Monogalactosyldiacylglycerol Deﬁciency in Arabidopsis Affects Pigment Composition in the Prolamellar Body and Impairs Thylakoid Membrane Energization and Photoprotection in Leaves 1[W][OA]

Monogalactosyldiacylglycerol (MGDG) is the major lipid constituent of chloroplast membranes and has been proposed to act directly in several important plastidic processes, particularly during photosynthesis. In this study, the effect of MGDG deﬁciency, as observed in the monogalactosyldiacylglycerol synthase1-1 ( mgd1-1 ) mutant, on chloroplast protein targeting, phototransformation of pigments, and photosynthetic light reactions was analyzed. The targeting of plastid proteins into or across the envelope, or into the thylakoid membrane, was not different from wild-type in the mgd1 mutant, suggesting that the residual amount of MGDG in mgd1 was sufﬁcient to maintain functional targeting mechanisms. In dark-grown plants, the ratio of bound protochlorophyllide (Pchlide, F656) to free Pchlide (F631) was increased in mgd1 compared to the wild type. Increased levels of the photoconvertible pigment-protein complex (F656), which is photoprotective and suppresses photooxidative damage caused by an excess of free Pchlide, may be an adaptive response to the mgd1 mutation. Leaves of mgd1 suffered from a massively impaired capacity for thermal dissipation of excess light due to an inefﬁcient operation of the xanthophyll cycle; the mutant contained less zeaxanthin and more violaxanthin than wild type after 60 min of high-light exposure and suffered from increased photosystem II photoinhibition. This is attributable to an increased conductivity of the thylakoid membrane at high light intensities, so that the proton motive force is reduced and the thylakoid lumen is less acidic than in wild type. Thus, the pH-dependent activation of the violaxanthin de-epoxidase and of the PsbS protein is impaired. et al., 2007). The difference absorption signal was measured using a KLAS-100 LED-array spectrophotometer (Heinz Walz), allowing the simultaneous measurement of light-induced difference absorption signals at six pairs of wave- lengths in the visible range of the spectrum between 500 and 570 nm. The ECS was deconvoluted from signals arising from Z, scattering effects, the C550- signal, and redox changes of the cytochromes, essentially as described by Klughammer et al. (1990), then normalized to the Chl content of the leaf disc. Leaves were illuminated for 10 min prior to each measurement to allow photosynthesis to reachsteady state. The maximum amplitude of the ECS was determined after illuminating the leaveswithsaturating light(2,100 m mol m 2 2 s 2 1 ). Saturating illumination was interrupted by short intervals of darkness (15 s), and the dark-interval relaxation kinetics of the ECS and cytochrome f reduction were measured. The 15-s dark intervals weresufﬁcient for complete relaxationof the pmf .Toobtain lightsaturationcurves of the pmf ,dark-interval relaxation kinetics after illumination at subsaturating light intensities were measured. Pmf partitioning into D pH and the electrochemical component ( DC ) was resolved by analyzing the slowly relaxing phase of the ECS as described by Takizawa et al. (2007).

Galactolipids are the most abundant nonproteinaceous constituents of plastid membranes. Two galactolipids, monogalactosyldiacylglycerol (MGDG) and digalactosyldiacylglycerol (DGDG), are abundant in plant plastid membranes, where they account for as much as approximately 50 and approximately 20 mol%, respectively, of total lipids (Douce and Joyard, 1990). Because of the different sizes of their head groups, MGDG and DGDG differ in shape and biophysical properties. DGDG has a cylindrical shape that is typical for most plastid lipids and is considered a bilayer-prone lipid. In contrast, MGDG has a conical shape due to its smaller head group and high content of unsaturated fatty acids, giving it nonbilayer-forming characteristics (Webb and Green, 1991;Bruce, 1998).
The synthesis of MGDG predominantly takes place in the inner envelope of chloroplasts (Block et al., 1983;Miège et al., 1999). The enzyme MGDG synthase 1 (MGD1) catalyzes the transfer of Gal from UDP-Gal to sn-1,2-diacylglycerol, thus producing MGDG (Benning and Ohta, 2005). The MGD1 gene, as well as the genes for two other MGDG synthases, MGD2 and MGD3, have been identified in Arabidopsis (Arabidopsis thaliana; Miège et al., 1999;Awai et al., 2001). MGD1 is expressed ubiquitously, whereas MGD2 and MGD3 are expressed preferentially in flowers and roots (Awai et al., 2001). Expression is uniformly high for MGD1 during all developmental stages, whereas expression levels of MGD2 and MGD3 are much lower (Awai et al., 2001).
We previously identified an MGDG-deficient mutant (mgd1-1) that carries a T-DNA insertion in the promoter region of MGD1 (Jarvis et al., 2000). The T-DNA insertion causes a 75% reduction in MGD1 mRNA expression, which correlates with a similar reduction in MGDG synthase activity in the mutant. Concentrations of MGDG in mutant leaves are reduced by 42% relative to the wild type. The mgd1 mutant is less green than the wild type, containing only approximately 50% of the normal amount of chlorophyll (Chl) and displays disturbed plastid ultrastructure. These data led to the conclusion that MGD1 is responsible for the synthesis of most MGDG in Arabidopsis. Thus, mgd1-1 is highly suitable for studies on processes in which MGDG is proposed to play an active role, such as chloroplast protein targeting, pigment-protein complex synthesis, and xanthophyll cycle activity.
Chloroplast protein import studies using the Arabidopsis dgd1 mutant, which is approximately 90% deficient in DGDG, revealed a significant decrease in import efficiency (Chen and Li, 1998). Furthermore, a large number of in vitro studies suggested that MGDG plays an important role during chloroplast protein import (Endo et al., 1992;van't Hof et al., 1993;Pinnaduwage and Bruce, 1996). However, some more recent studies were not consistent with this hypothesis (Inoue et al., 2001;Schleiff et al., 2003). So far, in vivo protein transport studies using an MGDG-deficient mutant have not been reported.
The prolamellar bodies (PLBs) of etioplasts have a high lipid to protein ratio compared to thylakoids. While MGDG is the dominant lipid, NADPH:protochlorophyllide (Pchlide) oxidoreductase (POR) is the most abundant protein in PLBs (Selstam and Sandelius, 1984). An interaction between POR and MGDG has been proposed to stabilize the formation of PLBs (Klement et al., 1999;Engdahl et al., 2001;Selstam et al., 2002). Furthermore, POR is thought to suppress the photooxidative damage caused by inactive or ''free'' Pchlide acting as a photosensitizer (Runge et al., 1996;Erdei et al., 2005), which is especially important during de-etiolation (Franck et al., 1995). Indeed, overexpression of POR in Arabidopsis plants considerably reduced photooxidative damage (Sperling et al., 1997). Because PLBs can still be observed in mgd1-1 plants (Jarvis et al., 2000), it is of interest to determine the effect of the MGDG deficiency on the POR-Pchlide-NADPH complex composition in etioplasts.
Several photosynthesis-related reactions are thought to be dependent on galactolipids. For example, the Arabidopsis dgd1 mutation, which causes DGDG deficiency, affects the composition of the water-oxidizing complex (Reifarth et al., 1997), both photosystems (Dörmann et al., 1995;Härtel et al., 1997), and the levels of xanthophyll cycle pigments (Härtel et al., 1998). The xanthophyll cycle pigment violaxanthin is de-epoxidized into antheraxanthin (A) and then into zeaxanthin (Z) by the enzyme violaxanthin deepoxidase (VDE). The two latter pigments are believed to facilitate dissipation of excess light energy in the PSII antenna bed during short-term, high-light stress, a process called nonphotochemical quenching (qN; Gilmore and Yamamoto, 1993;Horton et al., 2008). MGDG is the lipid that is most efficient at promoting VDE activity (Yamamoto, 2006); in vitro studies with purified, heterologously expressed VDE have shown that its activity strongly increases in the presence of MGDG (Latowski et al., 2004). In dgd1, the deepoxidation kinetics were accelerated relative to wild type, which was explained by a slight increase in MGDG levels in dgd1 (Härtel et al., 1998). However, no data on xanthophyll cycle activity in an MGDGdeficient mutant have been presented to date.
The aim of this study was to investigate the in vivo roles of MGDG in various processes in which it is putatively involved, as discussed above. The recently characterized mgd1 null mutant (mgd1-2) has extremely severe defects in chloroplast and plant development (Kobayashi et al., 2007), precluding its use in measurements of chloroplast functions such as those described here. For this reason, we have utilized the original mgd1-1 mutant, which possesses a leaky or knockdown allele and accumulates approximately 40% less MGDG than wild-type plants (Jarvis et al., 2000).

