Plastoglobular protein 18 is involved in chloroplast function and thylakoid formation

The plastoglobular protein PG18 plays an important role in thylakoid formation and its loss has a significant impact on photosynthesis. This emphasizes the general importance of plastoglobules in chloroplast biogenesis.

Abstract

Plastoglobules are lipoprotein particles that are found in different types of plastids. They contain a very specific and specialized set of lipids and proteins. Plastoglobules are highly dynamic in size and shape, and are therefore thought to participate in adaptation processes during either abiotic or biotic stresses or transitions between developmental stages. They are suggested to function in thylakoid biogenesis, isoprenoid metabolism, and chlorophyll degradation. While several plastoglobular proteins contain identifiable domains, others provide no structural clues to their function. In this study, we investigate the role of plastoglobular protein 18 (PG18), which is conserved from cyanobacteria to higher plants. Analysis of a PG18 loss-of-function mutant in Arabidopsis thaliana demonstrated that PG18 plays an important role in thylakoid formation; the loss of PG18 results in impaired accumulation, assembly, and function of thylakoid membrane complexes. Interestingly, the mutant accumulated less chlorophyll and carotenoids, whereas xanthophyll cycle pigments were increased. Accumulation of photosynthetic complexes is similarly affected in both a Synechocystis and an Arabidopsis PG18 mutant. However, the ultrastructure of cyanobacterial thylakoids is not compromised by the lack of PG18, probably due to its less complex architecture.

Introduction

Plastids of higher plants, mosses, and algae contain lipoprotein particles known as plastoglobules (PGs) (Lichtenthaler, 1968). PGs are surrounded by a lipid monolayer membrane, are found in both non-photosynthetic and photosynthetic tissues, and are associated with the thylakoids in chloroplasts. The particles are filled with hydrophobic molecules, and a number of different proteins have been identified which are associated with the lipid monolayer (van Wijk and Kessler, 2017). Strikingly, PGs are of a highly dynamic nature, and their numbers and sizes have been shown to increase considerably in different developmental stages. For instance, they were found to accumulate in etiolated tissues, decrease upon illumination and greening, and they appear to accumulate again during senescence where they play an important role in thylakoid breakdown (Tevini et al., 1977; Tevini and Steinmuller, 1985). Moreover, exposure to nitrogen starvation, drought, or light stress stimulates formation and growth of PGs (Eymery, 1999; Gaude et al., 2007; Zhang et al., 2010; Li, 2015).

In contrast to the thylakoid membrane, PGs are composed of neutral lipids (triacylglycerol, phytol esters, and free fatty acids), prenylquinones (α-tocopherol, plastoquinone, plastochromanol, and phylloquinone), as well as carotenoids—the last category being especially prevalent in the PGs of chromoplasts (van Wijk and Kessler, 2017). PGs have been purified from maize and Arabidopsis, and subjected to MS (Lundquist et al., 2012b; Huang et al., 2013). All of the ~30 proteins identified to date seem to be very specific and are almost exclusively localized to PGs; only a few are thought to be recruited to PGs under particular conditions, for example enzymes involved in jasmonate biosynthesis or chlorophyll degradation (Lippold et al., 2012; Lundquist et al., 2013). The most abundant PG proteins belong to the family of plastid-specific fibrillins (FBNs), which represent ~53% of the total protein content and are suggested to function as structural components (Grennan, 2008). Another large portion (~19%) of the PG proteome is composed of members of the ‘activity of BC1 complex kinase’ (ABC1K) family, which may phosphorylate other PG proteins and thus regulate their activity (Lundquist et al., 2012a). Apart from these abundant protein classes, only a few other PG proteins have been investigated in detail. One of these, the enzyme tocopherol cyclase [vitamin E deficient 1 (VTE1)], functions in the formation of tocopherol and plastochromanol-8 (Vidi et al., 2006). Moreover, the phytol ester synthases 1 and 2 (PES 1 and 2), which play a role in the chlorophyll degradation pathway, have been characterized from PGs (Lippold et al., 2012). Another metabolic enzyme is the NADP(H) dehydrogenase C1 (NDC1), which acts as a reductase in the phylloquinone biosynthetic pathway (Eugeni Piller et al., 2011; Fatihi et al., 2015). Most of the remaining PG core proteins have not yet been studied at the functional level. Interestingly, none of these proteins possesses typical transmembrane domains, as would be required for insertion into a lipid bilayer. It is therefore assumed that PG proteins insert short hydrophobic domains into the lipid monolayer or associate with the membrane via protein–protein interactions.

However, PGs may not be exclusive to higher plants. In fact, the Chlamydomonas reinhardtii eyespot and PGs resemble each other in appearance and protein composition (Kreimer, 2009). A number of common homologous proteins have been identified, such as proteins with plastoglobulin domains and ABC1 kinases (Schmidt et al., 2006; Lundquist et al., 2012a; Eitzinger et al., 2015). Moreover, lipid-rich droplets whose metabolite and small molecule composition is similar to that of PGs have also been observed in several cyanobacteria (Stanier et al., 1988; Peramuna and Summers, 2014). Furthermore, homologs of plant FBNs have been found in Synechocystis sp. PCC 6803 (hereafter Synechocystis), and loss-of-function mutants are characterized by light sensitivity and accumulation of photoprotective pigments (Cunningham et al., 2010). This suggests that PGs in Arabidopsis and lipid droplets in cyanobacteria may be evolutionarily related and could share conserved roles.

In this study, we investigated the role of a plastoglobular protein, which we termed pastoglobular protein 18 (PG18; AT4G13200), due to its molecular weight. Initially identified as UNKNOWN1 in Lundquist et al., 2012b, PG18 is an as yet uncharacterized member of the core proteome of PGs. PG18 has no predicted functional domains and is one of the less abundant PG proteins (Lundquist et al., 2012b). Nevertheless, the loss-of-function mutant in Arabidopsis shows a distinctly pale green phenotype and a defect in the formation of thylakoid membranes, as well as in the accumulation of thylakoid membrane protein complexes. Additionally, functional loss of the homolog of PG18 in cyanobacteria has comparable effects, supporting the idea that its role in thylakoid biogenesis has been conserved in higher plants.

