CGL160-mediated recruitment of the coupling factor CF₁ is required for efficient thylakoid ATP synthase assembly, photosynthesis, and chloroplast development in Arabidopsis

Abstract

Chloroplast ATP synthases consist of a membrane-spanning coupling factor (CF_O) and a soluble coupling factor (CF₁). It was previously demonstrated that CONSERVED ONLY IN THE GREEN LINEAGE160 (CGL160) promotes the formation of plant CF_O and performs a similar function in the assembly of its c-ring to that of the distantly related bacterial Atp1/UncI protein. Here, we show that in Arabidopsis (Arabidopsis thaliana) the N-terminal portion of CGL160 (AtCGL160N) is required for late steps in CF₁-CF_O assembly. In plants that lacked AtCGL160N, CF₁-CF_O content, photosynthesis, and chloroplast development were impaired. Loss of AtCGL160N did not perturb c-ring formation, but led to a 10-fold increase in the numbers of stromal CF₁ subcomplexes relative to that in the wild type. Co-immunoprecipitation and protein crosslinking assays revealed an association of AtCGL160 with CF₁ subunits. Yeast two-hybrid assays localized the interaction to a stretch of AtCGL160N that binds to the DELSEED-containing CF₁-β subdomain. Since Atp1 of Synechocystis (Synechocystis sp. PCC 6803) could functionally replace the membrane domain of AtCGL160 in Arabidopsis, we propose that CGL160 evolved from a cyanobacterial ancestor and acquired an additional function in the recruitment of a soluble CF₁ subcomplex, which is critical for the modulation of CF₁-CF_O activity and photosynthesis.

IN A NUTSHELL

Background: Thylakoid ATP synthases are impressive molecular engines that harness the light-driven proton gradient to generate ATP during photosynthesis. Their molecular mode of operation and atomic structure have been elucidated, but their assembly process is still under investigation. Specific auxiliary factors assist in ATP synthase assembly and prevent the accumulation of dead-end products or deleterious intermediates. CGL160 is one such factor and consists of a membrane and an N-terminal domain. The membrane domain of CGL160 is distantly related to bacterial Atp1 proteins, which are also present in cyanobacteria. Previous studies demonstrated that CGL160 promotes efficient formation of the membranous c-ring of thylakoid ATP synthases in Arabidopsis thaliana.

Question: What is the function of the green lineage-specific N-terminal domain of CGL160 in thylakoid ATP synthase assembly, and what is the evolutionary relationship between CGL160 and Atp1?

Findings: Here, we showed that the N-terminal domain of CGL160 is required for the late steps in thylakoid ATP synthase assembly and recruits the stromal ATP synthase intermediate coupling factor CF₁. The assembly step is critical for chloroplast development in the dark, ATP synthase activity, and photosynthesis in A. thaliana. We also revealed that Atp1 from the cyanobacterium Synechocystis spec PCC 6803 could functionally replace the membrane domain of CGL160 in A. thaliana. These results indicated that Atp1 operates in c-ring assembly in cyanobacteria and that CGL160 evolved from its cyanobacterial ancestor Atp1. However, CGL160 acquired an additional function in linking a soluble ATP synthase intermediate to a membranous subcomplex.

Next steps: The next steps are to identify all auxiliary factors required for the assembly of thylakoid ATP synthases and to understand their precise function in ATP synthase formation. Detailed knowledge of the factors and the assembly process could provide elegant strategies for adjusting proton circuits and altering the ATP budget in crops or other photosynthetic organisms.

Introduction

F-type ATP synthases, which utilize chemiosmotic membrane potentials to generate ATP, are central actors in the energy metabolism of bacteria, mitochondria, and chloroplasts. These biological nanomotors share a largely conserved structure, consisting of a soluble F₁ and a membrane-bound F_O moiety. Bacterial and chloroplast ATP synthases (CF₁-CF_O) are closely related with respect to size and subunit composition (Groth and Pohl, 2001; Vollmar et al., 2009; Hahn et al., 2018) and, in contrast to the multimeric mitochondrial ATP synthases, exist as monomers in thylakoid membranes (Daum et al., 2010). In the chloroplasts of land plants, CF₁-CF_O complexes reside exclusively in stroma lamellae and grana-end membranes, because the ∼16-nm stromal extension of CF₁ prevents its incorporation into the tightly packed grana stacks (Daum et al., 2010).

During photophosphorylation, CF₁-CF_O complexes couple the light-driven generation of the trans-thylakoid proton-motive force (pmf) to ADP phosphorylation. The membrane-embedded proteolipid c₁₄-ring, together with the non-covalently bound central stalk γε, form the motor unit and drive rotary catalysis by CF₁. The peripheral stator consists of the subunits a, b, and b′, and is connected to the (αβ)₃ unit by the δ subunit, which acts as a flexible hinge between CF₁ and CF_O (Murphy et al., 2019). Protons are translocated from the luminal to the stromal side via two aqueous channels in the CF_O-a subunit. During translocation, each proton enters the access channel and binds to a conserved glutamate residue in subunit CF_O-c. The c₁₄ motor executes an almost complete rotation before releasing the proton into the stroma through the exit channel (Hahn et al., 2018). The counterclockwise rotation of the central stalk in the vicinity of the hexamer triggers alternating nucleotide-binding affinities in the β subunits that ultimately drive ATP generation (reviewed in von Ballmoos et al., 2009; Junge and Nelson, 2015).

As a result of extensive organellar gene transfer during plant evolution, three CF₁-CF_O subunits (b′, γ, δ) are encoded in the nuclear genome, while the remaining CF₁-CF_O genes are organized into two plastid operons. Consequently, two different gene-expression systems must be tightly coordinated with the chloroplast protein import machinery for efficient CF₁-CF_O biogenesis. Several CF₁-CF_O auxiliary factors involved in plastid gene expression have been identified, including proteins involved in mRNA processing (ATPF EDITING FACTOR 1, AEF1), mRNA stabilization (PENTATRICOPEPTIDE REPEAT 10, PPR10; BIOGENESIS FACTOR REQUIRED FOR ATP SYNTHASE 2, BFA2), and translation initiation (ATPASE SPECIFIC DEFECT 4, ATP4; TRANSLATION DEFICIENT ATPase 1, TDA1; Pfalz et al., 2009; Eberhard et al., 2011; Zoschke et al., 2012; Yap et al., 2015; Zhang et al., 2019). Moreover, CF₁-CF_O assembly factors ensure correct complex stoichiometry and prevent the accumulation of dead-end products or harmful intermediates that could lead to wasteful ATP hydrolysis or pmf dissipation.

As in the case of the bacterial assembly model, plastid CF₁-CF_O complexes are constructed from different intermediates or modules (reviewed in Rühle and Leister, 2015). CF₁ assembly was first examined using in-vitro reconstitution assays, and was shown to be initiated by α/β dimerization in a chaperone-assisted process (Chen and Jagendorf, 1994). CF₁ formation depends on CONSERVED IN THE GREEN LINEAGE AND DIATOMS 11 (CGLD11 or alternatively BIOGENESIS FACTORS REQUIRED FOR ATP SYNTHASE 3, BFA3), which is specific to green plants, interacts with the hydrophobic catalytic site of the β-subunit and may prevent aggregation or formation of unfavorable homodimers (Grahl et al., 2016; Zhang et al., 2016). Moreover, PROTEIN IN CHLOROPLAST ATPASE BIOGENESIS (PAB; Mao et al., 2015) and BIOGENESIS FACTORS REQUIRED FOR ATP SYNTHASE 1 (BFA1; Zhang et al., 2018) have been proposed to be required for efficient incorporation of the γ subunit into CF₁.

Less is known about CF_O assembly, and only one accessory factor—CONSERVED ONLY IN THE GREEN LINEAGE160 (CGL160)—has been identified so far (Rühle et al., 2014). Absence of CGL160 in the Arabidopsis (Arabidopsis thaliana) mutant cgl160-1 is associated with a significant reduction (70%–90%) in wild-type (WT) CF₁-CF_O levels, and CF_O-c subunits accumulate as monomers. Moreover, split-ubiquitin assays have provided evidence that AtCGL160 interacts with CF_O-c and CF_O-b. It was therefore concluded that AtCGL160 is required for efficient formation of the c-ring in chloroplasts and shares this function with its distantly related bacterial counterpart ATP SYNTHASE PROTEIN 1 (Atp1 or alternatively named UncI because it is encoded by the first gene in the unc operon; Suzuki et al., 2007; Ozaki et al., 2008). Furthermore, AtCGL160 was suggested to participate in CF₁ assembly into the holo-complex, based on CF₁ subcomplex co-migration and crosslinking experiments using a putatively specific anti-AtCGL160 antibody (Fristedt et al., 2015).

In this study, the function of the N-terminal domain that is conserved in all CGL160 proteins from the green lineage was investigated in Arabidopsis. The results demonstrate that this domain (AtCGL160N) mediates the critical connection of CF₁ to CF_O assembly modules by interacting with subunit β. Thus, CGL160 emerges as a key auxiliary factor that not only promotes CF_O formation but is also involved in late CF₁-CF_O assembly steps.