RESULTS AND DISCUSSION
Protein Targeting Into and Across Chloroplast Membranes Is Not Affected in the mgd1 Mutant Based on in vitro studies, MGDG has been suggested to play an important role in protein import into chloroplasts and in the insertion of chloroplast outer envelope membrane proteins (van't Hof et al., 1991(van't Hof et al., , 1993Chupin et al., 1994;Pilon et al., 1995;Pinnaduwage and Bruce, 1996;Bruce, 1998;Schleiff et al., 2001;Hofmann and Theg, 2005). While previous analyses using the Arabidopsis dgd1 mutant clearly demonstrated the importance of DGDG for efficient chloroplast protein import (Chen and Li, 1998;Aronsson and Jarvis, 2002), the role of MGDG in protein targeting processes has not previously been tested in vivo. To address this issue, import studies with mgd1 chloroplasts were done using three nucleus-encoded precursor proteins: the small subunit of Rubisco (pSS; a photosynthetic preprotein), the 50S ribosomal subunit protein L11 (pL11; a nonphotosynthetic preprotein), and subunit II of CF 0 of the photosynthetic ATPase (pCF 0 II; a preprotein that engages the spontaneous thylakoid targeting pathway after chloroplast import; Fig. 1A). Import efficiency was measured using time course experiments. The appearance of the smaller, mature form of each protein, due to transit peptide cleavage, was taken to indicate that import had occurred (Fig. 1A). All three proteins were imported and, in all cases, import was time dependent. Surprisingly, protein import efficiency in mgd1 chloroplasts was similar to that in wild-type chloroplasts; none of the proteins tested showed any significant difference in import efficiency between wild-type and mutant chloroplasts (Fig. 1B).
Next, we assessed the possibility that protein insertion into the outer envelope membrane is affected by mgd1 because MGDG has been suggested to facilitate protein recognition during such targeting processes (Pinnaduwage and Bruce, 1996). We used two different outer envelope membrane proteins, atToc33 and atToc34, both of which lack a transit peptide (like most outer envelope proteins; Hofmann and Theg, 2005). However, we again observed no defect in protein targeting efficiency in the mgd1 mutant. Both proteins inserted equally well into the outer envelope membrane of wild-type and mutant chloroplasts (data not shown), and the inserted proteins withstood extraction by alkaline treatment to the same extent ( Fig. 1C, top) such that no protein was extracted to the soluble supernatant fraction (Fig. 1C, bottom), clearly demonstrating that insertion efficiency was not affected by the mgd1 mutation. These results are similar to those reported previously for OEP14 for the dgd1 mutant (Chen and Li, 1998), suggesting that galactolipids may not play an important role in the insertion of outer envelope proteins in vivo. However, OEP14 is a signalanchored protein, whereas atToc33 and atToc34 are tail-anchored proteins, and it is presently not clear if these different protein classes use the same or different insertion mechanisms (Hofmann and Theg, 2005).
Because MGDG is the major lipid constituent of thylakoid membranes and because CF 0 II targeting follows the spontaneous thylakoid insertion pathway that does not appear to depend on any proteinaceous factors (Michl et al., 1994;Jarvis and Robinson, 2004), we went on to conduct pCF 0 II thylakoid insertion experiments. Following chloroplast import reactions, plastids were treated with thermolysin to avoid possible contamination from envelope-bound CF 0 II, and then thylakoids were isolated and analyzed (Fig. 1D). However, the results revealed that the spontaneous insertion pathway was not significantly affected by the mgd1 mutation because equal amounts of protein were inserted into wild-type and mutant thylakoids. Thus, either the level of galactolipids in mgd1 is sufficient to retain a competent spontaneous insertion pathway, or, alternatively, the role of lipids (and of MGDG specifically) in this pathway is not significant.
Immunoblots were prepared to study the levels of proteins involved in the chloroplast import mechanism (atToc33 and atToc75-III), as well as of some substrates of the import machinery (atToc75-III, Lhcb2, PsaD, PsbS, plastocyanin [PC], and VDE; note that D1, PsbB, PsaC, and b-ATPase are encoded by the plastome and so do not need to be imported), in wild-type and mgd1 plants (Fig. 2). The chloroplast import machinery consists of several different components; the atToc33 and atToc75-III proteins tested here contribute Figure 1. Chloroplast protein translocation and insertion assays. A, Import of different precursor proteins into wild-type and mgd1 mutant Arabidopsis chloroplasts. Import reaction time points are defined as follows: a, 3 min (1 min for pL11); b, 6 min (2 min for pL11); c, 10 min (3 min for pL11). Note that for pL11, a shorter time series was used (1-3 min instead of 3-10 min for the other precursors) because pL11 is known to have a faster import rate (Aronsson and Jarvis, 2002). TM, Translation mixture; p, precursor protein; m, mature protein. B, Quantification of the import data shown in A. The amount of imported, mature protein in mgd1 chloroplasts is expressed as percentage of the amount of imported protein in wild-type chloroplasts (set to 100%). Values shown are means (6SD) derived from at least three assays. C, Direct insertion into the outer envelope membrane. In vitro-translated atToc33 and atToc34 proteins were incubated with wild-type and mgd1 chloroplasts for 30 min under import conditions and then subjected to treatment with Na 2 CO 3 , pH 11.5. The treated samples were then divided into a soluble (S) and an envelope-enriched (E) fraction prior to analysis. D, Direct insertion into the thylakoid membrane by the spontaneous pathway. In vitro-translated pCF 0 II was incubated with wild-type and mutant chloroplasts under import conditions for 30 min. At the end of the import reaction, thermolysin treatment was used to remove any unimported preprotein, and then thylakoid membranes were isolated.
to the receptor and channel-forming functions, respectively, in the outer envelope membrane (Bédard and Jarvis, 2005). For protein import to proceed efficiently, as was seen for pSS, pL11, and pCF 0 II (Fig. 1, A and B), one would expect the expression of translocon components like these to remain largely unchanged. As shown in Figure 2, both atToc33 and atToc75-III are expressed in wild-type and mgd1 plants at similar levels. Similarly, all proteins tested that must be imported into the chloroplast through the TOC/TIC machinery were also present at normal levels in the mgd1 mutant (Fig. 2).
Thus, our data suggest that MGDG may not be as important for chloroplast protein targeting as was suggested on the basis of earlier, in vitro studies. The many proteins normally present in chloroplast membranes are missing in the artificial membrane systems used for such in vitro experiments, and so the MGDG dependency observed in those studies may have been an overestimation. Alternatively, our negative results may be related to the fact that mgd1-1 contains only 42% less MGDG than the wild type, or indicate that some of the other lipids compensate for the loss of MGDG in the mutant (Jarvis et al., 2000). It is conceivable that a larger MGDG reduction would cause significant targeting defects. However, in the extreme case of the mgd1-2 null mutant (Kobayashi et al., 2007), plant and chloroplast development are so badly affected that analysis and interpretation would both be extremely problematic.
Our results (in particular, Fig. 1, A and B) contrast with those reported previously for the dgd1 mutant (Chen and Li, 1998;Aronsson and Jarvis, 2002). This may indicate that DGDG is relatively more important for chloroplast import than MGDG. This possibility is supported by the observation that DGDG was the only galactolipid associated with the TOC translocon complex (Schleiff et al., 2003) and by the fact that in mgd1 the level of DGDG is not altered (Jarvis et al., 2000). An alternative explanation for the different results obtained with dgd1 is that this mutant has a much more severe lipid defect: the dgd1 mutant has a 90% reduction in DGDG levels (Dörmann et al., 1995).