Materials and methods

Plant material and growth conditions

The T-DNA insertion line GK-439D01 (PG18 accession number: At4g13200) was obtained from the GABI-Kat collection, and homozygous mutants were screened with oligonucleotides given in Supplementary Table S1 at JXB online. For complementation the coding sequence of PG18 was cloned into pK7FWG2 under the control of the 35S promoter (Karimi et al., 2002). The construct was introduced into Agrobacterium tumefaciens strain GV3101, and pg18 mutants were transformed by floral dip (Clough and Bent, 1998).

Unless indicated otherwise, Arabidopsis (Arabidopsis thaliana) wild-type (WT) Columbia ecotype and the mutants were grown on soil under normal light conditions (NL, 16/8 h light/dark, 21 °C and 120 µmol photons m⁻² s⁻¹). For in vivo labeling experiments, plants were grown on half-strength Murashige and Skoog (MS) medium under the same light regimes as above for 21 d. Pea plants (Pisum sativum L., cv. ‘Arvica’, Prague, Czech Republic) were grown on sand in a climate chamber under a 16 h light (220 μmol photons m⁻² s⁻¹) and 8 h dark regime at 21 °C.

Synechocystis strains, growth conditions, and analysis

WT and mutant Synechocystis cells were grown on solid or in liquid BG 11 medium (Rippka, 1979) supplemented with 5 mM glucose (unless indicated otherwise) at 30 °C under continuous illumination at a photon irradiance of 30 μmol photons m⁻² s⁻¹ of white light. Doubling times were determined after 2 d and 4 d of photoheterotrophic and photoautotrophic growth, respectively.

To generate the insertion mutant synpg18⁻, fragments flanking the synpg18 gene (sll1769) were amplified from WT genomic DNA and the resulting fragments synpg18-up and synpg18-down were cloned into the pJET1.2 vector (Thermo Scientific). The synpg18-down fragment was inserted into pJET-synpg18-up via SalI and EcoRI, and a kanamycin resistance cassette was subsequently inserted between the upstream and downstream fragments via BamHI. Oligonucleotides are given in Supplementary Table S1. Synechocystis WT cells were transformed as described, and segregation was achieved by growth on increasing concentrations of kanamycin and confirmed by PCR (Eaton-Rye, 2004). Isolation of whole-cell proteins was performed as described (Rengstl et al., 2011). Chlorophyll contents of Synechocystis WT and mutant cells were measured according to Wellburn and Lichtenthaler (1984), and F_v/F_m values were determined with a FluorCam 800 MF (Photon System Instruments, Drasov, Czech Republic).

Analysis of chloroplast and cyanobacteria ultrastructure

Plant leaves were cut into small pieces (≤1×1×1 mm) in 75 mM cacodylate buffer containing 2 mM MgCl₂ and 2.5% glutaraldehyde. After over-/normal pressure infiltration with fixation buffer and storage of the samples overnight at 4 °C, the samples were post-fixed with 1% osmium tetroxide for 1 h. This step was followed by dehydration in a graded acetone series: samples were successively incubated in 10% acetone for 15 min, 20% acetone supplemented with 1% uranyl acetate for 30 min, and in 40, 60, and 80% acetone for 20 min each. Finally, the samples were put into 100% acetone at least twice (for 5 min, then overnight). Afterwards, the plant tissue was infiltrated with Spurr’s resin and polymerized at 63 °C for at least 16 h. Prior to fixation, Synechocystis cells were concentrated via gentle centrifugation. Chemical fixation was also carried out by resuspending the resulting pellet with 2.5% glutaraldehyde in 75 mM cacodylate buffer containing 2 mM MgCl₂ and keeping the samples at 4 °C overnight. The post-fixation with 1% osmium tetroxide and the following dehydration, infiltration, and embedding were carried out as mentioned above for Arabidopsis.

After thin sectioning, the material was examined on a Zeiss EM 912 with an integrated OMEGA filter for TEM. The acceleration voltage was set to 80 kV and the microscope was operated in the zero-loss mode. Images were acquired using a 2k×2k slow-scan CCD camera (TRS Tröndle Restlichtverstärkersysteme, Moorenweis, Germany).

SDS–PAGE and immunoblotting

Total protein extraction as well as thylakoid isolation were performed as described (Schwenkert et al., 2006; Patil et al., 2018). For PG18 extraction, thylakoids were resuspended in lysis buffer (5 mM HEPES-KOH, pH 8.5 mM EDTA) and incubated on ice for 30 min. Sonicated thylakoids were mixed with salt solutions [final concentration of 1 M NaCl, 100 mM Na₂CO₃, 3 mM urea, or 1% lithium dodecylsulfate (LDS), respectively] and incubated for 30 min on ice. The samples were centrifuged for 5 min at 6000 g and 4 °C to separate the soluble and membrane fraction of the thylakoids. SDS–PAGE and immunoblotting were performed as described (Schwenkert et al., 2006). For PG18 antisera production, full-length purified PG18 was injected into rabbits (Biogenes). Antisera against Arabidopsis proteins were purchased as indicated (see Supplementary Table S2). ATP synthase and D1 antisera were provided by S. Greiner, TIC110 and FBPase antisera were from B. Bölter, and PsaG and OE33 antisera were from Jörg Meurer.

Analysis of protein extracts by Blue Native (BN)–PAGE

BN–PAGE was performed as described previously (Schwenkert et al., 2006). Thylakoid membranes equivalent to 100 µg of protein were separated on a 6–15% acrylamide gradient. Lanes were excised, denatured, and subjected to SDS–PAGE in the second dimension. Gels were silver stained according to Blum et al. (1987). In vivo labeling was performed according to Meurer et al. (1998). Twenty-one-day-old plants of the WT and mutant were incubated for 20 min and isolated proteins were subjected to BN–PAGE as described above. Proteins were detected by autoradiography.

Agrobacterium-mediated transient expression of fluorescent proteins in tobacco

The Agrobacterium tumefaciens strain AGL1 was transformed with PG18 (At4g13200) in the vector pK7FWG2 (Karimi et al., 2002) and used to infiltrate 4- to 6-week-old Nicotiana benthamiana leaves as described by Schweiger et al. (2012). Fluorescence was observed with a confocal laser scanning microscope at 20 °C (Leica TCS SP5).