Results

The N-terminal moiety of AtCGL160 is required for efficient photosynthesis and CF₁-CF_O functionality

CGL160 was identified based on its coregulation with photosynthetic genes in ATTED-II (Obayashi et al., 2009) and its affiliation to the GreenCut suite of proteins (Merchant et al., 2007; Karpowicz et al., 2011). The C-terminal transmembrane segment of CGL160 (∼15 kDa) is distantly related to bacterial Atp1/UncI (Rühle et al., 2014; Fristedt et al., 2015), whereas the larger N-terminal portion of the protein sequence is only conserved in algae, bryophytes, and vascular plants (Supplemental Figure S1). This latter domain of ∼200 amino acids (aa) in Arabidopsis (AtCGL160N) includes a predicted, putative N-terminal chloroplast transit peptide (TP) of 46 aa (Emanuelsson et al., 1999). However, the alignment of AtCGL160 and CGL160N sequences from other species in the green lineage (Supplemental Figure S1) revealed that AtCGL160_29–46aa is conserved in vascular plants, indicating incorrect chloroplast TP prediction by ChloroP. Moreover, mass spectrometry has identified several phosphorylated peptides which are derived from positions 106–134 (Reiland et al., 2009; Reiland et al., 2011; Roitinger et al., 2015). Indeed, two conserved putative phosphorylation sites were found in the multiple sequence alignment of CGL160 homologs from species across the green lineage, which correspond to positions S111 and S126 in AtCGL160 (Figure 1, Supplemental Figure S1). Earlier studies have provided experimental evidence for the localization of AtCGL160 to the thylakoid membrane (Rühle et al., 2014; Tomizioli et al., 2014; Fristedt et al., 2015). To gain further insights into the topology of AtCGL160, a protease protection assay was carried out (Figure 1, B and C). In the case of topology 1, all trypsin cleavage sites in AtCGL160 reside in the lumen of the thylakoid and remain fully protected from proteolytic degradation (Figure 1B). Conversely, the stromal orientation of AtCGL160N predicted for topology 2 would expose trypsin cleavage sites and lead to degradation products of less than 2 kD (Figure 1A). To test the accessibility of native AtCGL160N, WT thylakoids were isolated and treated with trypsin for 10 min (Figure 1C). As expected, the luminal PSII subunit PsbO was not affected by the enzyme, whereas the stromally exposed PSI subunit PsaD was susceptible to the protease. AtCGL160N was also efficiently digested, leaving no detectable proteolytic cleavage products, which is consistent with protrusion of the entire N-terminal domain into the stroma, as shown in Topology 2 (Figure 1B).

Figure 1

Topology of AtCGL160 and trypsin cleavage-site prediction. A, Transmembrane (TM) domain predictions were obtained from the AtCGL160 UniProt protein accession O82279. Putative trypsin cleavage sites are highlighted in dashed lines and amino acid positions are indicated. The topology was drawn for the full-length sequence of AtCGL160 with Protter (Omasits et al., 2014). The precise length of the AtCGL160 chloroplast TP is unknown. ChloroP chloroplast TP prediction (1–46 aa) and manual annotation (1–28 aa) based on sequence comparison (Supplemental Figure S1) is depicted in green and bright green, respectively. Two conserved serine residues (S111 and S126) are marked in yellow. B, Representation of two putative AtCGL160 topologies. The four TM domains are indicated as black boxes. Accessible trypsin digestion sites are highlighted by red stars. C, Immunoblot of thylakoid membranes of the WT (Col-0) fractionated by SDS-PAGE, untreated (0 min) or treated with trypsin for 1, 2, 5, and 10 min. Blots were probed with antibodies against the lumen-oriented PSII subunit PsbO (antibody AS05 092, Agrisera), the stroma-exposed PSI subunit PsaD (antibody AS09 461, Agrisera) and AtCGL160 (antibody AS12 1853, Agrisera).

To dissect the function of the N-terminal portion of AtCGL160, three different constructs under control of the 35S promoter were cloned, transformed into the Atcgl160-1 background and screened for complementation (Figure 2, Supplemental Figure S2A). Plants that overexpressed the full-length coding sequence (CDS) of AtCGL160 served as controls (P_35S:AtCGL160), while the other two genotypes expressed either the CDS of the N-terminal (P_35S:AtCGL160N) or the C-terminal segment (P_35S:AtCGL160C) of the protein (Supplemental Figure S2B). In the case of P_35S:AtCGL160C plants, targeting of the truncated version to chloroplasts was achieved by fusing the CDS of the AtCGL160-derived ChloroP-predicted, putative chloroplast TP (1–46 aa) to that of AtCGL160C (Figure 2A).

Figure 2

Growth phenotype and leaf variegation of P_35S:AtCGL160, P_35S:AtCGL160N, and P_35S:AtCGL160C plants under short-day conditions. A, Schematic representations of reintroduced AtCGL160 coding sequences. Plants lacking AtCGL160 were transformed with overexpressor constructs harboring the coding sequences for the full-length AtCGL160 (P_35S:AtCGL160) and its N- (P_35S:AtCGL160N) and C-terminal (P_35S:AtCGL160C) segments. Transcription was under the control of the 35S CaMV promoter and targeting to the chloroplast was mediated by the ChloroP-predicted TP of AtCGL160 (TP). Amino acid positions are indicated and predicted TM domains (TM1–TM4) are schematically shown as black boxes. B, Leaf morphology of Col-0, atcgl160-1, P_35S:AtCGL160, P_35S:AtCGL160N, P_35S:AtCGL160C, and atcgld11-1 plants. C, Leaf areas of six individual plants per genotype were determined 4 weeks after germination. The horizontal lines represent the median, and boxes indicate the 25th and 75th percentiles. Whiskers extend the interquartile range by a factor of 1.5×, and outliers are represented by dots. The effect of the deletion of AtCGL160N in P_35S:AtCGL160C plants on growth under short-day conditions was tested by paired sample t test (two-sided). Statistically significant differences are marked with asterisks (*P < 0.05, **P < 0.01, and ***P < 0.001).

Since the commercially available AtCGL160 antibody (Agrisera AS12 1853) displayed non-specific binding to either CF₁-α or CF₁-β (Supplemental Figure S3A), an AtCGL160 antibody with no significant cross-reactions to other thylakoid proteins was generated (Supplemental Figure S3B). In the first step of antibody production, the N-terminal part of AtCGL160 conserved in vascular plants (AtCGL160_29–206aa) was fused to the maltose-binding protein and injected into rabbits. In the second step, antibodies specific for AtCGL160_29–206aa were affinity-purified from rabbit antisera using an immobilized fusion protein consisting of AtCGL160_29–206aa and glutathione S-transferase. As expected, when the resulting antibody fraction was tested in immunodetection assays, it showed only one distinct signal in the WT sample, which was enriched in extracts of P_35S:AtCGL160, but was absent in the atcgl160-1 sample (Supplemental Figure S3B).

As was previously demonstrated in complementation analyses with P_35S:AtCGL160-eGFP lines (Rühle et al., 2014), overexpression of the full-length AtCGL160 rescued the atcgl160-1 phenotype (Supplemental Figure S2A), as indicated by WT-like growth and restored leaf morphology under short-day conditions (Figure 2, B and C). P_35S:AtCGL160N failed to complement the mutant phenotype (Figure 2, B and C, Supplemental Figure S2A) and AtCGL160N could not be detected in either thylakoid (Supplemental Figure S3B) or leaf extracts (Supplemental Figure S3C). Since AtCGL160N transcripts were present in WT-like amounts in P_35S:AtCGL160N plants (Supplemental Figure S2B), the lack of AtCGL160N is probably due to proteolytic degradation owing to its inability to associate correctly with thylakoids. Nevertheless, P_35S:AtCGL160N plants were retained and served as an additional AtCGL160 knockout control. P_35S:AtCGL160C plants with similar overexpression rates to P_35S:AtCGL160 plants (Supplemental Figure S2B) were characterized by a significant increase in leaf area compared to the mutant background atcgl160-1 but were growth-retarded with respect to the WT control. Interestingly, like atcgl160-1, P_35S:AtCGL160C plants developed a variegated phenotype in old leaves, which was not found either in the WT or in the CF₁ assembly mutant atcgld11-1 (Grahl et al., 2016) under short-day conditions (Figure 2B).

To analyze the leaf phenotype in more detail, we carried out electron microscopic analyses of Col-0, atcgl160-1 and P_35S:AtCGL160C plants (Figure 3). In these genotypes, the chloroplast ultrastructure in preparations from green leaf sections was unchanged with regard to grana number per chloroplast and grana height distribution (Figure 3, A–F, Supplemental Figure S4). These observations in atcgl160-1, together with previous ultrastructural analyses of the CF₁ assembly mutant line atcgld11-1 (Grahl et al., 2016) and spinach (Spinacia oleracea) chloroplasts (Daum et al., 2010), support the idea that CF₁-CF_O complexes are not physically involved in thylakoid curvature formation. Examination of white leaf sections in atcgl160-1 and P_35S:AtCGL160C revealed the absence of thylakoids in plastids, accompanied by the appearance of plastoglobuli in densely packed stromal clusters (Figure 3, G–J). Furthermore, large vesicles were observed, which also point to increased catabolic activity and degradation processes in atcgl160-1 and P_35S:AtCGL160C plastids. Another finding was the inclusion of mitochondria in degraded plastids, which was also observed, to a lesser extent, in white leaf sectors of P_35S:AtCGL160C.

Figure 3

Plastid ultrastructure in white leaf sectors is altered in the absence of AtCGL160N under short-day growth conditions. Electron micrographs of samples from green leaf sections obtained from Col-0 (A and B), atcgl160-1 (C and D), and P_35S:AtCGL160C (E and F) plants. The ultrastructure of chloroplasts was further examined in samples of white leaf sections obtained from atcgl160-1 (G and H) and P_35S:AtCGL160C (I and J) plants. The photos on the right show enlargements of the images on the left. The scale bar corresponds to 2 µm in (A, C, E, and G), 1 µm in (I) and 0.5 µm in (B, D, F, H, and J).

To test whether disruption of AtCGL160N impairs photosynthesis and CF₁-CF_O activity, measurements of chlorophyll a fluorescence and electrochromic shift (ECS) were carried out on Col-0, atcgl160-1, P_35S:AtCGL160, P_35S:AtCGL160N, P_35S:AtCGL160C, and atcgld11-1 plants (Figure 4, A–D). As expected, the CF₁-CF_O assembly mutants atcgl160-1 and atcgld11-1 showed lower effective PSII quantum yields [Y(II)], higher heat dissipation (indicated as non-photochemical quenching, NPQ) and increased pmf, but lower proton conductivity (⁠${\mathrm{g}}_{\mathrm{H}}^{+}$⁠) through the thylakoid membrane compared to the WT control. P_35S:AtCGL160 and P_35S:AtCGL160N plants displayed similar levels of NPQ, pmf and ${\mathrm{g}}_{\mathrm{H}}^{+}$ to the WT and the CF₁-CF_O assembly mutant atcgld11-1, respectively. Notably, photosynthetic parameters in the P_35S:AtCGL160C line were only partially restored and Y(II) (Figure 4A), as well as proton flux (ν_H⁺) through the photosynthetic apparatus were decreased compared to that in Col-0 (Figure 4D).