Fluorescence Measurements Reveal Increased Pchlide-F656 Levels in Etiolated Plants
An interaction between POR and MGDG has been proposed to stabilize the formation of the etioplast structure in darkness and to increase the fluorescence of Pchlide at around 655 nm (Klement et al., 1999;Engdahl et al., 2001;Selstam et al., 2002). These observations, together with the alterations in plastid structure seen in mgd1 (Jarvis et al., 2000), led us to investigate the effect of MGDG deficiency on the assembly of plastid internal membranes, either developed in darkness as PLBs and prothylakoids or in the light as thylakoids. The fluorescence from pigmentprotein complexes consisting of Pchlide, NADPH, and POR can be used to assess how MGDG deficiency affects the plant's strategy for the use of incoming light and for protection against an excess of light.
The hook of dark-grown seedlings was subjected to rapid freezing in liquid nitrogen (77 K), and then fluorescence emission was recorded after excitation at 440 nm. Measurements of wild-type samples revealed a lower emission peak for Pchlide-F656 than for Pchlide-F631. By contrast, the mgd1 mutant had a higher peak at 656 nm, significantly increasing the ratio of Pchlide-F656 to Pchlide-F631 relative to the wild type (Fig. 3A). This pattern was even more pronounced when an excitation wavelength of 460 nm was used (data not shown).
Pchlide-F631 is the first form of the pigment to be produced during plastid development, and it is referred to as nonphotoactive Pchlide because it cannot participate in Chl formation (it is in a free or unbound state). By contrast, Pchlide-F656 corresponds to photoactive Pchlide bound to POR. Previous work has suggested that Pchlide-F631 (or an equivalent species) is a photosensitizer, rendering plants susceptible to photooxidative damage (Runge et al., 1996;Sperling et al., 1997;Erdei et al., 2005).
The photoactivity of Pchlide-F656 was verified by flash illumination. The dark-grown 5-d-old seedlings stored at 77 K were rewarmed to 253 K (220°C), flashed with strong light, and then immediately frozen in liquid nitrogen (77 K); thereafter, the emission spectra were recorded (Fig. 3B). The 2-fold higher levels of the chlorophyllide (Chlide)-F688 fluorescence peak in mgd1 can be taken as further evidence for the increased amount of photoactive Pchlide-F656 in the mutant (Fig. 3B). Measurements of dark-grown seedlings of different ages (5, 7, and 8 d old) always revealed an increase in the Pchlide-F656 form in the mutant, excluding the possibility that the differences in fluorescence were due to developmental stage differences (data not shown). To further verify the observed increase in formation of Chlide in mgd1, the degree of phototransformation was determined. A predetermined fluorescence ratio between Pchlide and Chlide of 1.09 (Ryberg and Sundqvist, 1982) was Figure 2. Immunodetection of different chloroplast proteins. Chloroplasts were isolated from 2-week-old wild-type and mgd1 mutant plants and used to prepare immunoblots. Equal (10 mg or 1 mg for PsbS) samples of total chloroplast protein were loaded per lane, and the blots were used to compare levels of several different proteins, as indicated. Antibodies were detected using an ECL system (Amersham Biosciences).
used to determine the degree of phototransformation of Pchlide to Chlide. The values obtained were always higher for mgd1; after one flash, mgd1 gave 51% and wild type gave 45%; after two flashes, the results were 65% and 49%; and, after three flashes, the results were 74% and 64%. Altogether, the data convincingly show that mgd1 has a higher yield of phototransformation of Pchlide into Chlide than the wild type. These results are consistent with the increased level of photoactive Pchlide-F656 observed in mgd1 (Fig. 3A).
Previous work demonstrated that overexpression of POR increases the Pchlide F656 to F631 ratio (equivalent to the F655 to F632 ratio; Franck et al., 2000) and protects the photosynthetic apparatus against photooxidative damage (Sperling et al., 1997). Thus, a low Pchlide F656 to F631 ratio is associated with susceptibility to oxidative damage. Our results suggest that a mechanism to avoid this scenario is activated in the mgd1 mutant. However, this mechanism does not appear to involve up-regulated expression of POR because steady-state levels of the protein were unaltered ( Fig. 3A, inset). The increased Pchlide F656 to F631 ratio in mgd1 may indicate that Pchlide-F656 formation is favored. This not only would serve to minimize the possibility for photooxidative damage but also would maximize the use of incoming light for Pchlide to Chlide transformation, enabling the proper conversion of PLBs and prothylakoids into thylakoids.
The observed differences in pigment-protein composition in dark-grown mgd1 plants (Fig. 3, A and B) are interesting because etioplasts in 5-d-old mgd1 plants were reported to have normal ultrastructure (Jarvis et al., 2000). Therefore, the increase in Pchlide-F656 is due neither to a lack of prothylakoids, which predominantly contain Pchlide-F631, nor to an increase in PLBs, which are highly enriched in Pchlide-F656 (Ryberg and Sundqvist, 1982). This is in agreement with our hypothesis that the accumulation of photoactive Pchlide-F656 (at the expense of nonphotoactive Pchlide-F631) is not simply a passive response but, rather, is part of an adaptive mechanism designed to protect the photosynthetic apparatus against excess light. This may be related to the fact that another photoprotective mechanism (the xanthophyll cycle) is compromised in the mgd1 mutant (see below).
We also investigated the role of MGDG in determining the composition of pigment-protein complexes in thylakoids of light-grown plants (Fig. 3C). No obvious differences in the fluorescence emission for PSII and PSI were observed in plants grown at low light intensity (approximately 100 mmol m 22 s 21 ; data not shown). However, when plants were grown at an elevated light intensity of approximately 200 mmol m 22 s 21 , the fluorescence at around 735 nm was decreased in mgd1 (Fig. 3C). The relative fluorescence intensity at 692 nm, which reflects PSII, was similar for mgd1 and wild type at the light intensity used (Fig.  3C). Because the ratio of the peaks at 692 and 735 nm is a measure for the antenna cross sections of the two photosystems relative to each other, these data suggest that either the accumulation of PSI reaction centers relative to PSII is reduced or that the distribution of light-harvesting complex (LHC) proteins between PSI and PSII is altered in mgd1 at elevated light intensities.

