation of ATP synthesis (Richter and McCarty, 1987; Hisabori et al., 2002; Johnson and McCarty, 2002; Nowak and McCarty, 2004; McCallum and McCarty, 2007). In vivo, the reversible reduction of the dithiol domain is achieved by thioredoxin f, which in turn is reduced by photosystem I via ferredoxin-thioredoxin reductase (Wolosiuk and Buchanan, 1977).

It has been reported that mutations of the cysteine residues of the dithiol domain or deletion of the whole domain affects the redox regulation of chloroplast ATPase activity (Samra et al., 2006). It has also been shown that transgenic plants expressing a permanently reduced form of the chloroplast γ–subunit generated by mutating the cysteine residues are not affected regarding the photosynthetic performance or growth (Wu and Ort, 2008). Despite the essential role of the dithiol domain for the redox regulation of the enzyme, it has been shown that amino acid residues in the vicinity of this domain also participate in the redox regulation of the enzyme (Konno et al., 2000; Samra et al., 2006; Wu et al., 2007). Accordingly, a recent report has shown that the γ2–ATP isoform of Arabidopsis remains reduced under physiological conditions, despite the presence of the dithiol domain, and can be oxidized only by strong agents as a consequence of amino acid changes in the proximity of the redox-active cysteines (Kohzuma et al., 2012). Furthermore, the deletion of either the entire plastid-specific 28 amino acid sequence, containing the dithiol domain, or some regions within the 28 amino acid sequence distal to the dithiol domain results in the loss of the redox regulation and the inhibitory effect exerted by the ε–subunit on the ATPase activity of the enzyme (Samra et al., 2006). In this respect, it is worth noting that the cyanobacteria γ–subunit, the ancestor of the nuclear-encoded viridiplantae plastid isoform, also exerts an important regulatory role in ATP synthase activity mediated by the inhibitory effect of the ε–subunit because of the plastid-specific-like segment, even though it is lacking the dithiol domain (Konno et al., 2006).

Taking into account the above aspects, it can be proposed that SlgATPp2 could behave as a permanently reduced γ–subunit. Furthermore, the amino acid changes observed in the otherwise highly conserved residues of the chloroplast-specific segment (Figure S2) suggests that the SlgATPp2 isoform could also attribute further alternative regulatory mechanisms to chromoplast ATP synthase. Reduction of the dithiol domain of the chloroplast γ-subunit significantly reduces the proton electrochemical potential threshold needed for the activation of the ATP synthase complex (McCallum and McCarty, 2007; Wu et al., 2007). Thus it can be proposed that the replacement of SlgATPp1 by SlgATPp2 during fruit ripening would allow the operation of the chromoplast ATP synthase complex at ΔpH values lower than those present in leaf and green fruit chloroplasts.

The expression analysis of SlgATPp2 in tomato fruits not producing carotenoids (r mutant) or unable to ripen (rin and nr mutants) has shown that the transcript accumulation profile is similar to that of wild-type fruits (Figure S1). This suggests that expression of the gene encoding SlgATPp2 is not regulated by ethylene but rather by different factors controlling fruit development and ripening. Moreover, the similar expression pattern observed in Solanum habrochaites fruits (Figure S1), a wild tomato relative that does not accumulate carotenoids upon reaching maturity, could reflect that the differentiation and evolution of the SlgATPp2 gene from its paralogue SlgATPp1 preceded the diversification of tomato species.

The cyanobacterial γ–subunit does not contain the nine amino acid sequences corresponding to the chloroplast dithiol domain present in mosses, algae and vascular plants, indicating that the insertion of this domain in the plastidial γ–subunit occurred after the gene transfer from cyanobacteria to the viridiplantae ancestors (Figure 7a). Algae genomes contain only one gene encoding the chloroplast γ–subunit, whereas specific classes of higher plants, like dicots, bear two paralogous genes. This suggests that the ancestor of the higher plant γ–subunit gene underwent duplication in specific plant taxa after the separation of algae and land plants. After duplication, the majority of dicot plants maintained two paralogous genes, both encoding for type–I subunits; however, it appears that a second (type–II) plastidial γ–ATP synthase subunit evolved from the diversification of the type–I subunit in specific plant species or families. This is supported by the phylogenetic tree presented in Figure 7b, showing that the subclades of the type–I and type–II γ–subunit paralogues are clustered in close proximity. Thus, whereas the type–I subunits have retained the dithiol domain as inherited from the viridiplantae descendant, the type–II γ–subunits have diversified in response to adaptive selection pressure by losing the cysteine residues of the dithiol domain (either by point mutation or deletion) and, therefore, the ability to be regulated through redox mechanisms. It is possible that the enzymes carrying this type of subunit have evolved to be able to function under the different physiological conditions found in non-photosynthetic plastids (such as tomato chromoplasts), which probably exhibit lower magnitudes of proton electrochemical potential across plastidial membranes because of the lack of photosynthetic activity. As indicated above, it has been reported that the Arabidopsis γ2–ATP isoform behaves as permanently reduced, despite carrying the conserved cysteine motif (Kohzuma et al., 2012). According to this, the presence or absence of the two cysteine residues in the dithiol domain may not be taken as a fixed criterion to define the biochemical role of the corresponding ATP synthase complexes, and suggests that other plant dithiol-containing γ–subunits could actually belong to the type–II group.