Figure 4

Lack of AtCGL160N perturbs photosynthesis and CF₁-CF_O integrity. A, PSII quantum yield (Y(II)) and heat dissipation (NPQ) of Col-0, atcgl160-1, P_35S:AtCGL160, P_35S:AtCGL160N, P_35S:AtCGL160C, and atcgld11-1 plants were determined using an Imaging-PAM system (Walz) and are displayed on a rainbow color scale (left panel; 0–1 for Y(II) and 0–4 for NPQ). Y(II) (middle panel) and NPQ (right panel) analyses from six plants per genotype and five leaves (n = 30) were analyzed at 185 µmol photons m⁻² s⁻¹. B, DIRK derived from ECS signals were recorded after 10 min of illumination from six individual plants grown under short-day conditions. To determine the pmf, total amplitude of the P515 differential absorption signal was normalized to a single turnover flash 4 min after the ECS measurement. C, Proton conductivity of the thylakoid membrane was determined from ECS signal relaxation rates, which were fitted to a first-order decay function. The inverse of the calculated rate constant was expressed as ${\mathrm{g}}_{\mathrm{H}}^{+}$ [s⁻¹]. Measurements were obtained from six individual plants grown under short-day conditions. D, The proton flux parameter ν_H⁺ was determined from the initial rate of decay of the ECS signal. Measurements were conducted on six individual plants grown under short-day conditions. E, Steady-state levels of immunodetected CF₁-CF_O marker subunits. After fractionation of thylakoid proteins on SDS-PAGE and transfer to PVDF membranes, blots were probed with antibodies against CF₁-β, CF₁-γ, CF_O-b, and CF_O-c. Coomassie Brilliant Blue (C.B.B.) staining is shown as loading control. For quantification, signals from four technical replicates of each marker subunit were normalized to signals detected in Col-0 samples. Horizontal lines represent the median, and boxes indicate the 25th and 75th percentiles. Whiskers extend the interquartile range by 1.5×. The effect of the deletion of AtCGL160N on photosynthetic parameters of P_35S:AtCGL160C plants shown in panels (A–D) was tested in paired-sample t tests (two-sided). Statistically significant differences are marked with asterisks (*P < 0.05, **P < 0.01, and ***P < 0.001).

To assess the integrity of the CF₁-CF_O complex in thylakoids, marker subunits were immunodetected in atcgl160-1, P_35S:AtCGL160, P_35S:AtCGL160N, P_35S:AtCGL160C, and atcgld11-1 plants, and quantified relative to Col-0 samples (Figure 4E). Levels of CF₁-β, CF₁-γ, CF_O-b and CF_O-c were restored to normal in P_35S:AtCGL160 but reduced to about 60%–65% of WT amounts in P_35S:AtCGL160C plants. Transformation with the P_35S:AtCGL160N construct had no effect on CF₁-CF_O subunit levels in the atcgl160-1 mutant.

Overall, overexpression of the Atp1/Unc1-like AtCGL160 domain alone (AtCGL160C) in the atcgl160-1 background only partially restored CF₁-CF_O amounts (Figure 4E) and activity (Figure 4C). Consequently, photosynthesis was downregulated and growth of P_35S:AtCGL160C plants was impaired (Figure 2C), which could be attributed to more highly activated ΔpH-dependent quenching mechanisms (Figure 4A). We deduced from these results that AtCGL160N might also be involved in CF₁-CF_O assembly at steps other than CF_O-c ring formation.

Stromal accumulation of CF₁ in the absence of AtCGL160N

To investigate the effects of deletion of AtCGL160N on CF₁-CF_O assembly, we performed BN/SDS-PAGE (2D-PAGE) analysis on thylakoids isolated from P_35S:AtCGL160 and P_35S:AtCGL160C plants grown under short-day conditions. Consistent with the accumulation of CF₁-CF_O marker subunits in Figure 4E, CF₁-β, CF_O-b, and CF_O-c levels were reduced in P_35S:AtCGL160C compared to the levels in plants that overexpressed the full-length CDS of AtCGL160 (Figure 5A). No accumulation of pre-complexes was observed, as amounts of free proteins, and components of the c-ring, CF₁, and the holo-complex were reduced uniformly. To assess the assembly status of the c-ring in more detail, we carried out 2D-PAGE with increased amounts of atcgl160-1 and P_35S:AtCGL160C thylakoids (Figure 5B). C-ring levels were considerably higher in P_35S:AtCGL160C than in the atcgl160-1 mutant background. We also examined CF₁ accumulation in the stroma of Col-0, P_35S:AtCGL160, P_35S:AtCGL160N, P_35S:AtCGL160C, and atcgld11-1 plants (Figure 5C), since CF₁-CF_O assembly takes place in a modular fashion and involves distinct thylakoid-integral and soluble intermediates. Strikingly, CF₁-β and CF₁-γ were enriched about 10-fold in the stroma of atcgl160-1, P_35S:AtCGL160N, and P_35S:AtCGL160C, but were detected in close to WT levels in P_35S:AtCGL160 and atcgld11-1 plants. In-depth 2D-PAGE analysis of CF₁ intermediates in atcgl160-1, and comparison with results from the co-migration database for photosynthetic organisms (PCom-DB, Takabayashi et al., 2017), revealed that in atcgl160-1 stromal CF₁-β and CF₁-γ were predominantly present in an α₃β₃γε complex that lacked subunit CF₁-δ (Supplemental Figure S5).

Figure 5

CF₁-CF_O assembly is perturbed in the absence of AtCGL160N. A, Thylakoid complexes from P_35S:AtCGL160 and P_35S:AtCGL160C plants were solubilized with n-dodecyl β-D-maltoside (1% [w/v]) and further separated by Blue-Native (BN, first dimension) and SDS-PAGE (SDS, second dimension). After protein transfer, PVDF membranes were probed with antibodies against CF₁-β, CF_O-b, and CF_O-c, and CF₁-β blots were subsequently exposed to anti-AtCGL160 antibodies (generated this study). Positions of the ATP synthase holo-complex (CF₁-CF_O) and the c-ring are indicated. Coomassie Brilliant Blue G-250 (C.B.B.) staining of PVDF membranes is shown as loading control. B, C-ring assembly in atcgl160-1 and P_35S:AtCGL160C plants. Increased amounts of thylakoid complexes (corresponding to 120 µg total chlorophyll) were solubilized and fractionated by BN/SDS-PAGE. Blots were probed with an antibody against CF_O-c. Positions of free c-monomers and the assembled c-ring are indicated. C, CF₁-β and CF₁-γ enrichment in stromal extract, which was isolated from Col-0, atcgl160-1, P_35S:AtCGL160, P_35S:AtCGL160N, P_35S:AtCGL160C, and atcgld11-1 rosette leaves. Signals of three CF₁-β and CF₁-γ immunodetection assays were quantified and are shown on a logarithmic scale. Horizontal lines represent the median, boxes indicate the 25th and 75th percentiles and whiskers extend the interquartile range by a factor of 1.5×.

We concluded that reintroduction of the transmembrane Atp1/Unc1-like domain of AtCGL160 restores c-ring formation, but leads to an overall reduction in CF₁-CF_O levels due to a defect in the attachment of CF₁ to a membrane-integral CF_O intermediate.

AtCGL160 interacts physically with CF₁-containing complexes

To pinpoint the role of AtCGL160 in the recruitment of CF₁ to a membrane-integral CF_O intermediate, protein interactions were assessed in co-immunoprecipitation (co-IP) assays (Figure 6, A and B). NP40-solubilized thylakoid proteins from P_35S:AtCGL160 plants grown under short-day conditions were chosen as co-IP input and pulled-down protein amounts were compared to those recovered in co-IP experiments carried out on thylakoid protein extracts of P_35S:AtCGL160C. Plants devoid of AtCGL160 were not considered for use as negative controls, since the reduction in CF₁-CF_O levels observed in atcgl160-1 (and P_35S:AtCGL160N; Figure 4E) might lead to misinterpretation of differential co-IP experiments. Quantitative data for precipitated proteins were obtained by tryptic digestion and subsequent peptide-fragment analysis using liquid chromatography coupled to mass spectrometry (Figure 6A). As expected, AtCGL160 was pulled down efficiently from P_35S:AtCGL160 extracts (log₂ FC ∼6.5). Moreover, all CF₁-CF_O subunits were identified in co-IPs (Figure 6, A and B) with high differential enrichment levels for the subunits α, β, γ, δ, ε, b, b′, and c (log₂ FC > 4.4). Subunit CF_O-a was co-immunoprecipitated at lower levels (log₂ FC ∼2.8). The pull-down of CF₁ subunits was confirmed by immunodetection assays of the two marker subunits CF₁-β and CF₁-γ, which were only detectable in co-IP output fractions obtained from P_35S:AtCGL160 samples (Supplemental Figure S6). Other known CF₁-CF_O assembly factors were not co-immunoprecipitated (Supplemental Table S1), indicating that AtCGL160 is associated with a late CF₁-CF_O assembly stage or the fully assembled complex from which other auxiliary factors had already dissociated.