MGDG Is Required for Efficient Photosynthesis and Photoprotection of Thylakoids
Chloroplast ultrastructure was previously reported to be altered in mgd1 (Jarvis et al., 2000). Not only is MGDG a major constituent of the lipid matrix of the photosynthetic membranes, but, as revealed by crystallization studies, it is also an integral component of PSI, PSII, and the cytochrome b 6 f complex (Jordan et al., 2001;Stroebel et al., 2003;Loll et al., 2005). Thus, the MGDG-deficiency of mgd1 might be expected to cause alterations in photosynthetic performance.
To assess this possibility, we first measured photosynthetic pigment composition in light-adapted plants. Pigments levels were analyzed by HPLC and are presented as nanomoles per gram fresh weight in Table I. The total amount of pigments is reduced by approximately 29% in mgd1 relative to wild type due to a 30% reduction in total Chl and a 23% reduction in carotenoids. In mgd1, Chl a is reduced by approximately 28%, and an even stronger reduction is observed for Chl b (35%); these changes are reflected in an increased Chl a to b ratio in the mutant (3.3 in wild type; 3.6 in mgd1). The increased Chl a to b ratio suggests a preferred maintenance of Chl a-rich reaction centers at the expense of Chl b-rich antenna complexes (which may be a mechanism to enable more efficient use of light energy with limited pigment resources). The quantification of photosynthetic complexes and low-temperature Chl a fluorescence measurements is certainly supportive of this interpretation (see below).
The major carotenoid pigment, lutein, was reduced in abundance in mgd1 by 33.7% (on a per fresh weight basis; Table I). Moreover, the amounts of all xanthophyll cycle pigments were reduced in mgd1 (Table I). V was reduced by 8.5%, Z was reduced by 37.5%, and neoxanthin was reduced by 33.7% (all on a per fresh weight basis). When the pigment concentrations were normalized to the amount of Chl a (mmol pigment/ mol Chl a; Table I), most pigment values were still reduced in mgd1 relative to wild type.
To characterize the function of the electron transport chain, the quantum yield of PSII, a measure of photosynthetic efficiency, was determined in dark-adapted leaves (F v /F m ). Measurements of F v /F m in wild type and mgd1 were indistinguishable (data not shown), indicating that a reduction in MGDG content of approximately 40% has no impact on photosynthetic efficiency in low-light-grown plants. Next, we determined light saturation curves of linear electron transport from PSII yield measurements. Because the absorption properties of wild-type and mgd1 leaves were unaltered, despite the approximately 30% reduction in pigments, and because quantification of the photosynthetic complexes indicated only a minor change in the stoichiometry of the photosystems and their antenna cross sections (see below), it is legitimate to calculate linear electron flux from PSII quantum efficiency (Genty et al., 1989). These measurements revealed a 30% reduction of linear electron flux capacity per leaf area in the mgd1 mutant (Fig. 4A). These data are independently supported by gas exchange measurements, revealing an approximately 25% reduction of light-and CO 2 -saturated leaf assimilation capacity per leaf area in the mutant (18.7 6 2.5 mmol CO 2 m 22 s 21 in the wild type; 13.9 6 3.7 mmol CO 2 m 22 s 21 in mgd1).
While the diminished linear electron flux and assimilation rate per leaf area point to a reduced efficiency of photosynthesis in mgd1 (Fig. 4A), the functional organization of the electron transport chain does not seem to be strongly affected by MGDG deficiency in mgd1; when electron transport rates are recalculated on a Chl basis (e.g. per photosynthetic unit), they were comparably high in mgd1 and the wild type. This was confirmed by quantifications of the photosynthetic com- plexes by means of immunoblotting (Fig. 2) and difference absorption spectroscopy (Table II).
Components analyzed by immunoblotting were as follows. For PSII, we analyzed the antenna complex polypeptide, Lhcb2, the two essential core complex proteins, D1 and PsbB, and PsbS localized at the periphery of PSII. For PSI, we analyzed PsaC and PsaD, which are localized at the periphery of PSI but which nevertheless represent reasonable indicators of PSI abundance, as in the absence of PsaC, no stable accumulation of PSI is possible (Takahashi et al., 1991). We also analyzed two proteins (PC and the b-subunit of the ATP synthase) not directly associated with either PSII or PSI but that are nonetheless important to maintain optimal photosynthesis. As shown in Figure 2, none of these proteins was expressed at significantly different levels in mgd1 relative to wild type. The results presented in Figure 1 are also consistent with this conclusion.
These conclusions are supported by difference absorption spectroscopy data (Table II). While the content of PSII was slightly increased in the mutant (approximately 120% of the wild-type level), the contents of the cytochrome b 6 f complex, PSI, and PC were similar in the wild type and mutant. The slight elevation of PSII content, which is below the resolution of our immunoblot analysis, may explain the increased Chl a to b ratio of the mutant; because the PSI content was unaffected, the ratio increase can only be explained by the loss of LHC antenna proteins of the photosystems, as alluded to earlier. This conclusion is in line with 77-K Chl a fluorescence emission signals determined on leaves of 28-d-old plants; the PSI fluorescence emission signal at 735-nm wavelength is slightly smaller in the wild type than in the mutant, which is in agreement with a slightly elevated PSII to PSI ratio in mgd1 thylakoids (Fig. 3C).
qN Is Impaired in the mgd1 Mutant In addition to having lower photosynthesis capacity per unit leaf area (Fig. 4A), Chl a fluorescence light response curves revealed that mgd1 has a much lower capacity for qN (qN is approximately 0.8 and 0.55 in wild-type and mgd1 leaves, respectively, at light intensities exceeding 1,000 mmol m 22 s 21 ; Fig. 4B). Con-sequently, the mgd1 mutant was clearly more susceptible to light stress. After leaves had been exposed to 1,000 mmol m 22 s 21 of actinic light for 1 h, the proportion of photoinhibited PSII was calculated from the dark-relaxation kinetics of qN. The qI component, the fraction of qN relaxing with a half-time .15 min (which correlates with PSII photoinhibition; Krause and Weis, 1991), was much larger in mutant than in wild-type leaves (0.25 versus 0.11, respectively), confirming that PSII in mgd1 is much more prone to photoinhibition under light-stress conditions. Thus, the reduction in the rapidly reversible component of qN (qE), which is related to thermal dissipation of excess energy in the PSII antenna bed, can be concluded to be even more dramatically impaired in mgd1 than suggested by their lower qN values in saturating light.
Reductions in the capacity of qN could be due to any of several possible defects in the mgd1 mutant because qE is essentially induced by acidification of the thylakoid lumen (which is itself influenced by numerous factors). Under standard growth conditions used during this study, the pH of the thylakoid lumen is predicted to be in the region of 6.5 (Takizawa et al., 2007). Upon high-light stress, it decreases and triggers photoprotective mechanisms via changes in the PSII antenna system. The PsbS protein, which is involved in thermal dissipation, is protonated and initiates a conformational change in the PSII-LHCII supercom- . Light response curves for linear electron transport and qN. A, Linear electron transport was calculated from PSII quantum yield in wild-type and mgd1 mutant plants according to Genty et al. (1989). Five-week-old plants grown in normal light (150 mmol m 22 s 21 ) were measured without further treatment. Data shown are means (6SD) derived from measurements of five different plants per genotype. Black symbols, wild type; white symbols, mgd1. B, qN was calculated using the data shown in A according to Krause and Weis (1991). plexes. This change in the PSII antenna bed is further stabilized by the de-epoxidation of V to Z by the luminal VDE (Horton et al., 2008). VDE is only activated when the pH of the thylakoid lumen falls below 6.8 (Takizawa et al., 2007). Hence, reductions in the capacity of qN could be due to reductions in the amounts of PsbS or VDE, or impairment of the function of VDE because all of these chloroplast components affect qE. It has been shown that MGDG is the major lipid that is directly associated with VDE activity (Yamamoto, 2006) and that VDE is dependent on MGDG for optimum activity in vitro (Latowski et al., 2004). Furthermore, any alteration in the thylakoid membrane's proton conductivity could directly affect qE, if the threshold acidification required for the activation of PsbS and VDE is not reached.

