Allan G. Rasmusson, Lund University, Sweden; Ian M. Møller, Aarhus University, Denmark

The electron transport chain (ETC) in the inner mitochondrial membrane of most eukaryotes consists of four large protein complexes, the rotenone-sensitive NADH dehydrogenase (complex I), succinate dehydrogenase (complex II), cytochrome bc₁ complex (complex III) and cytochrome c oxidase (complex IV). The electron transport activity of complexes I, III, and IV (the latter two often jointly called the cytochrome pathway) is coupled to extrusion of protons across the inner mitochondrial membrane. The electrochemical proton gradient that is formed and maintained in the process is used by the ATP synthase (also called complex V) to make ATP (Web Figure 12.3.A). The plant ETC, however, is highly branched. As a consequence, the plant oxidative phosphorylation system has several alternative pathways of electron transport via type II NAD(P)H dehydrogenases and the alternative oxidase (AOX), and additionally, an alternative proton transporter called the uncoupling protein (UCP). These proteins mediate bypasses around the proton-translocating multiprotein complexes (I–V): type II NAD(P)H dehydrogenases bypass complex I, AOX bypasses complexes III and IV, and UCP bypasses the ATP synthase (Siedow and Umbach 1995; Vanlerberghe and McIntosh 1997; Møller 2001; Millenaar and Lambers 2003; Fernie et al. 2004; Rasmusson et al. 2004, 2008; Vercesi et al. 2006; Vanlerberghe 2013). The type II NAD(P)H dehydrogenases and the AOX transport electrons without pumping protons, whereas the UCP allows proton flow from the intermembrane space to the matrix without ATP synthesis. The consequence of this is that flux through the energy conservation bypasses will mediate respiration that does not contribute to ATP production. At the same time, type II NAD(P)H dehydrogenases and AOX will not be controlled by cellular adenylate status. This Web Topic will describe the special features of these energy bypass proteins and discuss the consequences of their existence to the plant.

Web Figure 12.3.A Organization of the electron transport chain and ATP synthesis in the inner membrane of plant mitochondria. In addition to the four standard protein complexes found in nearly all mitochondria, the electron transport chain of plant mitochondria contains additional enzymes (depicted in green). None of these additional enzymes pumps protons. Additionally, the uncoupling protein directly bypasses the ATP synthase by allowing passive proton influx. NAD(P)H is expected to freely diffuse between the intermembrane space and the cytosol due to the presence of porins in the outer membrane. The multiplicity of bypasses in plants, where animals have only the uncoupling protein, gives a greater flexibility to plant energy coupling. Specific inhibitors—rotenone for complex I, antimycin for complex III, cyanide for complex IV, and salicylhydroxamic acid (SHAM) for the alternative oxidase—are important tools used to investigate mitochondrial electron transport. Some also have commercial uses, such as rotenone—which is used as an insecticide and to remove unwanted fish from lakes, without affecting plants. Because plants have the alternative pathways, they can survive exposure to inhibitors of the respiratory complexes. Abbreviations: AOX, alternative oxidase; ND(B1/B2/B4/A1/A2), various type II NAD(P)H dehydrogenases.

NAD(P)H Dehydrogenases in Plant Mitochondria

Except for yeast, the ETC of all eukaryotes—including plants—contains a proton-pumping NADH dehydrogenase, complex I, which is inhibited specifically by rotenone (see Web Figure 11.3.A). Additionally, plant, fungal, and protist mitochondria contain various versions of rotenone-insensitive enzymes capable of oxidizing NADH and/or NAD(P)H and passing the electrons on to ubiquinone. These NAD(P)H dehydrogenases are classified as type II, whereas complex I is a type I NAD(P)H dehydrogenase. Type II NAD(P)H dehydrogenases are relatively small, consisting of a single polypeptide of 45–60 kDa, whereas complex I is a complex of ca. 45 polypeptides with a total molecular mass approaching 1 MDa. The inner mitochondrial membrane in plants contains type II NAD(P)H dehydrogenases that face either the matrix or the intermembrane space, thus oxidizing NAD(P)H either from the matrix or the cytoplasm (see Web Figure 11.3.A). In general, photosynthetic organisms have more genes for type II NAD(P)H dehydrogenases than heterotrophic organisms (Møller 1997; Møller and Rasmusson 1998; Rasmusson et al. 1998; Kercher 2000; Møller 2001; Rasmusson et al. 2004, 2008). Genes encoding type II NAD(P)H dehydrogenases have also been found in some clades of “primitive” animals, for example sea squirts, sea urchins, and Hydra (Matus-Ortega et al. 2011).