Isolation of PGs

Intact chloroplasts from 3-week-old peas were isolated according to Waegemann et al. (1992). The chloroplasts were then separated in a discontinuous sucrose gradient according to Vidi et al. (2006). The gradient was fractionated by taking 1 ml samples, and the proteins in each fraction were precipitated with trichloroacetic acid.

Chl a fluorescence measurements and PSI activity

The kinetics of induction of Chl a fluorescence in WT and mutant leaves were measured using a pulse-modulated fluorometer (Imaging PAM and DUAL-PAM100; Walz). Leaves, dark adapted for at least 15 min, were used to analyze minimal (F₀) and maximal (F_m) fluorescence yields, the latter being determined by application of a saturating light pulse (1 s duration, 1000 μmol photons m⁻² s⁻¹, 4 min illumination between each pulse). The potential maximum quantum yield of PSII was measured as (F_m−F₀)/F_m=F_v/F_m (Schreiber et al., 1988).

PSI yield in leaves was measured as absorption changes at 820 nm induced by saturating pulses and far-red light (12 W m⁻² as measured with a YSI Kettering model 65 A radiometer) in the absence or presence of actinic light as indicated using the DUAL-PAM100 (Klughammer, 1994). The size of the intersystemic plastoquinone pool was calculated as the ratio of the areas induced by a single and multiple turnover flashes causing a single charge separation and the reduction of the entire plastoquinone pool, respectively (Schreiber, 1988) Other parameters were calculated using the algorithms provided in the DUAL-PAM100 software (Walz).

Pigment analysis

For pigment analyses, leaf samples were frozen in liquid N₂ and either used directly for pigment extraction or stored at –80 °C for up to 2 weeks until further use. Pigments were extracted by grinding frozen leaf material in a mortar after addition of 1 ml of 100% acetone. After a short centrifugation, the supernatant was filtered through a 0.2 μm membrane filter (GE Healthcare, Little Chalfont, Buckinghamshire, UK) and then subjected to HPLC analysis. Separation and quantification of pigments were done by reversed-phase chromatography as described in Färber et al. (1997).

Metabolite analysis

Lipids and non-polar metabolites were extracted as described in Hummel et al. (2011). For LC-MS analysis, the Dionex Ultimate 3000 UHPLC (Thermo Fisher Scientific) in combination with a timsTOF (Bruker Daltonik) was used. The dry extract of 50 mg of material was resolved in acetonitrile:isopropanol (7:3) and injected on a C8 reversed phase column (Ultra C8 100×2.1 mm; Restek) with 300 µl min⁻¹ flow at 60 °C. The solvents used are (A) water and (B) acetonitrile:isopropanol (7:3), both including 1% (v/v) ammonium acetate and 0.1% (v/v) acetic acid. The 26 min gradient started at 55% B, followed by a ramp to 99% B within 15 min. After a 5 min washing step at 99% B, the gradient was returned to 55% B and kept constant for 5 min equilibration.

For MS detection, an electrospray ionization (ESI) source was used in positive mode. Nitrogen was the dry gas, at 8 l min⁻¹, 8 bar, and 200 °C. The timsTOF mass spectra were recorded in MS mode from 50 m/z to 1300 m/z with 40 000 resolution, 1 Hz scan speed, and 0.3 ppm mass accuracy. Compounds were annotated in a targeted approach using the specific mass (m/z) at retention time and the isotopic pattern. All data were acquired by otofControl 4.0. The evaluation was performed by DataAnalysis 5.1, ProfileAnalysis 2.3, and MetaboScape 1.0. Corticosterone (0.2 mg ml^–1) was used as internal standard (IS) for normalization. Given values are relative ratios of the intensities×intensity of IS ⁻¹×g FW⁻¹.

Computational analyses

Sequences for PG18 from Arabidopsis and Synechocystis were obtained from TAIR (https://www.arabidopsis.org) and CyanoBase (http://genome.microbedb.jp/cyanobase), respectively. Homologs of PG18 from other species were collected from NCBI/BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and Phytozome (https://phytozome.jgi.doe.gov/pz/portal.html). Phylogenetic trees were generated by using the CLC Main Workbench software (CLC bio, Aarhus, Denmark). Alignments were generated by using the algorithm provided by CLC Main Workbench (developed by QIAGEN Aarhus).

Results

PG18 is localized to PGs

To investigate the subcellular localization of PG18 experimentally, we transiently expressed a PG18–green fluorescent protein (GFP) fusion protein under the control of the 35S promoter in tobacco leaves. Leaves were infiltrated with Agrobacterium containing the construct, and protoplasts were isolated 2 d after the transfection. GFP fluorescence, as well as chlorophyll autofluorescence, was detected in isolated protoplasts with a confocal laser-scanning microscope. The GFP signal was found exclusively in chloroplasts, where it appeared to form punctate structures (Fig. 1A).

Fig. 1.

PG18 is a membrane-associated protein in PGs. (A) Tobacco leaves were infiltrated with Agrobacterium containing a vector expressing PG18 fused to GFP (see the Materials and methods). Protoplasts were prepared from the infiltrated leaves and visualized with confocal microscopy. (B) A total protein extract from Arabidopsis was separated into a membrane (M) and a soluble fraction (S). PG18 was detected on immunoblots using specific antibodies. Antibodies raised against TIC110 (a membrane protein marker) and FBPase (a soluble protein marker) were used as controls. (C) Disrupted Arabidopsis thylakoids were extracted with salt-containing buffers and separated into supernatant (S) and pellet (P) fractions by centrifugation. The samples were fractionated by SDS–PAGE and analyzed immunologically using antibodies specific for PG18, Cyt f, and OE33, respectively. As a control for the specificity of the anti-PG18 antibody, isolated chloroplasts from the mutant line pg18 were also loaded onto the gel. (D) Chloroplast membranes were isolated from 3-week-old pea plants. The samples were then further fractionated on a discontinuous sucrose gradient (the number of the fraction is indicated above each lane). Immunoblotting was performed and the blot was probed for the presence of the marker proteins for thylakoids (LHCA1) and plastoglobuli (PGL35). (E) Arabidopsis thylakoids were isolated from plants grown on soil for 3 weeks. Trypsin digestion of thylakoids was performed on ice for 10 min and the samples were analyzed after immunoblotting. An antiserum directed against the ATP synthase α-subunit was used to ensure that the stromal side of the thylakoid was accessible to the enzyme, whereas the lumen protein OE33 served as a marker for the integrity of the thylakoids. PGL35 was used as a PG control.