VDE and PsbS Activities Remain Unchanged and Do Not Contribute to Changes of the Xanthophyll Cycle
To assess the different factors possibly contributing to alterations in qN and qE in mgd1, we first checked for changes in the levels of PsbS and VDE by immunoblotting. However, we detected no change in the abundance of either protein in the mutant (Fig. 2). Next, to compare in vivo parameters of the xanthophyll cycle in the mutant and wild type, we measured the kinetics of V de-epoxidation in leaves of plants that had first been adapted to complete darkness for 2 h and which were then exposed to strong light (approximately 1,000 mmol m 22 s 21 ) for up to 60 min. The measured amounts of V, A, and Z were normalized to the total pool size of xanthophyll cycle pigments and then plotted against the duration of exposure to high light (Fig. 5). Following dark adaptation, wild-type and mutant plants both contained high amounts of V, very low amounts of A, and no detectable Z (see 0-min data points; Fig. 5, A and B), due to the inactivation of VDE in darkness. Immediately after the onset of light, the level of V rapidly declined in both genotypes, while the levels of A and Z both increased (Z to a much higher value than A).
In wild-type leaves, the molar proportion of V initially dropped in the first 10 min to less than 40% and then remained constant throughout the remainder of the experiment (Fig. 5A). The proportion of V in mgd1 leaves also declined to approximately 40%, during the first 5 min, but in contrast to wild-type leaves, it progressively increased thereafter, reaching 55% after 60 min (Fig. 5B). For both genotypes, the molar proportion of A increased from 5% to approximately 20% to 25% during the first 10 min but then slightly decreased to approximately 15% to 20% after 60 min (Fig. 5, A and B). The changes in molar proportions of Z appeared to be directly opposite to, and were presumably due to, alterations in the abundance of V; levels of Z initially increased rapidly in both wildtype and mgd1 leaves but then remained at a higher level (approximately 45%-50%) in wild-type leaves, while they declined in mgd1 leaves before reaching a plateau during the latter part of the experiment (Fig.  5B). Thus, prolonged illumination with high light over a period of 60 min had a severe impact on the ratio of V to Z in mgd1 plants. These changes were reflected in differences in de-epoxidation status (DES; 5[A 1 Z]/ [V 1 A 1 Z]) between wild-type and mutant plants (Fig. 5C).
The reduced steady-state capacity of the xanthophyll cycle in mgd1 could be due to either impaired enzyme activity or reductions in the proton motive force (pmf) across the thylakoid membrane. However, the former possibility seems to be highly unlikely because reductions in enzyme activity should have affected both the rapid generation of A and Z during the first 5 min of light stress (Fig. 5) and the steadystate DES. Because no delay in V de-epoxidation was observed in mgd1 plants, a general impairment of VDE activity is unlikely to be the cause of the reduction in Z levels during the latter period of the light stress experiment (Fig. 5). Nevertheless, we examined VDE enzyme activities in wild-type and mgd1 samples in vitro under pH-induced dark conditions. However, as expected, no significant differences in VDE enzyme activities were detected between the wild-type and mutant samples (Supplemental Fig. S1). Because the lower qE and reduced levels of Z under light-stress conditions could not be explained by changes in PsbS or VDE protein levels, or by alterations in VDE activity, we next assessed the possibility that mgd1 causes alterations in the steady-state pmf across the thylakoid membranes.