Figure 6

AtCGL160 association with CF₁ subunits. A, Co-immunoprecipitation analyses with the newly generated antibody were carried out with solubilized thylakoids isolated from P_35S:AtCGL160, while P_35S:AtCGL160C plants served as the negative control. Co-immunoprecipitated proteins were further subjected to tryptic digestion, and peptides were analyzed by liquid chromatography coupled to mass spectrometry. Data for differentially enriched proteins are presented in a volcano plot. Each point indicates a different protein, ranked according to P-value (y-axis, −log₁₀ of P-values) and relative abundance ratio (x-axis, log₂ Fold Change P_35S:AtCGL160/P_35S:AtCGL160C). Protein candidates of interest are labelled in red. The dashed line indicates a negative log₁₀P-value of 1.5. Blue and red dots highlight AtCGL160 and CF₁-CF_O subunits, respectively. B, Schematic representation of differentially enriched subunits in a CF₁-CF_O cartoon. Relative amounts of co-immunoprecipitated CF₁-CF_O subunits are shown in colors on a log₂ FC scale from white (log₂ FC < 2) to red (log₂ FC > 6). Co-immunoprecipitation assays were carried out with three independent biological replicates. In each replicate, co-immunoprecipitation was performed with solubilized thylakoids isolated from ∼10 g leaf fresh weight of P_35S:AtCGL160 or P_35S:AtCGL160C plants. C, Co-migration of AtCGL160 with CF₁-CF_O in crosslinking experiments. Two-dimensional BN/SDS-PAGE analysis was used to compare untreated thylakoid extracts of the WT (Col-0) with extracts that had been crosslinked with DSP. Blots of the second dimension were probed with antibodies against AtCGL160 and CF_O-γ. The positions of CF₁-CF_O, the CF₁ intermediate, and the free protein fraction are indicated based on the mobility of α/β on the C.B.B. stained gel.

Since ALBINO 4 (ALB4), which is a member of the bacterial ALBINO 3 (ALB3)/OXIDASE ASSEMBLY 1 (Oxa1)/YidC protein insertase family, was previously proposed to participate in the linkage of a CF₁ to a CF_O assembly module (Benz et al., 2009), but was not pulled down in co-IP assays (Supplemental Table S1), the amount of thylakoid-associated CF₁ complexes was re-assessed in atalb4-1 (SALK_136199C) mutants and compared to levels identified in atcgl160-1 plants (Supplemental Figure S7). Thylakoids were isolated from Col-0, atcgl160-1 and atalb4-1 plants grown under short-day conditions and subjected to immunodetection assays of subunit CF₁-β. CF₁-β levels bound to atcgl160-1 thylakoids were reduced to 23 ± 13% of the WT level, while no deviation to the WT control could be observed for thylakoid-associated CF₁-β content in atalb4-1 samples (106 ± 30%).

To confirm the association of AtCGL160 with CF₁-containing complexes, crosslinking experiments were also carried out (Figure 6C). To this end, thylakoid membranes of WT plants were treated with the crosslinker dithiobis(succinimidyl propionate) (DSP), and subsequently subjected to 2D-PAGE and immunodetection of AtCGL160 and CF₁-CF_O marker subunits. In analyses with untreated thylakoid samples, AtCGL160 migrated predominantly in the monomer fraction. After crosslinking, AtCGL160 could be detected at a molecular mass range that corresponded to that of the CF₁-CF_O holo-complex.

In summary, co-IP of all CF₁-CF_O subunits with an AtCGL160-specific antibody, together with the observation that AtCGL160 co-migrated with the CF₁-CF_O holo-complex after DSP cross-linking, corroborates the involvement of AtCGL160 in the functional integration of CF₁ into the holo-complex at a late step in CF₁-CF_O assembly.

AtCGL160N interacts with CF₁-β in yeast two-hybrid assays

Interactions between the stroma-oriented AtCGL160N domain and individual CF₁ subunits were further examined by yeast two-hybrid experiments (Figure 7). Since AtCGL160_29–46aa is conserved in other vascular plants (Supplemental Figure S1) and might be part of the mature protein, it was considered in yeast two-hybrid assays. To this end, a construct coding for a fusion of AtCGL160N_29–206aa to the GAL4-binding domain (BD) was co-transformed into yeast cells together with constructs coding for GAL4 activation domain (AD) fusions to all CF₁ subunits (α, β, γ, δ, ε). Moreover, the BD-AtCGL160N interaction was tested with AD fusions to the soluble parts of the stator subunits b and b′, AtCGL160N, and CF₁ assembly factor AtCGLD11. As a result, only yeast cells carrying constructs for AD-CF₁-β and BD-AtCGL160N could grow on selective medium (Figure 7A). However, the reciprocal constructs BD-CF₁-β and AD-AtCGL160N did not interact (Supplemental Figure S8). To narrow down the CF₁-β interaction site, additional AD fusion constructs were cloned that encoded three different CF₁-β subdomains (Figure 7B) defined in earlier studies (Groth and Pohl, 2001; Zhang et al., 2016). Domain I comprises a thylakoid-distal β-barrel structure and interacts with CF₁-δ. Domain II harbors the catalytic site involved in ATP generation or hydrolysis. The thylakoid-proximal domain III contains the conserved “DELSEED” motif, which is required for CF₁-γ/ε-dependent regulation of ATP hydrolysis and synthase activity (Kanazawa et al., 2017; Hahn et al., 2018). When tested on restrictive medium, only cells harboring AD-CF₁-β_III together with BD-AtCGL160N could grow. In a reciprocal approach, coding sequences of AtCGL160N were deleted successively from the BD-AtCGL160N construct (Δ29–73, Δ74–105, Δ106–134, Δ135–160, and Δ161–206 aa) and tested for AD-CF₁-β interaction in yeast cells (Figure 7C). Only the Δ29–73 and Δ74–105 aa deletions resulted in an absence of growth, while yeast strains with deletion constructs of Δ106–134, Δ135–160, and Δ161–206 aa were able to proliferate on selective medium (Figure 7C). Thus, the interaction between AtCGL160 and CF₁ involves AtCGL160_29–105 and the thylakoid-proximal domain of CF₁-β_III, while the phosphorylation hotspot identified in the protein segment 106–134 aa (Figure 1A) is dispensable for the interaction.

Figure 7

AtCGL160N interaction studies in yeast two-hybrid assays. A, Interactions of AtCGL160 with CF₁-CF_O structural components exposed on the stromal side of thylakoids were tested by transformation of a construct that fuses AtCGL160N to the GAL4 DNA-BD (BD-AtCGL160). Cells were then co-transformed with constructs coding for GAL4 AD fused to CF₁-α, β, γ, δ, ε, CF_O-b_51–184 or CF_O-b′_109–219, as well as AtCGL160N or AtCGLD11. B, Interaction of AtCGL160N with structural domains of CF₁-β. Yeast cells carrying a construct coding for BD-AtCGL160 were transformed with constructs coding for AD-CF₁-β_I, AD-CF₁-β_II, and AD-CF₁-β_III. Structural domains of the CF₁-β are colored in red (Domain I), turquoise (Domain II), and blue (Domain III). The conserved DELSEED motif is shown in purple, and the L-shaped redox loop of CF₁-γ in orange. The atomic model of CF₁-CF_O was obtained from the PDB database (ID: 6fkh, Hahn et al., 2018) and formatted with ChimeraX (Pettersen et al., 2021). C, Mapping of the AtCGL160N interaction site. Consecutive regions (grey boxes) coding for segments of the soluble AtCGL160 domain were omitted from the BD-AtCGL160N vector and co-transformed with AD-CF₁-β into competent yeast cells. Transformations in all assays were verified by plating on permissive medium lacking Leu and Trp (-LT). Interactions were then tested on selective medium (-Leu/-Trp/-His/-Ade, [-LTHA]) by plating equal numbers of yeast cells in serial dilutions (10°, 10⁻¹, and 10⁻²). All experiments were performed at least twice in independent co-transformation assays. Note that the reciprocal constructs BD-CF₁-β and AD-AtCGL160N did not interact in yeast two-hybrid assays (Supplemental Figure S8).

Atp1 of Synechocystis (Synechocystis sp. PCC 6803) can functionally replace the membrane domain of AtCGL160 in Arabidopsis

The membrane domain of AtCGL160 shares moderate sequence similarity to Atp1 (SynAtp1) of Synechocystis and is encoded by the first gene in the atp1 operon (Supplemental Figure S9A). To investigate its function in cyanobacteria, a knockout of the SynAtp1 gene was generated. Since keeping the endogenous promoter in the homologous recombination cassette would result in unwanted recombination events without deletion of synatp1, it was replaced by the strong psbA2 promoter P_psbA2 (Supplemental Figure S9A). As a reference, a Synechocystis strain (P_psbA2:synatp1) was generated in which the intact synatp1 operon was under control of P_psbA2. Full segregation could be achieved for control strain P_psbA2:synatp1 but not for P_psbA2:Δsynatp1 even in the presence of 500 µg mL⁻¹ kanamycin (Supplemental Figure S9B). Reverse transcription-quantitative PCR (RT-qPCR) analyses revealed that synatp1 expression in P_psbA2:synatp1 and P_psbA2:Δsynatp1 was increased ∼4-fold and decreased to ∼10%, respectively, compared with expression levels of the WT control (Supplemental Figure S9C). Growth analyses (Supplemental Figures S9D and S10A) gave rise to similar maximal doubling times for the WT and the P_psbA2:synatp1 control strain under mixotrophic (∼9 h) and autotrophic (∼23 h) growth conditions. However, maximal doubling times of the non-segregated P_psbA2:Δsynatp1 were slightly longer under mixotrophic (∼10 h) and more pronounced under autotrophic (∼31 h) conditions (Supplemental Figure S10A). The effect of promoter replacement and SynAtp1 disruption was further investigated by gas exchange measurements using a Clark-type oxygen electrode (Supplemental Figure S10B). Neither oxygen consumption in the dark nor production rates at 400 µmol photons m⁻² s⁻¹ were altered in both strains compared to that in the WT control. Investigations of CF₁-CF_O marker subunit and CF_O-c multimer accumulation by immunodetection showed no apparent differences among the genotypes (Supplemental Figure S10C). We concluded from those results that the endogenous promoter of the synatp1 operon could be replaced by the strong psbA2 promoter and synatp1 overexpression did not affect respiration, photosynthesis or CF₁-CF_O accumulation. Moreover, SynAtp1 might have an essential function in Synechocystis, since segregation could not be achieved even in the presence of high selection pressure. As a consequence, residual genomic copies carrying the intact synatp1 gene under control of the endogenous synatp1 promoter and synatp1 expression levels of ∼10% were still sufficient in the non-segregated P_psbA2:Δsynatp1 strain to maintain CF₁-CF_O assembly at WT levels.