To determine whether PG18 is a soluble protein or is associated with any of the chloroplast’s membranes, total proteins isolated from leaves were separated into a membrane and a soluble fraction. PG18 was found exclusively in the membrane fraction by immunoblot analysis (Fig. 1B). We then treated the membrane fraction with 1 M NaCl, 0.1 M Na₂CO₃, and 3 M urea to ascertain whether PG18 is attached to the membrane by hydrophobic or electrostatic interactions and could therefore be removed by any of these agents (Fig. 1C). Incubation with buffer served as a negative control, and disruption of the membrane with 1% LDS was applied to achieve total solubilization of the membrane. Only the treatment with 1% LDS resulted in complete solubilization of PG18. However, a small amount of PG18 was found in the supernatant after treatment with 3 M urea (Fig. 1C). To assess the efficiency of these treatments, the blot was probed with OE33 antiserum. OE33 is a peripheral membrane protein in the thylakoid lumen and is at least partially extracted by each of the agents mentioned above, as they all damage the integrity of the thylakoid membrane (Fig. 1C) (Bhuiyan et al., 2015). Cytochrome f (Cyt f) was used as an example of an integral membrane protein and, as expected, it is solubilized only by treatment with 1% LDS (Fig. 1C). In Fig. 1C, pg18 protein extract of was loaded, showing specificity of the antiserum. These results were surprising, as they imply that PG18 behaves like an integral membrane protein, although its sequence appears not to contain any classical transmembrane domains. Several proteomics studies have identified PG18 in PGs (Friso et al., 2004; Peltier et al., 2006; Ytterberg et al., 2006) and we therefore set out to test this further. To this end, we fractionated pea chloroplasts into PGs and thylakoid membranes. Pea plants were chosen instead of Arabidopsis for this experiment because they provide larger amounts of leaf material. The specific reactivity of the polyclonal antibody raised against P. sativum PG18 has already been established (data not shown). Thylakoid membranes and PGs were separated on a sucrose density gradient, and selected fractions were subjected to SDS–PAGE and subsequently probed with antisera against LHCA1 (chlorophyll a-b binding protein 1) as a thylakoid marker, plastoglobulin 35 (PGL35) as a PG marker, and PG18 (Fig. 1D). As revealed by the distribution of the PG marker protein PGL35, the PG fraction was not contaminated with thylakoids, and the majority of PG18 was also detected in this fraction. Since PGs are known to be associated with the thylakoid membrane (Austin et al., 2006), it is not surprising that small amounts of both PG18 and PGL35 are detected in the thylakoid fraction. Additionally, in order to demonstrate that PG18 is located at the surface of the PGs, we treated intact thylakoids and their associated PGs with trypsin (Fig. 1E). As expected, PG18 was completely digested—as was the α-subunit of the ATP synthase, which is exposed on the stromal side of the thylakoids—while the luminal OE33 protein used as a control for thylakoid integrity remained intact. Taken together, the observed punctate structures of PG18–GFP, the subfractionation data, and the fact that PG18 has been assigned to PGs by MS analyses (Ytterberg et al., 2006; Lundquist et al., 2012b) led us to the conclusion that PG18 is indeed a PG-localized protein. Considering that none of the PG proteins identified so far harbors transmembrane domains, but are nevertheless most likely to be integrated into the lipid monolayer surface of PGs, the observation that PG18 behaves similarly to an integral membrane protein upon treatment with various chaotropic agents is compatible with the protein’s localization to PGs.

Arabidopsis pg18 mutants display a pale green phenotype, and alterations in thylakoid membrane structure and lipid content

To investigate the function of PG18, we isolated a homozygous T-DNA insertion line. The mutant was viable on soil, but displayed a pale green phenotype under NL conditions (21 °C, 120 µmol photons m⁻² s⁻¹ light intensity, 16/8 h light/dark). Since it has been observed that PGs become more abundant under light stress (Zhang et al., 2010), we also exposed mutant plants to increased light intensities (IL; 250 µmol photons m⁻² s⁻¹), namely 17.5 d growth under NL conditions followed by 3.5 d under IL conditions (Fig. 2A). The position of the T-DNA insertion as well as genotyping PCR is shown in Fig. 2B. To ensure that the phenotype correlated with the disruption of the PG18 gene, we complemented the mutant with a PG18–GFP fusion construct expressed under the control of a 35S promoter. A representative example of a complemented line is shown in Fig. 2A (right panel). The knock out of PG18 was verified at the protein level by immunoblot analysis (Fig. 2C). A specific band at 18 kDa was detected in the WT that was absent in the mutant. An antiserum against the chloroplast protein TIC110 was used as a loading control. In addition, proteins were extracted from the complemented mutant line and likewise probed with PG18 antiserum. A protein of 45 kDa was detected, corresponding to the expected size of PG18–GFP.

Fig. 2.

PG18 is required for normal development of A. thaliana. (A) WT, homozygous mutant (pg18), and complemented mutant plants (35S::PG18-GFP) were grown in long-day conditions for 3 weeks (normal light, NL) and for 17.5 d under 120 µmol photons m⁻² s⁻¹ followed by 3.5 d under 250 µmol photons m⁻² s⁻¹ (increased light, IL). (B) The position of the T-DNA in the first exon of PG18 and oligonucleotides used for genotyping by PCR are indicated. (C) Total protein extracts from WT and pg18 mutant plants were gel fractionated and immunoblots were prepared. The PG18 protein was detected with specific antibodies. TIC110, a protein found in the inner envelope of the chloroplast, was used as a loading control. The immunolabeled PG18–GFP is highlighted with an asterisk. (D) Total protein was extracted from plants grown under NL and IL conditions, and subjected to immunoblot analysis. The PG18 protein was detected using the specific antibodies mentioned above. The α-subunit of the chloroplast ATP synthase was used as a loading control. Antibodies against PGL35 were used to show PG accumulation under IL relative to NL conditions. (E) Plants were grown in long-day conditions for 3 weeks under NL and IL as explained in (A), and the leaves were excised for electron microscopy. Scale bars represent 500 nm. (F) Quantification of chloroplast area (in µm²). The bars represent the SE and asterisks represent significant differences (P<0.05) in comparison with the WT (Student’s t-test). (G) Numbers of PGs per chloroplast section. The bars represent the SEM and asterisks represent significant differences (P<0.05) relative to the WT based on the t-test.