Reduced Electrochromic Signal Amplitudes in mgd1
Confirm Impaired pmf Formation Across the Thylakoid Membrane To obtain information on the thylakoidal pmf, we measured electrochromic signals (ECS); these are carotenoid absorption shifts with kinetic properties that are proportional to the relative contributions of the electric (DC) and osmotic (DpH) components of the pmf and overall amplitudes that are proportional to the total pmf (Kramer et al., 2003;Cruz et al., 2005;Takizawa et al., 2007). The maximum pmf, determined from the dark-interval relaxation of the ECS after illumination with saturating light, was more than 60% lower in mgd1 leaves than in wild-type leaves (data not shown). Even after normalizing the signal to the Chl contents, a dramatic reduction (.40%) in the maximum pmf relative to wild type was observed ( Fig.  6A; see data points at 2,100 mmol m 22 s 21 actinic light intensity). To ensure that the observed reduction in the ECS amplitude in mgd1 did not result from changes in the optical properties of the leaves, we also measured cytochrome f reduction kinetics during the dark interval (data not shown). Any artefacts resulting from changes in the optical properties of the leaves should have had similar effects on the two signals because they are measured in the same wavelength range. However, after normalization of the cytochrome f difference absorption signals to leaf Chl contents, there were no significant differences; this is in accordance with the quantification of the cytochrome b 6 f complex in isolated thylakoids, which was also found to be unaltered in mgd1 (Table II). Therefore, the observed difference in the ECS amplitude clearly indicates that there is an approximately 40% reduction in pmf across the thylakoid membrane in mgd1, confirming that the mutant's capacity for thylakoid membrane energization is strongly reduced.

Reduced qN Capacity in mgd1 Is Due to Increased Conductivity of the Thylakoid Membranes
Next, we examined the relationship between pmf and actinic light intensity in wild-type and mutant leaves (Fig. 6A). The results showed that the pmf of the mutant leaves was almost as high as in wild-type leaves under light-limited conditions (200 mmol m 22 s 21 light). However, at higher light intensities, when qN becomes more important as a photoprotective mechanism, the increase in pmf was much weaker in mgd1 than in the wild type, resulting in the approximately 40% reduction in maximum pmf.
The reduced pmf under high light intensities indicates that the thylakoid lumen might be considerably less acidic in mgd1 than in wild type, impeding full activation of the VDE and PsbS proteins. This hypothesis could explain the mutant's reduced qN capacity. However, the induction of qN is not dependent on total pmf but only on the DpH component (Cruz et al., 2005;Takizawa et al., 2007). Because the partitioning of pmf into DpH and DC is highly variable between species and individuals (Takizawa et al., 2007) and the amplitude of the ECS reflects both the DC and DpH components of the pmf, we next had to discriminate between the contributions of the two components. This was done by measuring the slow phase (in the range of seconds) of ECS relaxation in darkness, which is attributable to counter-ion movement across the thylakoid membrane (Takizawa et al., 2007). Surprisingly, we found approximately 50:50 partitioning of the pmf between the DpH and DC components, regardless of the actinic light intensity, in both wild-type and mgd1 mutant leaves (Fig. 6B). Therefore, the reduced total pmf amplitude observed in mgd1 must correlate with a similar (.40%) reduction in maximum thylakoid lumen acidification. As described below, this would keep the thylakoid lumen pH above the threshold level required for full activation of PsbS and VDE.
The lumenal pH at maximum acidification is believed to be approximately 5.7 (Takizawa et al., 2007); below this pH, photosynthesis is strongly inhibited due to the loss of electron transfer activities of the cytochrome b 6 f complex, PSII, and PC (Krieger and Weis, 1993;Kramer et al., 1999). The pH of the thylakoid lumen in dark-adapted leaves is believed to be approximately 7.5, i.e. only slightly lower than the stromal pH (Takizawa et al., 2007). Considering these two values and the fact that the maximum total pmf and the DpH component are both reduced by approximately 40% to 50% in mgd1 plants, one can estimate that the lumenal pH in the mutant is in the region of 6.5. Because PsbS and VDE just begin to be activated when the lumen pH falls to around this value and full qE is only established when the pH falls below 6.0 (Takizawa et al., 2007), one can deduce that the reduced qN capacity of mgd1 is entirely explainable in terms of increased conductivity of the thylakoid membranes in saturating light.
Alteration of the steady-state energization of the thylakoid membrane may also explain the declines in Z content and DES observed in mgd1 after 5 min of high-light stress (Fig. 5). Immediately after illumination of dark-adapted leaves, the pmf is dramatically higher than under steady-state conditions because both the Calvin cycle (which consumes ATP) and the ATP synthase are still inactive, and their reductive activation by thioredoxin takes a few minutes in the light; a lower steady-state pmf is established only after ATP synthase activation because part of the pmf is then used for ATP synthesis. Therefore, in both wild-type and mgd1 mutant leaves, pmf is highest during the activation phase of photosynthesis and then subsequently declines to steady-state levels within 10 min following the onset of illumination. Hence, the pmf in the mgd1 mutant is only high enough to fully activate VDE during the first few minutes of illumination; later, under steady-state conditions, the pmf in mgd1 is below the optimal level for VDE, and so de-epoxidation rates are retarded and the equilibrium between Z and V starts to shift slightly toward V. In wild-type leaves, lower thylakoidal proton conductivity means that under 1,000 mmol m 22 s 21 actinic light (i.e. conditions used in Fig. 5), the steady-state pmf stays at a much higher level than in mgd1 (Fig. 6), and so the lumen retains sufficient acidity to maintain high VDE activity.
As yet, we can only speculate about possible reasons why the maximum pmf is reduced in mgd1. There are no apparent differences in pmf between wild-type and mutant plants under low-light intensities (up to 200 mmol m 22 s 21 ; Fig. 6A). This seems to preclude the possibility that the changes in the mutant are due to constitutive increases in the conductivity of the thylakoid membranes caused by alterations in levels of MGDG, because such changes would be expected to alter pmf under limiting light intensities as well as at high light intensities. Instead, structural changes of the thylakoids in response to increasing membrane energization may increase the conductivity of the mutant's thylakoid membranes to protons. Interestingly, the increased conductivity becomes apparent under conditions that initiate qE, which is known to involve major structural rearrangements of the PSII-LHCII supercomplexes (Horton et al., 2008). Therefore, MGDG may play an important role during this thylakoid membrane reorganization.