Since functional analyses with the non-segregated P_psbA2:Δsynatp1 Synechocystis strain turned out to be challenging, we pursued a strategy of cross-species complementation and replaced the membrane domain of AtCGL160 with SynAtp1 (Figure 8). Accordingly, an overexpressor construct coding for the fusion of AtCGL160N (1–206 aa) and SynAtp1 (2–117 aa) was transformed into the atcgl160-1 background (Figure 8A). Two independent P_35S:AtCGL160N-SynATP1 progenies were examined with respect to photosynthesis and chloroplast ATP synthase activity (Figure 8, B–G). Remarkably, both lines restored the photosynthetic phenotype to normal, as Y(II) and NPQ levels determined in light induction experiments were comparable to those of the WT control (Figure 8, B–D). The same observation could be made for the pmf and proton conductivity, which were recovered to WT levels in the two P_35S:AtCGL160N-SynATP1 lines (Figure 8, E and F). No effect on proton flux could be identified in P_35S:AtCGL160N-SynATP1 overexpressor plants (Figure 8G). Accumulation and thylakoid localization of AtCGL160N-SynAtp1 in the atcgl160-1 background could be confirmed by immunodetection using the AtCGL160N-specific antibody (Figure 8H). For both lines, the presence of AtCGL160N-SynAtp1 led to a partial restoration of thylakoid-associated CF₁ (70%–80%) and CF_O (80%–90%) to WT levels (Figure 8H).

Figure 8

Synechocystis Atp1 (SynAtp1) can functionally substitute for the membrane domain of AtCGL160 in Arabidopsis. A, Replacement of the AtCGL160 membrane domain by SynAtp1. TP, TM domains of SynAtp1 (TM1–TM4) and amino acid positions are indicated. B, Imaging-PAM analysis of two independent P_35S:AtCGL160N-SynATP1 overexpressor plants (atcgl160-1 background) grown under short-day conditions. Effective quantum yield of photosystem II (Y(II)) and NPQ were recorded 600 s after light induction (110 µmol photons m⁻² s⁻¹) and are displayed on a false color scale from 0 to 1 and 0 to 4, respectively. C, Y(II) analyses of P_35S:AtCGL160N-SynATP1 overexpressor plants. Y(II) was quantified from five plants per genotype and four measurements taken from individual leaves per plant (n = 20). D, NPQ analyses of P_35S:AtCGL160N-SynATP1 overexpressor plants. NPQ was quantified from five plants per genotype with four measurements taken from individual leaves per plant (n = 20). E, The pmf determined in ECS measurements. F, Thylakoid membrane proton conductivity ${\mathrm{g}}_{\mathrm{H}}^{+}$ [s⁻¹] derived from fast ECS kinetics analyses. G, Proton flux ν_H⁺ through the photosynthetic apparatus. ECS parameters (E–G) were analyzed on single leaves of 6–7 individual plants per genotype. For statistical analyses (C–G), the non-parametric Kruskal–Wallis test was performed, followed by pairwise Dunn’s tests. The P-values were adjusted on an experiment level using the Benjamini–Hochberg method. P-values (P ≤ 0.05) are displayed in bold. H, Immunodetection assays of AtCGL160N and CF₁-CF_O marker subunits in thylakoids isolated from two P_35S:AtCGL160N-SynATP1 overexpressor lines #1 and #2. The newly generated antibody was employed to detect the fusion protein. Coomassie brilliant blue G-250 staining (C.B.B.) of a PVDF membrane is provided as loading control. Thylakoid-associated CF₁ (CF₁-β and CF₁-γ) and CF_O (CF_O-b′ and CF_O-c) subunits were quantified from three biological replicates (3–6 pooled plants per replicate) and amounts (in %) were referred to WT samples. Horizontal lines in boxplots shown in C and E represent the median, boxes indicate the 25th and 75th percentiles and whiskers extend the interquartile range by a factor of 1.5×.

In summary, SynAtp1 could replace AtCGL160C and function in the chloroplast CF₁-CF_O assembly process. However, a very high level of the chimeric AtCGL160N-SynAtp1 protein was required to achieve partial restoration in CF₁-CF_O content in Arabidopsis. Conversely, a conserved role in CF_O assembly could be inferred for Atp1 in Synechocystis, as AtCGL160C alone was able to promote c-ring formation (Figure 5B).

Discussion

AtCGL160 evolved from the cyanobacterial ancestor Atp1

The assembly of protein complexes generally depends on a network of auxiliary factors that must exhibit a high degree of compatibility with their substrates. For instance, functional Rubisco was assembled in Escherichia coli only in the presence of the chloroplast chaperonin system CHAPERONIN 60 (Cpn60)/CHAPERONIN 20 (Cpn20), which could not be replaced by the bacterial GroEL (large protein encoded in the GroE operon)/GroES (small protein encoded in the GroE operon) system (Aigner et al., 2017). It is therefore particularly noteworthy that SynAtp1 was compatible with the CF_O assembly process in chloroplasts despite having only moderate sequence similarity to the membrane domain of AtCGL160 (Rühle et al., 2014). However, CF_O-c sequence conservation from cyanobacteria to plants with 80% identical amino acids is high (Lill and Nelson, 1991), and might account for functional cross-species complementation with a construct encoding an AtCGL160N-SynAtp1 fusion (Figure 8). From these results, we concluded that CGL160 evolved from the cyanobacterial CF_O assembly factor Atp1 and acquired an N-terminal domain with a distinct function in the green lineage.

AtCGL160N recruits a stromal α₃β₃γε complex for late CF₁-CF_O assembly steps

Despite structural similarities and comparable subunit compositions, the number of known assembly factors for ATP synthases is markedly higher in chloroplasts than in bacterial systems (reviewed in Zhang et al., 2020). Thus, the expanded molecular inventory for CF₁-CF_O assembly in chloroplasts might reflect the need for tight post-translational control of CF₁-CF_O formation, since the complex plays a central role in pmf utilization and regulation of photosynthesis (reviewed in Avenson et al., 2005). In this context, an important finding of previous studies was that disruption of full-length AtCGL160 (Rühle et al., 2014; Fristedt et al., 2015) was more detrimental to levels of functional ATP synthase than the loss of Atp1/UncI in non-photosynthetic bacteria (Gay, 1984; Liu et al., 2013). Furthermore, we show here that expression of P_35S:AtCGL160C in plants that lack AtCGL160N only partially restores CF₁-CF_O levels and activity (Figure 4). These observations prompted us to investigate the molecular function of the green-lineage-specific AtCGL160N in the CF₁-CF_O assembly process in more detail.

Several lines of evidence suggest that the N-terminal domain of AtCGL160 recruits a stromal CF₁ intermediate, while the C-terminal segment participates in c₁₄-ring assembly: (1) AtCGL160N protrudes into the stroma, as deduced from protease protection assays (Figure 1); (2) formation of the c₁₄ ring is restored in the presence of AtCGL160C alone, but CF₁ accumulates strongly in the stroma in the absence of AtCGL160N (Figure 5); (3) CF₁ subunits are differentially enriched in co-IP analyses performed with solubilized thylakoids isolated from P_35S:AtCGL160 plants (Figure 6, A and B); (4) AtCGL160 co-migrates with a large complex after DSP-mediated crosslinking (Figure 6C); and (5) AtCGL160N interacts with CF₁-β in yeast two-hybrid experiments (Figure 7).

A role for AtCGL160 in the incorporation of CF₁ into the holocomplex was previously proposed by Fristedt et al. (2015). This assumption was based on the observations that AtCGL160 co-migrated with CF₁ subcomplexes in BN/SDS-PAGE analyses and could be cross-linked to CF₁ subunits in WT protein samples. However, we detected AtCGL160 predominantly in the monomer fraction in untreated thylakoid preparations in this study (Figure 6C), as well as in previous work (Rühle et al., 2014), and co-migration of AtCGL160 with high-molecular-mass complexes was only observed after thylakoid proteins had been crosslinked with DSP (Figure 6C). Furthermore, the commercially available AtCGL160 antibody (AS12 1853, Agrisera) employed in the study of Fristedt et al. (2015) was found here to cross-react strongly with CF₁-α or CF₁-β (Supplemental Figure S3A), which complicates the interpretation of one-dimensional co-migration and crosslinking experiments in the absence of appropriate controls. Therefore, a new antibody was generated that does not cross-react with CF₁-CF_O subunits and thus provides a reliable means of probing the molecular interactions of AtCGL160 (Supplemental Figure S3B).

In addition to CGL160, ALB4 was previously suggested to be involved in joining a CF₁ to a CF_O module (Benz et al., 2009). Another study provided evidence that ALB4 and its paralog ALB3 physically interact with each other and show significant functional overlap in the membrane insertion of subunits of the Cyt b₆f complex (Trosch et al., 2015). Moreover, alleles of ALB4 (or alternatively SUPPRESSOR OF TIC40 LOCI 1, STIC1) have been identified as suppressors of the chloroplast protein import mutant tic40 (Bedard et al., 2017), and ALB4/STIC1 and SUPPRESSOR OF TIC40 LOCI 2 (STIC2) were shown to act together in thylakoid protein targeting in a pathway that also involves chloroplast SIGNAL RECOGNITION PARTICLE 54 (cpSRP54) and a chloroplast homolog of the bacterial SRP receptor FtsY (cpFtsY). In our study, we did not identify ALB4/STIC1 in co-IP experiments with anti-AtCGL160 antibodies (Figure 6, Supplemental Table S1), and amounts of thylakoid-associated CF₁-β in atalb4-1 mutants (SALK_136199C) grown under short-day conditions were unaltered (Supplemental Figure S7). Thus, ALB4/STIC1 does not act in concert with CGL160 in late stages of CF₁-CF_O assembly but serves as a general thylakoid protein biogenesis factor involved in folding or assembly of a specific subset of transmembrane proteins (Bedard et al., 2017).