Plant type II NAD(P)H dehydrogenases constitute three small protein families: NDA, NDB and NDC. All are present in mitochondria, but for particular proteins, functions in other compartments have been demonstrated and/or suggested by targeting analyses. In potato and Arabidopsis mitochondria, NDA and NDC proteins have been localized to the matrix side of the inner mitochondrial membrane, whereas NDB proteins are bound to the outer surface of the inner mitochondrial membrane (Rasmusson et al. 1999; Michalecka et al. 2004; Elhafez et al. 2006; Carrie et al. 2008).

The outer membrane of all mitochondria contains NADH-ferricyanide and NADH-cytochrome b reductase activities that are insensitive to rotenone and to the complex III inhibitor, antimycin A. However, the outer membrane activities are not linked to the inner membrane ETC and will not be treated further here.

There Are Three Types of NAD(P)H Dehydrogenase Activities on the Inner Matrix Surface of the Inner Membrane

Much of the information about the number of matrix-facing NAD(P)H dehydrogenases has come from experiments using inside-out inner membrane vesicles. Such vesicles have the matrix surface facing the medium and the external NAD(P)H dehydrogenases hidden inside the vesicles and unable to interfere with the assays. This method has been complemented with a technique where the membrane-permeabilizing peptide alamethicin is used to make a channel for the substrate through the inner membrane (see Web Topic 12.1). Experiments with these methods have shown the presence of three NAD(P)H dehydrogenase activities on the matrix side of the inner membrane (see Web Figure 12.3.A) (Møller and Palmer 1982; Rasmusson and Møller 1991; Melo et al. 1996; Agius et al. 1998; Johansson et al. 2004):

• Complex I pumps protons, is Ca²⁺-independent, and is inhibited by rotenone and diphenylene iodonium (DPI). Complex I has a high affinity for NADH [low K_m(NADH)].
• ND_in(NADH) does not pump protons, is Ca²⁺-independent, and is DPI-insensitive. The enzyme has a tenfold lower affinity for NADH than complex I [i.e., a tenfold higher K_m(NADH)]. This activity is most likely catalyzed by NDA1 and NDA2.
• ND_in(NADPH) does not pump protons, is Ca²⁺-dependent, and is sensitive to DPI. The protein catalyzing this activity has not been identified, but may be NDC1.

Because of the contrasting affinities of complex I and ND_in(NADH) for NADH, only complex I is expected to be engaged at low matrix concentrations of NADH. However, when the NADH concentration in the matrix increases (e.g., if the ETC is restricted by lack of ADP in the cell), ND_in(NADH) can become engaged. However, the exact concentrations of NADH in the matrix under different physiological conditions are not fully clarified. This is partly due to the complex kinetic regulation of NADH formation (Hagedorn et al. 2004) and partly due to the fact that a significant proportion of matrix NADH is protein-bound instead of free in solution (Kasimova et al. 2006).

When electrons are transported through the type II NAD(P)H dehydrogenases, only the two proton-pumping sites of the cytochrome pathway are involved in the flow of electrons from matrix NADH to oxygen. As a consequence, the theoretical ADP/O yield is 1.5, unless the AOX or the UCP are active, in which case the yield will be even lower.

Type II NAD(P)H dehydrogenase proteins of NDA-type are present in mitochondria, inside the inner membrane barrier, in potato and Arabidopsis (Rasmusson et al. 1999; Michalecka et al. 2003; Elhafez et al. 2006). Evidence from expression and mutant analysis as well as sequence comparisons collectively indicate that they are NADH dehydrogenases (Svensson and Rasmusson 2001; Moore et al. 2003; Michalecka et al. 2004; Elhafez et al. 2006; Escobar et al. 2006). However, the gene identity of the internal NADPH dehydrogenase has yet to be determined.