CONCLUSION
The availability of the mgd1 mutant enabled us to analyze the involvement of MGDG in various processes in vivo. The lack of a measurable effect of the mgd1 mutation on various chloroplast protein targeting pathways suggested that the residual MGDG in mgd1 is sufficient to maintain efficient protein transport (Fig. 1). Thus, while the possibility that MGDG is directly involved in chloroplast protein traffic at some level cannot be excluded at this stage, it is possible that the importance of the lipid in these processes has been somewhat overestimated. The low-temperature fluorescence measurements on dark-grown plants revealed the activation of a putative photoprotective mechanism in mgd1 (Fig. 3, A and B), suggesting that the mutation affects the plant's ability to deal with high light intensities. In accordance with this notion, we confirmed that MGDG deficiency has a negative effect on photosynthetic efficiency at elevated light intensities (Figs. 3C and 4). This is due to a reduction in qN capacity that we ascribe to increased conductivity of the thylakoid membranes, which reduces the maximum pmf in mgd1 by approx- Figure 6. Thylakoid membrane energization is strongly reduced in the mgd1 mutant at high light intensities. A, Light response curve of the total ECS, which is proportional to the steady-state pmf across the thylakoid membrane. The pmf across the thylakoid membrane during steady-state photosynthesis was estimated from the amplitude of the ECS during short intervals (15 s) of dark relaxation. Leaves of 5-weekold plants grown in normal light (150 mmol m 22 s 21 ) were preilluminated at each light intensity for 10 min to allow photosynthesis to reach steady state, and the total ECS was normalized to the Chl content of the leaf segment being measured. B, Contribution of the DpH to the total pmf across the thylakoid membrane. Estimates of the partitioning of pmf between its DpH and DC components were derived from measurements obtained during the slowly relaxing phase of the ECS. Data shown in A and B are means (6SD) of measurements of six different plants per genotype. Black symbols, wild type; white symbols, mgd1.
imately 40%. Consequently, the threshold pH for initiating qE is barely reached, and neither PsbS nor VDE becomes fully activated in the mutant. Impairment of thylakoid membrane energization under high-light conditions may be due to structural rearrangements that occur during the onset of qE. Consideration of our results alongside those obtained using the DGDGdeficient mutant dgd1 clearly indicates that MGDG and DGDG do not simply provide a membrane environment to physically support the photosynthetic complexes. Rather, these two galactolipids contribute directly to various photosynthesis-related processes with individual contributions that are quite distinct.

Plant Material and Growth Conditions
Arabidopsis (Arabidopsis thaliana) seeds (wild type and mgd1-1; both Columbia-0 ecotype) were surface-sterilized, sown on petri plates containing Murashige-Skoog medium, and cold-treated at 4°C (Aronsson and Jarvis, 2002). Thereafter, the seeds were transferred to a growth chamber; after 14 d, seedlings were transplanted to soil. The light regime was 16 h light/8 h dark with a light intensity of approximately 100 to 200 mmol m 22 s 21 , unless stated otherwise. The mgd1-1 mutant used throughout this study has been described previously (Jarvis et al., 2000). Together, DNA gel-blot analysis (Supplemental Fig. S2) and genetic, cosegregation analysis (Supplemental Table S1) demonstrated conclusively that there is only a single T-DNA insertion in the mutant and that this insertion is tightly linked to the mutant phenotype.
For the analysis of pigment-protein complexes (Fig. 3), plants were grown on soil either in complete darkness for 5 to 8 d or in the light regime given above for 28 d. Plants used for all other measurements were 35 to 42 d old. Plants used for xanthophyll cycle measurements were kept in darkness for 2 h before applying high-light stress. Light stress was performed for 0, 2, 5, 10, 30, or 60 min at 1,000 to 1,100 mmol m 22 s 21 .

Chloroplast Isolation and Protein Import
Chloroplasts from 14-d-old plants were isolated according to Aronsson and Jarvis (2002). Template DNA for the Arabidopsis precursors pSS, pL11, and pCF 0 II, as well as for atToc33 and atToc34, were amplified by PCR from cDNA clones using M13 primers. Coupled transcription/translation was performed using either a wheat (Triticum aestivum) germ extract system or a TNT T7 Quick for PCR DNA system (Promega) containing [ 35 S]Met and T7 RNA polymerase, according to the manufacturer's instructions (Promega). Import reactions were conducted as described (Aronsson and Jarvis, 2002). Briefly, each 150-mL import assay contained 10 7 chloroplasts, 5 mM MgATP, 10 mM Met, and translation mixture not exceeding 10% of the total volume. Import was carried out in white light (100 mmol m 22 s 21 ) at 25°C and was stopped by adding ice-cold HEPES-sorbitol buffer containing 50 mM EDTA. Samples were resolved, fixed, and visualized by fluorography. Quantification employed ImageQuant software (Molecular Dynamics).
For alkaline treatment, import reactions were scaled up three times (3 3 10 7 plastids) and conducted for 30 min prior to treatment with Na 2 CO 3 , pH 11.5. Thereafter, envelope membranes were enriched according to Baldwin et al. (2005).
For the thylakoid assays, a 30-min import reaction using 3 3 10 7 plastids was performed, followed by treatment with thermolysin (final concentration 0.1 mg mL 21 ) for 30 min at 4°C. Thylakoids were then isolated as described by Rawyler et al. (1992), except that UDP and p-hydroxymercuribenzoic acid were omitted from the buffers and MOPS was replaced with HEPES.

Protein Analysis
Concentrations of chloroplast (Fig. 2) and total protein (Fig. 3A, inset; isolated from 5-d-old dark-grown seedlings according to Kovacheva et al. [2005]) samples were determined using the D C -Protein Assay (Bio-Rad Laboratories). For each sample, an equal amount of protein (10 mg) was separated on SDS-PAGE (Laemmli, 1970) and blotted to nitrocellulose membrane.