It has been reported that PG proteins are strongly expressed under increased light; however, PG18 was not among the proteins showing increased accumulation (Ytterberg et al., 2006). To verify this, we monitored the levels of PG18 in plants grown under NL and IL conditions. Consistent with the MS-based quantification in Ytterberg et al. (2006), PG18 is equally abundant in NL and IL plants, while expression of the core PG protein PGL35 increases upon exposure to high light levels (Fig. 2C).

Since the pale phenotype indicated a defect in chloroplasts, we proceeded to analyze the phenotype at the ultrastructural level (Fig. 2E). Rosette leaves of 3-week-old WT, pg18 mutant, and complemented plants grown under NL and IL conditions were analyzed. In comparison with the WT, the cross-sectional area of pg18 chloroplasts was significantly smaller under both conditions (Fig. 2F). When examined at higher magnification, the stroma lamellae in pg18 mutant chloroplasts were shorter and appeared to be stacked, rather than present in single layers. Moreover, the stroma lamellae exhibited less branching between grana stacks, while the latter displayed a more compact organization and generally consisted of more layers in the mutant than in the WT or the complemented line. This effect was observed under both NL and IL conditions, although it was more pronounced under IL (Fig. 2F, lower panels). Notably, under both light conditions, PG abundance per chloroplast was also reduced significantly in the mutant chloroplasts in comparison with the WT (Fig. 2G). The number of PGs doubled under IL in the WT. Strikingly, this effect of exposure to IL was also observed in the mutant, in spite of the overall reduction in the numbers of PGs (Fig. 2G).

As a next step, we aimed to analyze changes in the lipid content of the WT and pg18 mutant. We analyzed the major thylakoid lipids monogalactosyldiacylglycerol (MGDG) and digalactosyldiacylglycerol (DGDG) and, albeit not statistically significant, we observed a slight reduction of DGDG under NL conditions in the mutant. A similar effect was caused by IL treatment already in the WT; however, no further reduction of these lipids was observed in the mutant under IL conditions (Supplementary Fig. S1). Sulfoquinovosyldiacylglycerol (SQDG) and the phospholipids phosphatidylglycerol and phosphatidylcholine remained essentially unchanged in the mutant (Supplementary Fig. S1).

Arabidopsis pg18 mutant plants show symptoms of light stress

To assess differences in thylakoid pigments that might be related to the mutant phenotype, the pigment content of pg18 leaves was analyzed (Fig. 3). Chl a and b were indeed reduced in the mutant, in agreement with its pale green phenotype (36% and 20% reduction for Chl a and b, respectively) under both growth light conditions. However, only the reduction in Chl a relative to the WT was significant (Fig. 3A, left panel), resulting in a significantly decreased Chl a:b ratio in the mutant (Fig. 3A, right panel).

Fig. 3.

Pigment analysis of the WT and pg18 mutants. Pigment extracts isolated from 3-week-old Arabidopsis plants grown in a long-day photoperiod under NL and IL conditions were quantified by HPLC. (A) Chlorophyll content on a fresh weight basis (left panel) and Chl a/Chl b ratio (right panel). (B) Content of β-carotene (Car), lutein (Lut), and neoxanthin (Nx) normalized to 1000 Chl (a+b). (C) Content and de-epoxidation state (DEPS) of the xanthophyll cycle pigments violaxanthin (Vx), antheraxanthin (Ax), and zeaxanthin (Zx). The xanthophyll amount (left panel) including the VAZ pool size (=sum of Vx+Ax+Zx) is normalized to 1000 Chl (a+b). The DEPS (right panel) was calculated as (Zx+0.5×Ax)/(Vx+Ax+Zx)×100. Asterisks represent significant differences (Student’s t-test, P<0.05) in comparison with the WT. Data show mean values (±SE), n=4.

Analysis of the carotenoid content revealed that β-carotene (Car) was significantly reduced in the pg18 mutant compared with the WT, irrespective of the light conditions during growth. In contrast, the levels of lutein (Lut) as well as neoxanthin (Nx) were increased in the mutant (Fig. 3B). Furthermore, the amount (VAZ pool size) and de-epoxidation state (DEPS) of the xanthophyll cycle pigments violaxanthin (Vx), antheraxanthin (Ax), and zeaxanthin (Zx) were analyzed, since both parameters are reliable indicators of light stress. Indeed, a significant increase of both the VAZ pool size (Fig. 3C, left panel) and the DEPS (Fig. 3C, right panel) was determined for pg18 compared with the WT, with a more pronounced increase under IL conditions, supporting the view that the mutant is more susceptible to light stress. No difference in pigment accumulation was observed for the 35S::PG18-GFP line as compared with the WT (data not shown).

Photosynthetic performance is affected in pg18 mutant plants

The photosynthetic performance of both photosystems was analyzed in 3-week-old plants grown under NL and IL conditions using a pulse amplitude-modulated (PAM) fluorimeter. Compared with the WT, the F₀ values were 3- and 2-fold increased in the pg18 mutant grown under NL and IL conditions, respectively, indicative of a disturbed PSII function (Supplementary Table S3; Supplementary Fig. S2A). In accordance with this, the maximum PSII quantum yield expressed as the ratio of the variable to the maximum fluorescence (F_v/F_m) was reduced to ~66% (NL) and 56% (IL) in pg18 as compared with the WT, again indicating defects of PSII in the mutant lines (Supplementary Table S3). The fluorescence dropped far below the F₀ level during light induction in pg18, which is typical for mutants directly affected in PSII (Supplementary Fig. S2A). Both PSI and PSII yields were reduced in the pg18 mutant, especially under increased light intensities (Fig. 4A, B). Moreover, the theoretically deduced electron transport rate (ETR) was dramatically reduced by >3-fold in NL and ~5-fold in IL in the mutant compared with the WT, which is an indication of an inefficient linear electron flow (Fig. 4C). The non-photochemical quenching (NPQ) of excitation energy in the pg18 mutant was twice as high as in the WT even under low light, which very probably reflects an increased proton gradient across the thylakoid membrane (Fig. 4D). As can be seen from an image of whole plants recorded with an imaging PAM, the F_v/F_m value was equally reduced in all rosette leaves of the mutant (Fig. 4E). No difference from the WT was observed for the 35S::PG18-GFP line (data not shown).