In the yeast Saccharomyces cerevisiae, complex I is absent, so a single type II NADH dehydrogenase (Ndi1) catalyses the oxidation of matrix NADH. The Ndi1 protein sequence is overall relatively similar to the plant NDA proteins. As the only type II NAD(P)H dehydrogenase in yeast, Ndi1 has been crystallized and its structure determined (Web Figure 12.3 B). It is a homodimer of two identical ≈50 kDa polypeptides, containing flavin adenine dinucleotide (FAD) as its only electron-carrying prosthetic group. The C-terminal domain of each subunit forms alpha helices that are partially embedded in the membrane. This positions the FAD close to the membrane surface, where it is accessible to both NADH from the aqueous matrix compartment and ubiquinone from the hydrophobic membrane interior (Iwata et al. 2012; Feng et al. 2012).

Web Figure 12.3.B Structures of a type II NADH dyhydrogenase and an alternative oxidase. The Ndi1 from S. cerevisiae (pdb: 4G9K; Iwata et al. 2012) and the alternative oxidase from Trypanosoma brucei (pdb:3VV9; Shiba et al. 2013) are both homodimers. The reactions are drawn for one monomer each. Each enzyme contains a single internal electron-carrying group; FAD for the Ndi1 and a di-iron center for the alternative oxidase. Both enzymes are partly embedded in the inner leaflet of the inner mitochondrial membrane, using alpha-helices that are parallel to the membrane plane. Because the proteins do not span the membrane, transfer of electrons between the hydrophilic redox couples NADH/NAD⁺ and H₂O/O₂, and the hydrophobic ubiquinone (UQH₂/UQ) cannot involve proton pumping across the membrane, and all the energy released by the reaction is instead given off as heat.

There Are Three types of External Dehydrogenases that Oxidise Cytosolic NAD(P)H

Plant mitochondria can oxidize cytosolic NADH and NADPH directly via external dehydrogenases. None of the external dehydrogenases pumps protons, which means that, at most, two proton-pumping sites are involved when the electrons move from the dehydrogenases to oxygen (see Web Figure 12.3.A). So, as for ND_in(NADH), the theoretical maximum ADP/O yield is 1.5.

The oxidation of exogenous NADH or NAD(P)H by isolated plant mitochondria requires Ca²⁺ and, as a result, is inhibited by EDTA and EGTA, chelators of di- and trivalent cations (Coleman and Palmer 1971; Møller et al. 1981). However, in some plant tissues, NADH oxidation is largely Ca²⁺-independent (Nash and Wiskich 1983, Escobar et al. 2006). The Ca²⁺ concentration needed for half-maximal activity of NADH oxidation is 0.1–0.5 μM, depending on ionic environment and respiratory state (Moore and Åkerman 1982; Rugolo et al. 1991). This is in the same range as the 0.1–0.2 μM free Ca²⁺ thought to be present in the cytosol of a resting plant cell. Therefore, the external NAD(P)H oxidation may become activated by the increased Ca²⁺ concentration associated with cellular excitation caused by external stimuli.

Among type II NAD(P)H dehydrogenases in plants, NDB-type proteins are present on the outside of the inner membrane in mitochondria (Rasmusson et al. 1999; Michalecka et al. 2004; Elhafez et al. 2006). NADPH oxidation is carried out by the NDB1 protein, which does not accept NADH and is completely dependent on Ca²⁺ (Michalecka et al. 2004; Geisler et al. 2007). In Arabidopsis, NDB2 and NDB4 are NADH dehydrogenases, the former being stimulated by Ca²⁺, but the latter being Ca²⁺-independent (Geisler et al. 2007). Therefore, the variation in Ca²⁺-dependence of NADH oxidation between plant organs is most likely due to variations in abundance of Ca²⁺-dependent and -independent NDB proteins.