AtCGL160 is critical for chloroplast development in the dark

It has long been thought that the hydrolytic activity of CF₁-CF_O needs to be inactivated in the dark to prevent futile ATP depletion (Ort and Oxborough, 1992). However, analysis of the constitutively redox-activated γ-subunit mutant gamera, in which a “dark pmf” is maintained, revealed increased stability of photosynthetic complexes upon prolonged darkness, suggesting that a certain degree of ATPase activity may be beneficial during the night (Kohzuma et al., 2017). Concomitantly, several processes have been proposed to depend on the maintenance of a dark pmf. These include thylakoid protein transport via the Tat- and Sec-dependent pathways, modulation of protease activity and ion homeostasis in the chloroplast. In this regard, a remarkable influence of AtCGL160 disruption on leaf variegation (Figure 2) and chloroplast development (Figure 3) was observed exclusively under short-day conditions. Surprisingly, this phenotype was not detectable in atcgld11-1 plants with a defect in CF₁ assembly and reduced amounts of CF₁-CF_O compared to those in atcgl160-1 (Figure 4E). However, the leaf phenotype correlated with the accumulation of a CF₁ intermediate in the stroma (Figure 5C). Thus, AtCGL160-mediated CF₁ recruitment might also be critical in preserving the dark pmf at night. Alternatively, stroma-enriched CF₁ complexes (Figure 5C) could alter the chloroplast ATP/ADP ratio by excessive hydrolytic activity, and disturb ATP-dependent nocturnal processes that ultimately lead to premature chloroplast degradation (Figure 3).

AtCGL160 is a central CF₁-CF_O assembly factor with multiple functions

Assembly of membrane-embedded ATP synthase modules and their subsequent association with F₁ subcomplexes are critical steps in bacterial and organellar ATP synthase biogenesis, as premature formation of the proton-translocating channel between the c-ring and the a-subunit (equivalent to the ATP9 ring and the ATP6 subunit in mitochondria) can lead to uncontrolled dissipation of the pmf (Birkenhäger et al., 1999; Franklin et al., 2004), and only efficient integration of F₁ triggers ATP production. In this context, molecular aspects of the assembly processes were recently elucidated for bacterial (reviewed in Deckers-Hebestreit, 2013), as well as yeast and human mitochondrial ATP synthases (reviewed in Song et al., 2018). One significant outcome was that, while ATP synthase assembly pathways and the repertoire of auxiliary factors differ among these systems, formation of the proton-translocating unit during the final assembly steps is common to all of them. Intriguingly, our data revealed a dual involvement of AtCGL160 in CF₁-CF_O assembly, namely in c-ring formation and the recruitment of a CF₁ intermediate (Figure 9). In fact, these two events were suggested to proceed sequentially in the assembly of bacterial ATP synthases (Deckers-Hebestreit, 2013). Since an E. coli strain lacking subunit δ accumulates a c₁₀α₃β₃γε subcomplex, it is assumed that cytoplasmic F₁ first binds to the c₁₀ ring, and c₁₀α₃β₃γε associates with the ab₂ module in a δ-dependent manner in the final assembly step (Hilbers et al., 2013).

Figure 9

SynAtp1 and AtCGL160 in CF₁-CF_O assembly. (1) SynAtp1 (upper panel) and AtCGL160 (lower panel) operate both in c₁₄ assembly. (2) Recruitment of a stromal CF₁ intermediate (α₃β₃γε) is assisted by the N-terminal domain AtCGL160N specific for the green lineage. Interaction is mediated by AtCGL160_29–105 (light blue) and subdomain CF₁-β_III (blue), while the phosphorylatable segment 106–134 aa (in red) is dispensable for interaction. CF₁-β_III contains the conserved DELSEED motif (purple). (3) The exact timing of assembly factor release is unknown. However, SynAtp1 and AtCGL160 interacted with CF_O-b in split-ubiquitin assays (Rühle et al., 2014) and might stay attached to a CF₁-CF_O assembly intermediate until CF_O-a or CF₁-δ incorporation. The atomic model of spinach CF₁-CF_O was retrieved from the PDB database (ID: 6fkh, Hahn et al. 2018) and then formatted using ChimeraX (Pettersen et al., 2021).

By analogy with the bacterial assembly pathway, AtCGL160 may facilitate the integration of a stator assembly module into the holo-complex. Indeed, the interaction of AtCGL160C and SynAtp1 with CF_O-b has been demonstrated in split-ubiquitin assays (Rühle et al., 2014). Moreover, CF_O-a was less highly enriched in co-IP analyses than other CF₁-CF_O subunits (Figure 6, A and B), which might argue for the release of AtCGL160 after functional incorporation of CF_O-a in the final steps of CF₁-CF_O assembly (Figure 9). In this scenario, AtCGL160 could act as a placeholder to prevent the premature formation of proton-translocating intermediates. A similar function has been described for the inner membrane assembly (INA) complex in yeast mitochondria, which binds to the c-ring, but also to a distinct assembly intermediate consisting of ATP6, ATP8, ATP10, ATP23, peripheral stalk subunits, and the F₁ domain (Naumenko et al., 2017). This ensures that the c-ring and subunit ATP6 are assembled into the proton-conducting unit in a controlled manner. However, due to a generally low turnover rate of CF₁-CF_O assembly (reviewed in Schöttler et al., 2014) and inefficient detection of distinct thylakoid-integral intermediates, a robust CF_O assembly map is still lacking, and “true” stator-containing assembly modules have not been described so far.

Nevertheless, a straightforward assembly mechanism for the recruitment of CF₁ can be derived from our study (Figure 9). After AtCGL160-assisted ring formation (Rühle et al., 2014), the stromally oriented AtCGL160N (Figure 1) binds to a CF₁ intermediate consisting of α₃β₃γε but not subunit δ (Figure 5C, Supplemental Figure S5). Recruitment is mediated through interaction of AtCGL160_29–105 with subunit CF₁-β; thus, the phosphorylatable AtCGL160 segment is dispensable for the interaction (Figure 7). Since AtCGL160 can be cross-linked to high-molecular-mass complexes that are larger than CF₁ (Figure 6C), AtCGL160 might remain attached to a putative c₁₄α₃β₃γε or bb′c₁₄α₃β₃γε intermediate. Its release could then be triggered by the incorporation of subunit CF_O-a or CF₁-δ in the final assembly steps.

At this stage, we cannot rule out the possibility that AtCGL160N might have regulatory functions beyond CF₁ recruitment, as it interacts with the thylakoid-proximal domain III of CF₁-β, which contains the conserved DELSEED motif (Figure 7B). Several regulatory mechanisms have been elucidated in which the subunit β and the DELSEED motif are implicated. For instance, the autoinhibitory subunit ε interacts with the DELSEED motif in bacteria (Tanigawara et al., 2012; Sobti et al., 2016), whereas in bovine (Cabezon et al., 2003) and yeast mitochondria (Robinson et al., 2013), the small protein IF₁ inhibits ATPase activity by binding at the α/β interface. In plants, a regulatory mechanism controls CF₁-CF_O activity also involving the DELSEED and an L-shaped, two β-hairpin containing motif with two conserved redox-sensitive cysteines in the CF₁-γ subunit (Hahn et al., 2018). By analogy with the role of IF₁, which was shown to inhibit ATPase activity during the assembly of human mitochondrial ATP synthases (He et al., 2018), AtCGL160N may regulate ATPase activity during CF₁-CF_O assembly via an as yet unknown mechanism.

Methods

Bioinformatics sources

Protein and gene sequences were downloaded from the Arabidopsis Information Resource server (TAIR; http://www.arabidopsis.org), Phytozome (https://phytozome.jgi.doe.gov/pz/portal.html), and the National Center for Biotechnology Information server (NCBI; http://www.ncbi.nlm.nih.gov/). TPs were predicted by ChloroP (http://www.cbs.dtu.dk/services/ChloroP/; Emanuelsson et al., 1999). Structural data were obtained from the PDB homepage (https://www.rcsb.org/) and processed with ChimeraX (https://www.cgl.ucsf.edu/chimerax/; Pettersen et al., 2021). Multiple sequence alignments were generated with the CLC workbench software (v8.1) and protein features were visualized with Protter (https://wlab.ethz.ch/protter/start/; Omasits et al., 2014). Co-migration of stromal proteins was examined with the online tool PCom-DB (http://pcomdb.lowtem.hokudai.ac.jp/proteins/top; Takabayashi et al., 2017). Boxplots were drawn with BoxPlotR (http://shiny.chemgrid.org/boxplotr/; Spitzer et al., 2014).

Plant material and growth conditions

Arabidopsis (A. Thaliana) T-DNA lines for atcgl160-1 (SALK_057229, Col-0 background), atcgld11-1 (SALK_019326C, Col-0 background), and atalb4-1 (SALK_136199C) were obtained from the SALK collection (Alonso et al., 2003). Plants were grown in potting soil (A210, Stender, Schermbeck, Germany) under controlled greenhouse conditions (70–90 µmol photons m⁻² s⁻¹, 16/8-h light/dark cycles), or in climate chambers (equipped with 17-W cool white fluorescent lamps; CLF Plant Climatics) on an 8-h light/16-h dark cycle for biochemical and physiological analyses. Fertilizer was added to plants grown under greenhouse conditions according to the manufacturer’s recommendations (Osmocote Plus; 15% nitrogen [w/v], 11% [w/v] P₂O₅, 13% [w/v] K₂O, and 2% [w/v] MgO; Scotts, Germany). Plants were watered with tap water. For domain-specific complementation assays, either the complete coding region of AtCGL160 (P_35S:AtCGL160) or parts of the CDS corresponding to amino acids 1–206 (P_35S:AtCGL160N) and 207–350 (P_35S:AtCGL160C) were cloned into the binary Gateway vector pB2GW7 (Karimi et al., 2002), placing the genes under control of the cauliflower mosaic virus (CaMV) 35S promoter. The putative, TP coding sequence (for amino acids 1–46) was fused to the AtCGL160C CDS in the case of the P_35S:AtCGL160C vector. For the cross-species complementation construct, the codon usage of the synatp1 CDS (UniProt ID P27196) was optimized for expression in Arabidopsis (IDT, Coralville, IA, USA) and fused to the AtCGL160N sequence by fusion PCR. The constructs were first transformed into Agrobacterium tumefaciens strain GV3101, and then into atcgl160-1 plants by the floral-dip method (Clough and Bent, 1998). T1 plants were selected by screening for Basta resistance. Basta positives were screened for equal amounts of the AtCGL160 transcript by RNA gel-blot hybridization or immunodetection of AtCGL160N as described below.