Experimental procedures

Chromoplast purification and ATP determination

Chromoplast purification from red ripe cherry and MicroTom tomato fruits was performed according to the method described by Angaman et al. (2012), adding 0.2% of BSA in the extraction buffer. Chromoplasts were incubated at 23°C in buffer A, containing 100 mm Hepes, pH 7.4, 10 mm MgCl2, 2 mm MnCl2, 10 mm KH2PO4, 1 mm NADPH, 1 mm NADP, 20 μm FAD, 2 mm pyruvate and 330 mm sorbitol, based on that described in Angaman et al. (2012). ATP levels were determined in the whole reaction mixture using the ATP Bioluminescence Assay Kit HS II (Roche, http://www.roche.com) and a GLOMAX 96 Microplate Luminometer (Promega, http://www.promega.com).

Oxygen uptake measurements

Oxygen consumption was measured with a liquid-phase Clark-type oxygen electrode (Qubit Systems, http://qubitsystems.com). Isolated chromoplasts were incubated in buffer A for 20 min and then 0.1–ml aliquots were transferred into the electrode chamber containing 0.4 ml of buffer B (300 mm sorbitol, 10 mm N-tris(hydroxymethyl)methyl-2 aminoethanesulfonic acid (TES), 2 mm MgCl2, 5 mm KH2PO4, 0.33 mm EDTA, pH 7.4), which was equilibrated with ambient air. The reaction was carried out at 25°C under constant stirring. NADPH (1 mm final concentration) and CCCP (1 μm final concentration) were added to the electrode chamber using a Hamilton syringe from a stock solution dissolved in distilled water and DMSO, respectively. Protein content was measured using the RC DC Protein Assay kit (Bio-Rad, http://www.bio-rad.com).

Identification and cloning of tomato γ–ATP synthase subunits

Plastid and mitochondrial ATP synthase γ–subunits were identified in the SOL Genomics Network database by Basic Local Alignment Search Tool (BLAST) search using known plant plastid or mitochondrial homologs as a query. Primers used for the full-length cloning of the corresponding cDNAs are listed in Table S2. First-strand cDNA was synthesized from tomato fruit pericarp total RNA, extracted following the method described by Bugos et al. (1995) using the RETROscript First Strand Synthesis Kit (Ambion, now Invitrogen, http://www.invitrogen.com). Amplified DNA fragments were cloned into pGEM-T Easy Vector (Promega) and their identity was verified by DNA sequencing.

Plastid localization experiments

cDNA fragments corresponding to the selected N–terminal regions containing the predicted transit peptides of the tomato gATP subunits according to different prediction software (e.g. signalp 4.0, chlorop 1.1, targetp 1.1 or wolf psort; Figure 3A) were amplified using the primers listed in Table S2 and cloned into the pGFP-MRC vector (Rodriguez-Concepcion et al., 1999) in frame with the GFP (Green Fluorescent Protein), and under the control of the CaMV35S promoter. The constructs generated were used for the transient expression of the chimeric proteins in fully expanded Arabidopsis leaves using tungsten particles (1.0 μm) and the Biolistic PDS-1000/He system (Bio-Rad). After incubation in the dark for 24 h at 22°C, samples were examined for fluorescence using a Leica TCS 4D confocal laser scanning microscope (Leica, http://www.leica.com).

Expression analysis of the tomato γ–subunits

Ailsa Craig tomato plants were grown in the glasshouse under a 16–h photoperiod. Northern blot analysis and hybridization experiments were performed according to Pateraki et al. (2004). Gene-specific probes were designed after comparison of the tomato γ–subunit cDNA sequences. Primers used for amplifying the corresponding cDNA fragments are listed in Table S2. [32P]dCTP radiolabeled probes were prepared with the ‘Ready-To-Go labeling beads’ (GE Healthcare, http://www.gehealthcare.com). Labeled blot signals were visualized, after exposure to a Phosphorimager screen, using the Personal Molecular Imager FX (Bio-Rad).

Transient silencing of SlgATPp1 and SlgATPp2 in tomato fruit

Silencing of the SlgATPp2 gene in tomato fruit was performed using a VIGS system, as described by Orzaez et al. (2009). The gene-specific fragments used for northern blot hybridizations were cloned in the pTRV2_RD vector for the generation of the pTRV2_RD_gATPp1 and pTRV2_RD_gATPp2 plasmids. The primers used are listed in Table S2. pTRV2_RD, pTRV2_RD_gATPp1 and pTRV2_RD_gATPp2 constructs were used for the agroinfiltration experiments. Transcript levels of SlgATPp1 and SlgATPp2 were determined using semi-quantitative RT-PCR. The primers used for the fragment amplification were also used in this case (Table S2).

Phylogenetic analysis

Phylogenetic analyses were performed using the Neighbor-Joining method in mega 5 (Tamura et al., 2011). The bootstrap consensus tree was inferred from 1000 replicates. Branches corresponding to partitions reproduced in less than 50% of bootstrap replicates are collapsed. The accession numbers of protein sequences analyzed are listed in Table S1.

Acknowledgements

We are grateful to Diego Orzaez for providing seeds of the Del/Ros1 tomato plants and the pTRV1 and pTRV2_RD vectors. We thank the staff of the Instal.lació Radioactiva and Servei de Microscòpia Confocal of the Facultat de Biologia and the Serveis Cientificotècnics and Serveis de Camps Experimentals of the Universitat de Barcelona. This work was supported by grants from the Spanish Ministerio de Ciencia e Innovación (BIO2009-09523 to AB, including European Regional Development Funds), the Spanish Consolider-Ingenio 2010 Program (CSD2007-00036 Centre for Research in Agrigenomics) and the Generalitat de Catalunya (2009SGR0026). MR was the recipient of a pre-doctoral fellowship from the Spanish Ministerio de Economia y Competitividad. We also thank Michael Phillips, Sigrid Naudé and Allison Maree Heskes for editorial help.

Ancillary