Mammalian mitochondria lack external NAD(P)H dehydrogenases. Instead, they have a Ca²⁺-dependent glycerol-3-phosphate dehydrogenase on the outer surface of the inner membrane, which donates electrons directly to the ubiquinone pool. The dihydroxyacetone phosphate formed is converted back into glycerol-3-phosphate by a cytosolic NAD-dependent glycerol-3-phosphate dehydrogenase. In this way, the two low-molecular-weight compounds, glycerol-3-phosphate and dihydroxyacetone phosphate, can shuttle reducing power (a collective term for high-energy electrons carried by, for example, NADH, NADPH, ferredoxin or an organic acid redox couple like malate/oxaloacetate) from the cytosol, across the outer membrane, and to the inner membrane to yield the same amount of ATP as would be produced by a plant-type oxidation via the external NADH dehydrogenase. Yeast mitochondria have both external NADH dehydrogenases and a glycerol-3-phosphate shuttle, and there is evidence that plants may also possess both systems (Shen et al. 2006).

Another way to oxidize cytosolic NAD(P)H, which is found in mammalian mitochondria, is to use the malate/aspartate shuttle to transfer reductant into the matrix as NADH that can be oxidized by the internal NAD(P)H dehydrogenases. However, it is not clear whether this shuttle can work to import cytosolic reductant into plant mitochondria (see Web Topic 12.5). In the opposite direction, NADH can be shuttled from the matrix to the cytosol via the malate/oxaloacetate shuttle. In this way, the external NADH dehydrogenase can also work as a bypass around complex I. Support for this bypass has been seen in the form of an upregulation of external NADH dehydrogenase activity in mutants where complex I or an internal type II NAD(P)H dehydrogenase has been inactivated (Sabar et al. 2000; Moore et al. 2003).

The total picture of this multitude of enzymes tells us that plants contain a highly flexible system. Reducing power can move between compartments of the cell and enter the ETC through different dehydrogenases, with different regulation and with different yields of energy in the form of ATP.

The Alternative Oxidase

The presence of the AOX is one of the features that sets plant mitochondria apart from mammalian mitochondria. This enzyme has therefore received a lot of attention from plant scientists, and we now have a fairly good understanding of the regulation of its structure and activity (Vanlerberghe and McIntosh 1997; Millar et al. 2011; Vanlerberghe 2013; Moore et al. 2013). AOX is generally present and expressed in plants and plant organs. It also occurs in some species of fungi and protists, and has even been found, contrary to previous beliefs, in several groups of “primitive animals” but not in vertebrates or arthropods (McDonald et al. 2009).

The AOX is a quinol-oxygen oxidoreductase and it does not pump protons. The oxygenase transfers electrons from ubiquinol (the reduced form of ubiquinone) to oxygen and generates water as the end product of the reaction (see Web Figure 12.3.B), so four electrons must be transferred to oxygen. The protein contains a catalytic di-iron center (Siedow and Umbach 2000; Berthold and Stenmark 2003; Shiba et al. 2013). The functional form of the enzyme is a dimer with the two polypeptides either covalently or non-covalently bound to each other (see Web Figure 12.3.B). The structure positions the protein so that it is embedded in the inner leaflet of the inner mitochondrial membrane, with the active site exposed to the matrix.

Web Figure 12.3.C The structure and regulatory features of AOX in plant mitochondria. Synthesis of AOX is regulated transcriptionally by several cellular conditions, including elevated levels of ROS, organic acids, and sugars, in response to stress and during development. Additionally, AOX is modified posttranscriptionally by a mitochondrial thioredoxin system that mediates a conversion between a relatively inactive form with a disulfide bridge between the monomers and a more active form with free thiol groups. Finally, in the AOX from many plant sources, a high amount of substrate will increase the activity directly by a posttranslational binding of pyruvate (Pyr) to the reduced enzyme and by increases in the concentration of the substrate ubiquinol (UQH₂) inside the inner membrane.