Transmission electron microscopy

Leaf pieces of about 1.5 × 1.0 mm were cut with a new double edge razor blade (Feather, Osaka, Japan) and immediately immersed in fixation buffer (0.1 M sodium phosphate buffer, pH 7.4, 2.5% [v/v] glutaraldehyde, 4% [v/v] formaldehyde) at room temperature. A mild vacuum (about 20 mbar) was applied until the leaf pieces sank, the fixation buffer was replaced with fresh one and the samples were fixed overnight at 4°C. After three 10-min washes in sodium phosphate buffer (pH 7.4), the samples were osmicated with 1% osmium tetroxide and 1.5% potassium ferricyanide in 0.1 M sodium phosphate buffer (pH 7.4) for 60 min at 4°C. The samples were rinsed 3 times for 10 min each time in distilled water and incubated in 1% uranyl acetate (in distilled water) at 4°C overnight. After three washes of 10 min each in distilled water the samples were dehydrated using increasing concentrations of ethanol and infiltrated, with propylene oxide as an intermediate solvent, in glycid ether 100 (formerly Epon 812; Serva, Heidelberg, Germany) following standard procedures. Polymerization was carried out for 40–48 h at 65°C. Ultrathin sections (∼60 nm) were cut with a diamond knife (type ultra 35°; Diatome, Biel, Suisse) on an EM UC7 ultramicrotome (Leica Microsystems, Wetzlar, Germany) and mounted on single-slot Pioloform-coated copper grids (Plano, Wetzlar, Germany). The sections were stained using uranyl acetate and lead citrate (Reynolds, 1963) and viewed with a JEM-1400 Plus transmission electron microscope (JEOL, Tokyo, Japan) operated at 80 kV. Micrographs were taken using a 3.296 × 2.472 pixel charge-coupled device camera (Ruby, JEOL). The number of grana was determined in 25 chloroplasts per genotype from green leaf sections, where a granum was defined by at least two stacked thylakoid membranes. The grana height was analyzed using 417 (Col-0), 402 (atcgl160-1), and 406 (P_35S:AtCGL160C) grana.

Chl a fluorescence measurements

In vivo Chl a fluorescence of whole plants was measured using an imaging Chl fluorometer (Imaging PAM, Walz, Effeltrich, Germany). Plants were dark-adapted for 20 min and exposed to a pulsed, blue measuring beam (4 Hz, intensity 3, gain 3, damping 2; F_O) and a saturating light flash (intensity 10) to calculate Fv/Fm. PSII quantum yield [Y(II) = (Fm′ − F)/Fm′] and NPQ [NPQ = (Fm − Fm′)/Fm′] for Col-0, atcgl160-1, P_35S:AtCGL160, P_35S:AtCGL160N, P_35S:AtCGL160C, and atcgld11-1 plants were determined in light saturation curve analyses by gradually (every 180 s) increasing the light intensity (0, 20, 55, 110, 185, 280, 335, 395, 460, 530, 610, 700 µmol photons m⁻² s⁻¹), with only the data at 185 µmol photons m⁻² s⁻¹ presented in this study (Figure 4A). Y(II) and NPQ of P_35S:AtCGL160N-SynATP1 and control plants were recorded 600 s after light induction experiments (110 µmol photons m⁻² s⁻¹; Figure 8, B–D).

ECS measurements

ECS measurements were performed using the Dual-PAM-100 (Walz, Effeltrich, Germany) equipped with a P515/535 emitter-detector module (Schreiber and Klughammer, 2008). The measurement was carried out at 23°C under ambient CO₂ conditions. Plants grown in short-day conditions for 4 weeks were light-adapted, and detached leaves were illuminated for at least 10 min with 129 µmol photons m⁻² s⁻¹ red light. After illumination, dark-interval relaxation kinetics (DIRK) were measured in the millisecond to second range. Values for pmf (ECSt), and proton conductivity (⁠${\mathrm{g}}_{\mathrm{H}}^{+}$⁠) were calculated as described (Cruz et al., 2001; Schreiber and Klughammer, 2008). Briefly, the maximum amplitude of the inverse electrochromic band-shift kinetic was measured in the second range and normalized to a single saturating P515 pulse (ECS_ST) measured after 4 min of dark incubation. For proton conductivity, electrochromic band-shift kinetics were recorded in the millisecond range in five consecutive periods of darkness (2 s), separated by light intervals of 30 s (Schreiber and Klughammer, 2008). Averaged signals were fitted to a single exponential decay function and the reciprocal value of the ECS decay time constant was used to estimate the proton conductivity ${\mathrm{g}}_{\mathrm{H}}^{+}$ (Kanazawa and Kramer, 2002). Proton flux (ν_H⁺) was determined from the initial rate of decay (1–20 ms) of the ECS signal.

AtCGL160 antibody generation

Rabbit antibodies were generated against AtCGL160 that had been heterologously expressed in Escherichia coli and then purified. To this end, the coding sequence corresponding to AtCGL160_29–206 was cloned into the pMal-c5x vector (New England Biolabs) and purified on amylose columns (New England Biolabs) according to the manufacturer’s instructions. The protein was injected into rabbits for antibody production (Pineda, Berlin, Germany). To reduce epitope cross-reactions, the antiserum was purified on a column crosslinked with heterologously expressed AtCGL160_29–206 fused to the glutathione-S-transferase (GST) tag. Purified antibody was employed at a dilution of 1:1,000. Signals were detected by enhanced chemiluminescence (Pierce™ ECL Western Blotting Substrate, Thermo Scientific) using an ECL reader system (Fusion FX7; PeqLab, Erlangen, Germany).

Nucleic acid analysis

Total RNA from snap-frozen leaves was extracted with an RNeasy Plant Mini Kit (Qiagen) according to the supplier’s instructions. Samples equivalent to 20 µg total RNA were fractionated by electrophoresis in formaldehyde-containing agarose gels (1.2%), blotted onto nylon membranes (Hybond-N+, Amersham Bioscience) and fixed by UV radiation (Stratalinker UV Crosslinker 1800). To control for equal loading, abundant RNAs on nylon membranes were stained with methylene blue solution (0.02% [w/v] methylene blue, 0.3 M sodium acetate pH 5.5). To detect gene-specific transcripts, DNA fragments amplified from cDNA were labelled with radioactive [α-³²P]dCTP and subsequently used as probes in hybridization experiments (see Supplemental Table S2 for primer information). Signals were detected with the Typhoon Phosphor Imager System (GE Healthcare).

Protein analysis

Leaves from 4-week-old plants grown under short-day conditions were harvested shortly after the onset of the light period, and thylakoid membrane-enriched samples were isolated according to Rühle et al. (2014). Crosslinking of thylakoids was performed by incubation with 2.5 mM DSP (Thermo Scientific). After incubation for 20 min on ice, crosslinking was quenched with 60 mM Tris/HCl (pH 7.5). Chl concentrations were determined as described in Porra et al. (1989). For immunotitrations, thylakoid membrane pellets were resuspended in loading buffer (100 mM Tris/HCl pH 6.8, 50 mM dithiothreitol [DTT], 8% [w/v] SDS, 24% [w/v] glycerol and 0.02% [w/v] bromophenol blue). Denaturation for 5 min at 70°C and protein fractionation on Tricine-SDS-PAGE gels (10% gels supplemented with 4M urea) was carried out according to Schägger (2006). Immunodetections were performed as described below. Sample preparation for BN-PAGE was performed with freshly prepared thylakoids as described in Peng et al. (2008). First, membranes were washed twice in wash buffer (20% glycerol, 25 mM BisTris/HCl pH 7.0). Then, samples were treated with wash buffer containing 1% (w/v) n-dodecyl β-D-maltoside and adjusted to 1 mL mg⁻¹ Chl for 10 min on ice. After centrifugation (16,000 × g, 20 min, 4°C), supernatants were supplemented with 1/10 volume of BN sample buffer (100 mM BisTris/HCl pH 7.0, 750 mM ε-aminocaproic acid, 5% [w/v] Coomassie G-250). BN-PAGE gels (4%–12% gradient) were prepared as described in Schägger et al., (1994). Solubilized samples corresponding to 60 µg Chl were loaded per lane and gels were run at 4°C overnight. To separate complexes into their subunits, BN-PAGE strips were treated with denaturing buffer (0.2 M Na₂CO₃, 5% [w/v] SDS, 50 mM DTT) for 30 min at room temperature and loaded on Tricine-SDS-PAGE gels. Gels were subsequently subjected to immunoblot analysis with antibodies against CF₁-CF_O subunits and AtCGL160, as described below.

For analysis of the stromal CF₁ intermediate, intact chloroplasts from 4-week-old plants were isolated according to Rühle et al. (2021). After lysis in 25 mM HEPES/KOH (pH 7.5) containing 5 mM MgCl₂ for 30 min on ice, the stromal fraction was separated from membranes by centrifugation at 35,000 × g for 30 min (4°C). Protein concentration was measured using the Bradford Protein Assay (Bio-Rad). Stromal BN analysis was performed according to Reiter et al. (2020). In brief, chloroplast-enriched pellets were resuspended in BN washing buffer and mechanically disrupted by passage through an 0.45-mm syringe. The stromal fraction was separated from membranes by centrifugation at 35,000 × g for 30 min (at 4°C). Total soluble protein (100 µg) was mixed with 1/10 volume of BN sample buffer before fractionation in the first dimension as described above.