Fig. 4.

Loss of PG18 affects photosynthetic performance. Arabidopsis plants (3 weeks old) grown under NL and IL conditions were subjected to chlorophyll fluorescence analysis. (A) Quantum yields of PSI, Y(I). (B) Quantum yield of PSII, Y(II). (C) Electron transport rate (ETR) in μmol electrons m⁻² s⁻¹. (D) Non-photochemical quenching (NPQ). Data represent mean values (±SE), n=3. (E) Maximum quantum yield of PSII (F_v/F_m) in 3-week-old WT and mutant plants. The color scale at the bottom indicates the signal intensities.

As compared with the WT, the quantum yield of non-photochemical energy dissipation due to PSI donor Y(ND) site limitation was increased in the mutant, whereas the acceptor side limitation Y(NA) was decreased, indicating that the electron transport towards PSI is limiting the overall photosynthetic electron transport (Supplementary Table S3). We further investigated the amount of electrons present within the plastoquinone pool involved in the photosynthetic electron transport, since a substantial portion of the plastoquinone is known to be present in PGs (Ksas et al., 2018). Interestingly, the size of the plastoquinone pool shared by both photosystems was twice as high in pg18 as in the WT (Supplementary Table S3). However, metabolite measurements performed with entire leaves showed no notable difference in the plastoquione amounts (Supplementary Table S4).

Next, we analyzed the ability to recover from photoinhibition by measuring the F_v/F_m value in response to 2 h exposure to 1000 µmol photons m⁻² s⁻¹ and a subsequent 6 h dark recovery phase. Relative to the initial dark F_v/F_m ratio, the high light-induced reduction (and thus photoinhibition of PSII) was the same for WT and pg18 plants, and the recovery rate was almost comparable with that of the WT, demonstrating a rather efficient repair system in the pg18 mutant (Supplementary Fig. 2B). Additionally, the translation efficiency was assessed by in vivo radiolabeling of WT and mutant pg18 plants with ³⁵S-labeled methionine for 20 min. The labeled proteins were separated by BN–PAGE. In line with the recovery experiment results, the translation as well as the complex assembly rate did not show differences (Supplementary Fig. 2C). Since the effect on the ETR and the quantum yield of both photosystems might be related to altered steady-state levels, stoichiometry, or assembly of the thylakoid membrane complexes in the mutant, we went on to investigate them in more detail.

Loss of PG18 has an impact on the accumulation of thylakoid membrane protein complexes

In order to monitor the abundance of the thylakoid membrane proteins in the WT and the pg18 mutant, we performed immunoblot analyses with antisera against subunits of the photosystems, the ATP synthase, and the Cyt b₆f complex. The results revealed that components of the ATP synthase (α and γ subunits), and the PsaG, PsaD, PsaF, and LHCA2 subunits of PSI were notably reduced as compared with WT levels (Fig. 5A, B). A slight reduction was also observed for the PSII core components D1 and CP47. However, mutant levels of Cyt f were unchanged relative to the WT. Unexpectedly, levels of LHCB2 were increased by ~20%. To analyze the assembly status of thylakoid complexes, thylakoid membranes were solubilized with 1% β-dodecylmaltoside and separated in BN gels, followed by SDS–PAGE in the second dimension. In accordance with the immunoblot results, the mutant pg18 showed higher accumulation of LHCBs compared with the WT, and a slight reduction in subunits of ATP synthase and PSI (Supplementary Fig. 3).

Fig. 5.

Loss of PG18 has an impact on the accumulation of thylakoid membrane complexes. (A) Thylakoids were isolated from plants grown on soil for 3 weeks under long-day conditions. Proteins in the thylakoid membranes were immunodetected using antibodies against subunits of PSI, PSII, ATP synthase, and the Cyt b₆f complex. Loading of 100% corresponds to 20 μg of protein, and the same amount was loaded for the mutant. CBB, Coomassie brilliant blue. (B) Quantitative analysis of immunoblots using ImageJ. Results are plotted as a percentage of WT levels, and are based on mean values for three replicates in each case. Error bars represent the SE with n=3.

Loss of the Synechocystis homolog of PG18 affects cyanobacterial fitness

PGs are not exclusive to plants, but are also found in cyanobacteria as lipid droplets (Stanier et al., 1988). Therefore, we asked whether homologs of PG18 are present in other organisms, since other PG proteins—such as FBNs—have been identified in cyanobacteria (Cunningham et al., 2010). Interestingly, PG18 turns out to be conserved from cyanobacteria to higher plants (Supplementary Fig. 4A; Supplementary Data S1, S2).

In order to investigate the function of the PG18 homolog in cyanobacteria, we generated a knock out mutant in Synechocystis by inserting a kanamycin resistance cassette into the sll1769 ORF (Supplementary Fig. 4B). However, in this mutant (synpg18⁻), the ultrastructure of the thylakoid membrane is apparently normal (Fig. 6A). Nevertheless, the growth rate of the synpg18⁻ strain was significantly reduced compared with the WT under photoautotrophic conditions, whereas no differences were observed under photoheterotrophic conditions (Table 1). In accordance with what was observed in Arabidopsis, the chlorophyll content of the synpg18⁻ mutant was reduced relative to the WT, as was photosynthetic performance—as reflected in the parameter F_v/F_m: 0.27 for synpg18⁻ as against 0.42 for the WT (Table 1). Since photosynthesis was affected in the synpg18⁻ mutant, we analyzed the accumulation of thylakoid membrane proteins. The synpg18⁻ mutant showed significantly lower levels of the PSI subunit PsaD and Cyt f of the Cyt b₆f complex, whereas other subunits were present in normal amounts (Fig. 6B).