When electrons from oxidation of matrix NADH flow through the AOX, at least two of the three sites of proton translocation (complexes III and IV, see Web Figure 12.3.A) are bypassed, so energy conservation in the form of ATP is much smaller when the AOX is active. The activity of the AOX must therefore be carefully regulated. The genomes of many plant species encode several AOX isoforms; for example, five genes encoding AOX proteins are present in Arabidopsis. The relative amounts of AOX isoforms, and their mRNAs, vary with development, cellular redox status, and external factors (Wagner and Krab 1995; McCabe et al. 1998; Millenaar and Lambers 2003; Clifton et al. 2006).

Apart from regulation of AOX gene expression, which determines levels of AOX protein, the enzyme’s activity is also regulated posttranslationally (see Web Figure 12.3.C). The two subunits of the AOX dimer can be covalently linked through a disulfide bridge between cysteine residues. The reduced, non-covalently linked dimer is the active form, and this form appears to be the predominant one in vivo (Vanlerberghe et al. 1999). A mitochondrial thioredoxin system regulates AOX activity (Gelhaye et al. 2004). Thioredoxins regulate enzymes by reducing disulfide bridges, and it is likely that the regulation of AOX involves the redox-active cysteines. The presence of thioredoxin regulation connects the AOX activation level to the mitochondrial NADPH pool and the oxidation of citric acid cycle metabolites by NADP-utilizing enzymes (Vanlerberghe and McIntosh 1997; Møller and Rasmusson 1998).

The substrate supply may regulate AOX activity both directly and indirectly. The activity of many AOX enzymes, including, for example, soybean AOX, is stimulated by 2-oxo organic acids—in particular, pyruvate (see Web Figure 12.3.C), which activates AOX by interacting with the same cysteines that are involved in dimerization (Rhoads et al. 1998; Vanlerberghe et al. 1998). This activation may be a feed-forward mechanism that upregulates the capacity of the ETC when the glycolytic end-product pyruvate is available. Both the cytochrome pathway and the AOX pathway use ubiquinol as the substrate. In the absence of pyruvate, the cytochrome pathway is saturated at 40% Q_red/Q_tot (40% of the UQ is reduced), at which point the AOX pathway is not engaged at all. The AOX pathway starts accepting electrons at 50% reduction and is not fully saturated at 80% (Web Figure 12.3.D). In the presence of pyruvate, the AOX pathway becomes engaged at a much lower Q_red/Q_tot (20%), whereas the cytochrome pathway is unaffected. The regulation of AOX by pyruvate has, however, been questioned based on the observation that the normal in vivo concentration of pyruvate appears to be higher than the concentration of pyruvate required to give maximal stimulation of AOX activity in vitro (Millenaar et al. 1998).

Web Figure 12.3.D Dependence of respiratory rate on quinone redox state in purified soybean cotyledon mitochondria oxidizing succinate in the absence (top graph) or presence (bottom graph) of pyruvate. Q_r is ubiquinol (reduced ubiquinone) and Q_t is total ubiquinone (reduced + oxidized). Myxothiazol is an inhibitor of complex III and, in its presence, only the AOX is active. (From Hoefnagel et al. 1995.)

Plant Mitochondria Also Contain an Uncoupling Protein

Mammalian mitochondria contain neither type II NAD(P)H dehydrogenases nor AOX. However, they have a UCP that increases the proton permeability of the inner mitochondrial membrane and, in that way, dissipates the proton gradient. This is another mechanism for reducing the ATP production and increasing heat production, for example, for thermoregulation (see also Web Essay 12.6). Surprisingly, plant mitochondria also contain a UCP (Vercesi et al. 1995; Laloi et al. 1997; Jezek et al. 2000; Fernie et al. 2004; Vercesi et al. 2006). We do not know for certain why plant mitochondria should require two different mechanisms for achieving the same end result. The UCP is related to anion carriers. In vitro, UCP's transport activity is dependent on free fatty acids and is inhibited by nucleotides like ATP and GTP but stimulated by superoxide (Considine et al. 2003). The latter result indicates that the UCP may protect against oxidative stress (see Web Essay 12.7).