Immunoblot analyses

Proteins fractionated by gel electrophoresis were transferred to polyvinylidene difluoride membranes (PVDF; Immobilon-P, Millipore) using a semi-dry blotting system (BioRad) as described in the supplier’s instructions. After blocking with TBST (10 mM Tris/HCl pH 8.0, 150 mM NaCl and 0.1% [v/v] Tween-20) supplemented with 3% (w/v) skim milk powder, the membranes were incubated with antibodies at 4°C overnight. Antibodies used in this study were obtained from Agrisera (CF₁-β: AS05 085, 1:5,000; CF₁-γ: AS08 312, 1:5,000; CF_O-b: AS10 1604, 1:5,000; CF_O-c: AS09 591, 1:3,000; AtCGL160: AS12 1853, 1:1,000; PsbO: AS05 092, 1:10,000 and PsaD: AS09 461, 1:1,000).

Yeast-two-hybrid experiments

Yeast-two-hybrid assays were carried out using the Matchmaker Two-Hybrid System Kit (Clontech). The AtCGL160_29–206aa CDS without the signal peptide (see Supplemental Table S2 for primer information) was cloned into the bait vector pGBKT7 (BD-AtCGL160N), whereas the coding sequences of CF₁-α, -β, -γ, -δ, -ε, the soluble domains of CF_O-b (51–184 aa) and b' (109–219 aa), AtCGL160N and the CF₁ assembly factor AtCGLD11 were cloned into the prey vector pGADT7 (named AD-CF₁-α, AD-CF₁-β, AD-CF₁-γ, AD-CF₁-δ, AD-CF₁-ε, AD-CF_O-b, AD-CF_O-b', AD-AtCGL160N, and AD-AtCGLD11). As in the case of AtCGL160, signal peptide sequences were omitted from the nucleus-encoded subunits CF₁-γ, CF₁-δ, CF_O-b', and AtCGLD11. For testing interaction between AtCGL160N and CF₁-β in a reciprocal approach, the CDS of CF₁-β was also cloned into the bait vector pGBKT7 (BD-CF₁-β). For binding-domain analysis of CF₁-β, the respective CDS was subdivided into three parts, according to Groth and Pohl (2001), and cloned into pGADT7. In the case of AtCGL160N binding-site analysis, sequences coding for 29–73, 74–105, 106–134, 135–160, and 161–206 aa were deleted from the BD-AtCGL160 vector using a site-directed mutagenesis kit (NEB). Primers are listed in Supplemental Table S2. Bait and prey vectors were co-transformed into AH109 yeast strains (Clontech) following the manufacturer’s instructions. Co-transformants were selected on synthetic dropout (SD) medium (Clontech) lacking leucine and tryptophan (-LT). In order to identify protein interactions, double transformants were grown on SD medium lacking leucine, tryptophan, histidine, and adenine (-LTHA).

Co-immunoprecipitation

Freshly extracted thylakoids corresponding to ∼10 mg chlorophyll were resuspended in 500 µL extraction buffer (50 mM Tris/HCl pH 7.5, 150 mM NaCl, 1 mM MgCl₂, 5% [w/v] glycerol, 1% [v/v] Nonidet P40 [NP40], 0.2 mM phenylmethylsulfonyl fluoride [PMSF]) and solubilized for 30 min on ice. After centrifugation at 35,000 × g for 30 min and 4°C, the supernatant was added to 20 µL Dynabeads (Thermo Scientific), equilibrated with equilibration buffer (50 mM Tris/HCl pH 7.5, 150 mM NaCl, 5% [w/v] glycerol, 0.05% [v/v] NP40) and labelled with AtCGL160 antibody according to the manufacturer’s instructions. The suspension was incubated with rotation for 3 h at 4°C, washed 3 times with equilibration buffer, and twice with the same buffer but omitting NP40. Proteins were eluted with 100 µL 0.1 M glycine pH 2.0 for 10 min and neutralized with 100 µL 0.1 M ammonium bicarbonate. After treatment with 10 µL of 45 mM DTT and 10 µL of 0.1 M iodoacetamide, samples were digested with 1.5 µg of trypsin at 37°C overnight. Peptides were desalted with home-made C18 stage tips (Rappsilber et al., 2003), vacuum-dried to near dryness and stored at –80°C. LC MS/MS run and data analysis were performed as described in Reiter et al. (2020). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2022) partner repository with the dataset identifier PXD032230.

Generation and analysis of Synechocystis sp. PCC 6803 strains

Glucose-tolerant Synechocystis sp. PCC 6803 (Synechocystis) WT and mutant strains were grown in BG11 medium (Rippka et al., 1979) optionally supplemented with 5 mM glucose. Liquid cultures were kept at 30°C under continuous illumination (∼60 µmol photons m⁻² s⁻¹) on shakers (120 rpm). Cultures of P_psbA2:synatp1 and P_psbA2:Δsynatp1 strains were supplemented with 50 µg mL⁻¹ kanamycin. Synechocystis strain P_psbA2:synatp1 and P_psbA2:Δsynatp1 were generated by homologous recombination. Constructs were cloned using the Golden Gate cloning strategy (Engler et al., 2008) and assembled into the destination vector pICH69822 (Icon Genetics GmbH, Halle, Germany). Selection of positive transformants and segregation were conducted in the presence of increasing kanamycin concentrations (up to 500 µg mL⁻¹). DNA extraction and segregation analyses were carried out using the PHIre Plant Direct PCR-Kit according to the manufacturer’s instructions (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Total RNA was isolated from mixotrophically grown cells as previously described (Dann and Leister, 2019). RNA concentration was determined using a Nanodrop 2000 spectrophotometer (PeqLab, Erlangen, Germany). After adjusting RNA concentration and examination of RNA integrity by gel electrophoresis, samples were treated with DNase according to the manufacturer’s instructions (RNase-free DNase I, New England Biolabs). Synthesis of cDNA was carried out with 250 ng RNA using the iScript cDNA Synthesis Kit (BioRad, Munich, Germany). Quantitative real-time PCR analysis with primers described in Supplemental Table S2 was performed on a BioRad CFX Connect Real-Time system with the iQ SYBR Green Supermix (BioRad). Expression of synatp1 was quantified in technical triplicates, normalized to rrn16S expression (Pinto et al., 2012) and referred to synatp1 expression of the wildtype. Oxygen consumption and production rates were determined with the Clark-type oxygen electrode Oxytherm+ system (Hansatech Instruments Ltd, UK) at 30°C. In brief, oxygen uptake rates (nmol O₂ min⁻¹${\text{OD}}_{730}^{-1}$⁠) of cell cultures at the end of their exponential growth phase (OD₇₃₀∼0.2–0.4) were recorded for 10 min in the dark. Photosynthetic activity was determined by measuring the oxygen production rate (nmol O₂ min⁻¹${\text{OD}}_{730}^{-1}$⁠) at 400 µmol photons m⁻² s⁻¹ for 5 min. Amounts of thylakoid ATP synthase in P_psbA2:synatp1 and P_psbA2:Δsynatp1 were analyzed by immunodetection. Thylakoids were isolated as described (Gandini et al., 2017), adjusted to 2 µg Chl a (which corresponds to 100% in Supplemental Figure S8), and subjected to Tricine-SDS-PAGE and immunodetection assays as outlined in the main text.

Accession numbers

AtCGL160 (AT2G31040, UniProt identifier O82279), AtCGLD11 (At2G21385), AtALB4/AtSTIC1 (AT1G24490), SynAtp1 (Sll1321)

Supplemental data

The following materials are available in the online version of this article.

Supplemental Figure S1. Multiple alignment of the N-terminal portions of CGL160 sequences identified in species belonging to the green lineage.

Supplemental Figure S2. Screening of P_35S:AtCGL160, P_35S:AtCGL160C, and P_35S:AtCGL160N plants.

Supplemental Figure S3. Immunodetection of AtCGL160 in Col-0, atcgl160-1, P_35S:AtCGL160, P_35S:AtCGL160N, P_35S:AtCGL160C, and atcgld11-1 plants.

Supplemental Figure S4. Quantification of grana number and height in Col-0, atcgl160-1, and P_35S:AtCGL160C green leaf sector samples.

Supplemental Figure S5. Characterization of the stromal CF₁ complex in atcgl160-1 plants.

Supplemental Figure S6. Immunoblot analysis of AtCGL160 co-immunoprecipitation assays.

Supplemental Figure S7. Quantification of thylakoid-bound CF₁-β subunits in atalb4-1 Arabidopsis mutant lines.

Supplemental Figure S8. Interaction of BD-CF₁-β and AD-AtCGL160N in yeast two-hybrid assays.

Supplemental Figure S9. Lack of segregation in a synatp1 knockout strain of Synechocystis sp. PCC 6803 (Synechocystis) indicated an essential function for SynAtp1.

Supplemental Figure S10. The endogenous promoter of the atp1 operon in Synechocystis could be functionally replaced by the strong psbA2 promoter (P_psbA2).

Supplemental Table S1. AtCGL160 co-immunoprecipitation experiments.

Supplemental Table S2. Primers used in this study.

Supplemental Data Set S1. Statistical analyses.

Acknowledgments

We thank Tim Scheibenbogen, Michael Berger, and Tanja Neufeld for technical assistance with Yeast-two-hybrid experiments, Tim Dreißig for technical assistance with heterologous expression of AtCGL160N, and Paul Hardy for critical comments on the manuscript.

Funding

This work was funded by the German Science Foundation (DFG, Research Unit FOR2092, project number 239484859, grant GE 1110/9-1 for S.G. and RU 1945/2-1 for T.R.).

Conflict of interest statement. The authors declare no conflicts of interest.

B.R. and T.R. designed research. B.R., L.R., G.M., S.G., and T.R. carried out experiments. B.R., D.L., and T.R. prepared the article. T.R. supervised the whole study.

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plcell) is: Thilo Rühle ([email protected]).