Fig. 6.

Loss of SynPG18 function affects PSI and Cyt b₆f protein accumulation. (A) Transmission electron micrographs of WT and synpg18⁻ mutant Synechocystis cells. Scale bars=250 nm. (B) Quantification of immunoblots using ImageJ. Results are plotted as a percentage of WT levels, and are based on mean values for three replicates in each case, and the error bars represent the SE with n=3. (C) Whole-cell protein extracts from Synechocystis WT and synpg18⁻ (30 µg of protein for 100%) were separated by SDS–PAGE and analyzed by immunoblots using antibodies against PSI (PsaD), PSII (D1), ATP synthase (α- and β-subunits), Cyt f, and phycocyanin.

Table 1.

Physiological characteristics of the synpg18⁻ mutant

Strain	Doubling time (h)	Chlorophyll content (μg OD750−1)	F v/Fm
WT	8.15±0.40a 17.44±0.28b	2.22±0.34	0.41±0.01
synpg18 −	8.05±0.21a 19.34±0.13b	1.79±0.10	0.27±0.02

^a,b Doubling times in the ^apresence or ^babsence of 5 mM glucose, measured under continuous illumination at 30 μmol photons m⁻² s⁻¹ and CO₂-limiting conditions. The photosynthetic parameter F_v/F_m was obtained from chlorophyll fluorescence measurements. Data are mean values ±SD of three independent replicates.

Discussion

Several lines of evidence indicate that PGs play an important role in thylakoid development. Their number decreases during de-etiolation, while mutants with defects in thylakoid formation often accumulate larger numbers of PGs (Kroll et al., 2001; Babiychuk et al., 2008). Their function may consist of providing metabolites, such as carotenoids and prenylquinones. Additionally, they may be responsible for providing triglycerides for membrane lipid synthesis.

In this study, we have addressed the role and localization of PG18. PG18 is a protein of unknown function that was identified in PGs of higher plants by MS (Lundquist et al., 2012b). With the help of a PG18-specific antiserum, we confirmed that PG18 is indeed localized to PGs. Surprisingly, despite the absence of obvious hydrophobic regions in its sequence, PG18 behaves as a membrane protein. A previous study on the fibrillin PGL34, which likewise does not contain typical transmembrane domains, revealed that most deletion variants tested failed to localize to PGs when transiently expressed as GFP fusions (Vidi et al., 2007). This suggests that the association of PG proteins with the PG monolayer might rely on their folded conformation rather than a particular hydrophobic domain.

Our characterization of a loss-of-function mutant of PG18 clearly demonstrates that the lack of this protein has a severe impact on plant fitness throughout development. Mutant plants show a light green phenotype and are smaller than the WT. Since PGs have been shown to accumulate under light stress (Zhang et al., 2010), we assessed whether the mutant line exhibits a stronger phenotype under such conditions. Mutants for other PG proteins typically show enhanced phenotypes when subjected to light stress (Porfirova et al., 2002; Youssef et al., 2010; Fatihi et al., 2015). However, the phenotype of pg18 remained essentially the same under NL and IL conditions. This indicates that PG18 plays a more general role in chloroplast biogenesis. This notion is supported by the fact that pg18 plants have smaller chloroplasts and fewer PGs per chloroplast than the WT. Notably, the number of PGs was increased under IL conditions in both WT and pg18 chloroplasts, although the latter always contained fewer PGs per chloroplast. This indicates that PG18 is not involved in promoting PG accumulation under light stress.

The pale green leaves of pg18 plants point to a reduction in chlorophyll content. This assumption was verified by analyzing the chlorophyll content of leaves from plants grown under NL and IL conditions, and the most significant reduction was noted in Chl a. The carotenoid fraction revealed a strong reduction in β-carotene in the pg18 mutant compared with the WT under both NL and IL conditions. This reduction as well as the reduced Chl a/b ratio support an increased PSII/PSI ratio, as also shown by analysis of the protein levels. Moreover, the increase in Vx and Nx (on a chlorophyll basis) is in good agreement with the observed relatively higher amounts of LHCII.

Analysis of the pigments of the xanthophyll cycle in the pg18 mutant showed that levels of the photoprotective pigments Ax and Zx were increased relative to the WT, and further enhanced under IL conditions. These pigments are synthesized from Vx under light stress and contribute to energy dissipation (NPQ) and thus to photoprotection of PSII (Gilmore et al., 1995; Farber et al., 1997; Jahns and Holzwarth, 2012). Interestingly, levels of the photoprotective prenyllipid α-tocopherol, which is stored in PGs, increased in the WT after IL treatment. This effect, however, was not observed in pg18 when comparing NL and IL conditions (Supplementary Table S4). Taken together, pg18 mutant plants showed symptoms of light stress even under normal light conditions, which probably explains the accumulation of larger grana stacks as well as more LHCII in their chloroplasts.

Chlorophyll fluorescence analyses showed that photosynthetic performance in general was affected, with both photosystems showing lower levels and quantum yields than the WT, and these deficits became somewhat more pronounced under increased light intensities. Interestingly, we observed elevated levels of phylloquinone in the mutant, which might result from reduced PSI levels, thus leading to an excess of phylloquinone (Supplementary Table S4). In turn, this shows that phylloquinone, which is stored in PGs, is not limiting. The severely increased F₀ value in pg18 mutants can partially be explained by the over-representation of LHCB2, which is not functionally connected to PSII but contributes to the fluorescence. The increased NPQ is in accordance with the reduced levels of the ATP synthase causing accumulation of lumenal protons and thus increased dissipation of heat via qE. This observation correlates with higher levels of Zx and Ax found in the mutant plants relative to the WT. These pigments are known to modulate NPQ by deactivating excited states in the PSII antenna, and also acting as antioxidants in the thylakoid membrane (Havaux et al., 1991; Havaux and Niyogi, 1999; Nilkens et al., 2010). The decreased electron transport was also confirmed by the increased PSI donor side limitation, although levels of this photosystem were more reduced than those of PSII or the Cyt b₆f complex, indicating strong down-regulation of PSII by increased NPQ. Furthermore, the increased size of the intersystemic electron pool indicates that levels of PQ molecules involved in photosynthesis are unchanged but that less active photosystems share the same PQ pool size as the WT. Complementation of the mutant line rescued both the photosynthetic activity and reduced pigment accumulation to WT levels (data no shown).