The Physiological Impact of Energy Bypass Proteins

Since plant respiration consumes roughly half of the photosynthetic assimilate, plant respiration is one of the major processes on earth. A substantial part of the total respiratory flux proceeds via pathways that bypass energy conservation and produces little or no ATP. Superficially viewed such respiration would be of little use to plant growth and maintenance. However, research especially in recent years has provided several insights into the ways in which respiratory energy bypass pathways may be beneficial to plants.

The respiratory rates of some thermogenic (heat-producing) flowers are the highest among plants and, in fact, exceed even those of warm-blooded animals. The external NADH dehydrogenase, AOX, and UCP appear particularly active in thermogenic flowers, where they are primarily responsible for heat production (Web Essay 12.6). At the same time it is clear that the respiratory rate of most plant tissues is insufficient to generate an appreciable increase in tissue temperature.

Thermogenesis is restricted to relatively few plant species. In non-thermogenic plants, synthesis of AOX is up-regulated at the transcriptional level by several biotic and abiotic stresses, including cold stress (Fiorani et al. 2005; Vanlerberghe 2013). However, the cold induction is clearly not to heat the tissue. During cold stress, reactive oxygen species (ROS) are formed, and other stress treatments that lead to elevated ROS also induce AOX (Clifton et al. 2006). ROS (e.g., superoxide and hydrogen peroxide) are formed as by-products of electron transport under aerobic conditions (see Web Essay 12.7). ROS can cause damage to proteins, lipids, and DNA and the cell must therefore limit their formation (Møller et al. 2007). AOX helps prevent overreduction of the ETC and thus lowers ROS production (Purvis and Shewfelt 1993; Millenaar et al. 1998). Supporting this, Maxwell et al. (1999) demonstrated the importance of the AOX for limiting ROS formation in living plant cells. Overexpression of AOX in cultured cells lowered the steady state level of ROS, whereas suppression of AOX increased it. Similar results have been observed in intact plants (Cvetkovska and Vanlerberghe 2013). These results point to the mitochondrion as an important site of ROS generation and ROS signaling in plant cells (Rhoads and Subbaiah 2007; Vanlerberghe 2013). With the discovery of ROS stimulation of plant UCP and the increased ROS resistance in plants overexpressing UCP (Brandalise et al. 2003; Considine et al. 2003), the picture has become more complex. By decreasing the resistance for electron transport through the proton-pumping complexes, UCP activity may, like AOX, prevent overreduction and ROS formation (Fernie et al. 2004). Another observation that supports UCP having a role in ROS protection is that AOX activity is inhibited by an ROS product (Winger et al. 2005). For further information on mitochondrial involvement in ROS metabolism, see Web Essay 12.7.

Type II NAD(P)H dehydrogenases and AOX may both act as reductant overflow enzymes by reoxidizing surplus NADH with oxygen (Millenaar and Lambers 2003). In other words, the enzymes would improve redox homeostasis by preventing overreduction of the electron transport chain. Reductant overflow can prevent inhibition of the citric acid cycle under conditions where the ADP concentration is low and thus the proton-pumping complexes are restricted by a high membrane potential. In this way, the citric acid cycle can supply carbon skeletons for biosynthesis (e.g., 2-oxoglutarate for nitrogen assimilation; see textbook Figure 12.14), even if the ATP level in the cell is high (Noctor and Foyer 1998). There are other cases where NADH dehydrogenases and AOX in concert may allow an oxidation of surplus NADH. Examples include switching from nitrate (for which a large amount of reductant is needed for assimilation) to ammonium as a nitrogen source (Escobar et al. 2006), and during phosphate deficiency (Sieger et al. 2005). A further indication of cooperative function of NADH dehydrogenases and AOX is that NDB-type NADH dehydrogenases are strongly coexpressed with alternative oxidases (Clifton et al. 2005; Clifton et al. 2006; Ho et al. 2007). In this case, reductant would be shuttled from the mitochondrial matrix to the intermembrane space, where the NADH would be oxidized on the outer surface of the inner mitochondrial membrane.