Fluorescence Spectroscopy
Fluorescence emission spectra from cotyledons of dark-grown plants or leaves from light-grown plants were measured at 77 K using a Fluorolog-3 spectrofluorometer (Spex Instruments S.A.). The emission was measured as photon emission per unit interval of wavelength. Emission spectra were recorded with an integration time of 0.5 s and the excitation wavelength set to 440 nm, as indicated in the figures. Both excitation and emission monochromators were used with a slit width of 3 nm. The spectra were corrected for the spectral sensitivity of the photomultiplier. Samples were inserted in cylindrical glass cuvettes and stored in liquid nitrogen prior to measurements. Small parts of the hypocotyl accompanied the cotyledons, but the fluorescence from the hypocotyl is negligible compared to the fluorescence from the cotyledons (data not shown). All spectra were smoothed 10 times using a fixed-bandwidth, sharp-cutoff, three-point, low-pass linear digital filter. The spectra shown are averages of five to 10 spectra.
The degree of phototransformation of Pchlide to Chlide in Arabidopsis seedlings was determined after irradiating the samples with three flashes and extracting the pigments into acetone to measure the fluorescence spectra at room temperature. The amount of Chlide found was used to calculate the amount of Pchlide before the flash using a ratio of 1.09 for the fluorescence yield of Pchlide:Chlide. The 1.09 ratio was determined using a PLB preparation (Ryberg and Sundqvist, 1982) measured in darkness, after irradiation and extraction into a 80% (v/v) acetone solution.

Chl Fluorescence Measurements
Chl fluorescence was recorded with a pulse-amplitude modulation fluorometer (Dual-PAM; Heinz Walz). Plants were dark-adapted for 1 h, and fluorescence light response curves were recorded after a 5-min exposure to the photosynthetically active radiation as indicated. PSII quantum yield and linear electron flux were calculated (Genty et al., 1989).

Difference Absorption Spectroscopy
The PSII, cytochrome b 6 f complex, and PSI contents were determined in thylakoids, isolated according to Schö ttler et al. (2004) as follows. PSI was quantified from measurements of P 700 difference absorption signals at 830-to 870-nm wavelengths obtained from solubilized thylakoids, according to Schöttler et al. (2007), using a Dual-PAM Chl fluorometer (Heinz Walz). PSII and the cytochrome b 6 f complex were determined from difference absorption measurements of cytochromes b 559 (PSII) and cytochromes f and b 6 . For these measurements, thylakoids equivalent to 50 mg Chl mL 21 were destacked in a low-salt medium to improve the optical properties of the samples (Kirchhoff et al., 2002). All cytochromes were oxidized by application of 1 mM Na 3 Fe(CN) 6 . Subsequent addition of 10 mM sodium ascorbate resulted in the reduction of cytochrome f and the high-potential form of cytochrome b 559 , while cytochrome b 6 and the low-potential form of cytochrome b 559 were only reduced upon addition of dithionite. At each redox potential, absorption spectra between 540 and 575 nm were determined using a V-550 spectrophotometer with a head-on photomultiplier (Jasco) in which the monochromator slit width was set to 1 nm. The acquired difference absorption spectra were deconvoluted using reference spectra and difference absorption coefficients as described by Kirchhoff et al. (2002). The PSII contents were calculated from the sum of the difference absorption signals arising from the low and high potential forms of cytochrome b 559 .
The relative stoichiometries of PC per P 700 were determined using a PC version of the Dual-PAM spectroscope (Dual-PAM-S; Heinz Walz; Schö ttler et al., 2007). Measurements were performed on intact leaves prior to thylakoid isolation because some PCs are released from the thylakoid lumen during the isolation, so data obtained on isolated thylakoids are not fully quantitative. The relative PC contents were multiplied by the P 700 contents determined in thylakoids to obtain estimates of absolute PC contents.
The ECS peak at 517 nm was used as an in vivo measure of the proton motive force across the thylakoid membrane (Kramer et al., 2003;Takizawa et al., 2007). The difference absorption signal was measured using a KLAS-100 LED-array spectrophotometer (Heinz Walz), allowing the simultaneous measurement of light-induced difference absorption signals at six pairs of wavelengths in the visible range of the spectrum between 500 and 570 nm. The ECS was deconvoluted from signals arising from Z, scattering effects, the C550signal, and redox changes of the cytochromes, essentially as described by Klughammer et al. (1990), then normalized to the Chl content of the leaf disc. Leaves were illuminated for 10 min prior to each measurement to allow photosynthesis to reach steady state. The maximum amplitude of the ECS was determined after illuminating the leaves with saturating light (2,100 mmol m 22 s 21 ). Saturating illumination was interrupted by short intervals of darkness (15 s), and the dark-interval relaxation kinetics of the ECS and cytochrome f reduction were measured. The 15-s dark intervals were sufficient for complete relaxation of the pmf. To obtain light saturation curves of the pmf, dark-interval relaxation kinetics after illumination at subsaturating light intensities were measured. Pmf partitioning into DpH and the electrochemical component (DC) was resolved by analyzing the slowly relaxing phase of the ECS as described by Takizawa et al. (2007).

Photosynthetic Pigment Measurements
Leaf samples were homogenized in liquid nitrogen, and then pigments were extracted at 4°C under low-light conditions using, first, 700 mL of 80% (v/v) acetone, and second, 700 mL of 100% acetone. The two supernatants were combined, and then a 20-mL sample was used immediately for HPLC analysis according to Thayer and Björkman (1990) and Gilmore and Yamamoto (1991), with some alterations. Pigments were separated on a C18 reversephase column (Nucleosil 120A C18, 3 mm, 250 3 3 mm, with integrated precolumn; Knauer) using the following solvents: solvent A, acetonitril:methanol: 0.1 M Tris-HCl, pH 8.0 (72:8:3); solvent B, methanol:ethylacetate (68:32). The gradient used was: 0 to 25 min 100% A / 100% B and 25 to 32 min 100% B, followed by 10 min equilibration in 100% solvent A prior to the next measurement, all at a flow rate of 0.4 mL min 21 . Detection was carried out at 440 nm, and peaks were identified by comparing their UV spectra and their retention times with commercially available standards.
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number AAF65066.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Measurements of VDE activity in wild-type and mgd1 thylakoids.
Supplemental Figure S2. DNA gel blot indicating that mgd1 contains a single T-DNA insertion.
Supplemental Table S1. Cosegregation of the mgd1 phenotype and the T-DNA insertion.