Moreover, to investigate the light sensitivity of PSII and the ability to repair it after photoinhibition via de novo synthesis and incorporation of D1 under IL conditions, we performed light stress recovery analysis. Strikingly, both the WT and pg18 were equally affected by light stress, but the WT recovered slightly faster than pg18. Therefore, the pleiotropic effects on photosynthetic performance in pg18 chloroplasts probably result from assembly defects or altered stoichiometry of the photosynthetic complexes and/or antenna proteins. In line with this, analysis of the accumulation of thylakoid proteins revealed a strong reduction in the ATP synthase and PSI complexes. In agreement with the reduction in PSI, levels of Chl a were sharply reduced in the pg18 mutant compared with the WT, which is also reflected in the fact that the ratio of Chl a to Chl b remains essentially the same under NL and IL conditions in pg18. Furthermore, our finding that the stromal lamellae are shorter and less branched in the pg18 mutant than in the WT (Fig. 2E) correlates with the reduction in levels of PSI and ATP synthase, as less membrane area is available for their integration. In summary, the overall stoichiometry of the complexes in the thylakoid membrane is affected in the pg18 mutant. Interestingly, there is a clear accumulation of LHC proteins in the mutant line, which is compatible with the fact that pg18 plants accumulate more photoprotective LHC-binding pigments such as Zx (Johnson et al., 2007). This effect can also be observed in the high level of NPQ in the pg18 mutant, which is further enhanced under IL, in accordance with the increased levels of Zx in the mutant relative to the WT under both NL and IL conditions. Hence, PG18 seems not to intervene directly in the assembly of any particular complex in the thylakoid membrane, but rather affects the composition of some of them, possibly by modulating the structural organization of the thylakoid membrane during its biogenesis.

Despite the lack of identifiable functional domains, suggesting that PG18 is not itself an enzyme, loss of PG18 has a significant impact on the composition and architecture of the thylakoid membrane. Considering that PGs play an important role in mobilizing lipids for incorporation into the thylakoid membrane (Deruere et al., 1994; Simkin et al., 2007), an alteration in PGs could be expected to affect thylakoid complexes, as has been observed when the lipid content of the thylakoid membrane is altered (Zhou et al., 2009; Kansy et al., 2014). The observed higher MGDG:DGDG ratio in the mutants correlated with the increased grana formation and stacking of thylakoid membranes (Lee, 2000).

Levels of PG18 were not found to be specifically up-regulated under stress conditions, nor does it accumulate to a greater extent in PGs isolated after high light treatment (Ytterberg et al., 2006). These findings indicate that it is a constitutive component of PGs, possibly fulfilling a general role in PG maintenance or interacting with other PG proteins. This inference is supported by the observation that mutants for other PG proteins do not show a phenotype under normal conditions. Phenotypes only become manifest when PG mutants are exposed to stresses, such as high light intensities (Porfirova et al., 2002; Singh et al., 2010; Youssef et al., 2010; Martinis et al., 2013; Avendano-Vazquez et al., 2014; Fatihi et al., 2015). Moreover, PG18 was also found in chromoplasts of red pepper (Ytterberg et al., 2006), suggesting that its function is not restricted to chloroplasts. It is noteworthy, however, that PG18 is phosphorylated, which might lead to conformational changes or otherwise have an impact on its activity, possibly depending on different developmental stages or stress conditions (Wang et al., 2013; Lohscheider et al., 2016). Interestingly, co-expression analysis previously showed that PG18 is co-expressed with a number of components of the chloroplast redox network, such as STN7 and the thioredoxins (Lundquist et al., 2012b). Redox regulation is known to be important for regulation of photosynthesis, especially under short-term acclimation to different light conditions. It is therefore highly feasible that PGs, and PG18 in particular, are involved in regulation via redox-controlled processes.

In accordance with what we observed in Arabidopsis, loss of SynPG18 has similar effects on Synechocystis fitness. Although no alterations in thylakoid ultrastructure were observed in the mutant cyanobacteria, growth rate and chlorophyll content were lower than in the WT, as in the case of A. thaliana. The reduction in growth rate becomes manifest only when cyanobacteria were grown under photoautotrophic conditions, which is compatible with the assumption that SynPG18 is involved in photosynthetic performance. Additionally, F_v/F_m was reduced in the synpg18⁻ mutant. Examination of the composition of the photosynthetic complexes revealed reduced amounts of PsaD and Cytf, while other proteins such as D1 and ATP synthase subunit α were present in WT levels. This result differs slightly from what we observed in A. thaliana, where ATP synthase subunits were more strongly reduced and Cytf was unchanged. This interspecies difference can probably be explained from an evolutionary perspective. It can be assumed that in cyanobacteria PG18 has a role in thylakoid biogenesis that has been conserved throughout evolution, while new features leading it to affect ATP synthase accumulation in green plants could have been acquired subsequently. This assumption is supported by the evidence that, in plants, PGs have multiple roles in development, senescence, and light stress, making them more complex than their counterparts in cyanobacteria.

In summary, investigation of the pg18 mutant underlines the importance of PGs in the formation of thylakoid membranes and shaping their protein complex composition. Elucidating its exact role will be a challenging task to address in the future.

Supplementary data

Supplementary data are available at JXB online.

Data S1. Species and accession numbers.

Data S2. Protein sequence alignment of PG18 in based on the accessions given in Supplementary Data S1.

Fig. S1. Lipid analysis in the WT and pg18.

Fig. S2. Photosynthetic performance and photoinhibition/recovery in the WT and pg18.

Fig. S3. Loss of PG18 function affects PSI and ATP synthase complexes.

Fig. S4. Conservation of PG18.

Table S1. List of oligonucleotides used in this research.

Table S2. List of antisera used in this research.

Table S3. Photosynthetic parameters.

Table S4. Measurements of prenyllipids.

Glossary

Abbreviations

FBN

fibrillin

PG

plastoglobule

PG18

plastoglobular protein 18