The concept of overflow is also relevant for photorespiratory conditions (see textbook Chapter 8), under which there is a massive flow of glycine that has to be oxidized by the mitochondria. As much as half of the NADH produced in this process is probably exported, but the other half has to be oxidized promptly (Krömer 1995). There are several reasons to believe that a reductant overflow via energy bypass proteins is involved. The ND_in(NADH) and AOX capacity in leaf mitochondria is higher than in mitochondria from heterotrophic tissue (Dry and Wiskich 1985; Igamberdiev et al. 1997), and both proteins are down-regulated by dark treatment (Svensson and Rasmusson 2001). Consistent with these observations, AOX activity in vivo (see Web Essay 12.9 for a description of the method) is up-regulated by light, AOX capacity is induced by greening of etiolated tissue, and expression of the nda1 gene (encoding an internal NAD[P]H dehydrogenase) in potato and Arabidopsis is light-dependent (Atkin et al. 1993; Ribas-Carbo et al. 2000, 2008; Svensson and Rasmusson 2001; Michalecka et al. 2003; Escobar et al. 2004; Rasmusson and Escobar 2007). However, only some of the light-responsive NADH dehydrogenase and AOX genes in Arabidopsis were responsive to high light treatment, where overflow load should be at the maximum (Yoshida et al. 2007, 2009). Apart from NADH dehydrogenases and AOX, overflow may also be mediated by the cytochrome pathway together with UCP, which may improve reoxidation of photosynthetically derived surplus NADH. In support of this idea, mutants lacking UCP have a decreased photosynthetic rate (Sweetlove et al. 2006), and cytochrome oxidase and AOX activity in vivo are both up-regulated in response to high light (Florez-Sarasa et al. 2011).

Mutant tobacco plants lacking the proton-pumping complex I can still survive, apparently aided by the action of NDin(NADH), which oxidizes matrix NADH with a lower ATP yield. These plants, however, show decreased photosynthetic rates, indicating that the ND_in(NADH) cannot fully compensate for the missing complex I (Noctor et al. 2004, 2007). Also, external NAD(P)H dehydrogenases have been suggested to support photosynthesis. These enzymes may reoxidise NAD(P)H that is shuttled out of the chloroplast in order to balance the NAD(P)H to ATP ratio in the stroma (Krömer 1995). The potato NDB1 external NADPH dehydrogenase is able to modulate the cellular NADPH/NADP-ratio, but an importance for photosynthesis has not been observed (Liu et al. 2008). In summary, there are many indications that mitochondria—especially the energy bypass proteins—have roles in supporting photosynthesis, for example by oxidizing excess NAD(P)H (see also Raghavendra and Padmasree 2003).

A special involvement of the AOX pathway has been observed in the acidification/deacidification cycle of CAM plants. CAM plants fix CO₂ in the form of malate during the night. This is done by cytosolic PEP carboxylase, and the malate produced is stored in the vacuole. The CO₂ fixed is released during the daytime by malic enzymes, located in cytosol and mitochondria (see textbook Chapter 8). Robinson et al. (1992) used the oxygen discrimination technique to measure the contribution of the AOX pathway to whole-leaf respiration in the CAM plant, Kalanchoe daigremontiana. They found that the respiratory rate of the leaves increased by 35% in the early hours of the day. This increase was due to increased AOX activity. A detailed time course showed that the AOX activity increased two- to threefold during the early day at the same time as the malate was metabolized. As soon as the malate had disappeared by mid-day, the AOX activity returned to the low level observed during the night (Web Figure 12.3.E). A reasonable conclusion is that a substantial amount of the malate from the vacuole is oxidized via malic enzyme in the mitochondria. This generates NADH in the matrix, which is oxidized to a large extent via the AOX. The released CO₂ can then be refixed by the Calvin cycle in the chloroplast.

Web Figure 12.3.E The participation of AOX in malate oxidation in Kalanchoe diagremontiana during a night–day cycle. (A) Changes in flux through the AOX pathway in leaf discs. (B) Changes in malate concentration of leaf discs. The solid bar at the bottom of the figure represents the dark period. (From Robinson et al. 1992.)