One-helix protein 2 is not required for the synthesis of photosystem II subunit D1 in Chlamydomonas

Abstract

In land plants and cyanobacteria, co-translational association of chlorophyll (Chl) to the nascent D1 polypeptide, a reaction center protein of photosystem II (PSII), requires a Chl binding complex consisting of a short-chain dehydrogenase (high chlorophyll fluorescence 244 [HCF244]/uncharacterized protein 39 [Ycf39]) and one-helix proteins (OHP1 and OHP2 in chloroplasts) of the light-harvesting antenna complex superfamily. Here, we show that an ohp2 mutant of the green alga Chlamydomonas (Chlamydomonas reinhardtii) fails to accumulate core PSII subunits, in particular D1 (encoded by the psbA mRNA). Extragenic suppressors arose at high frequency, suggesting the existence of another route for Chl association to PSII. The ohp2 mutant was complemented by the Arabidopsis (Arabidopsis thaliana) ortholog. In contrast to land plants, where psbA translation is prevented in the absence of OHP2, ribosome profiling experiments showed that the Chlamydomonas mutant translates the psbA transcript over its full length. Pulse labeling suggested that D1 is degraded during or immediately after translation. The translation of other PSII subunits was affected by assembly-controlled translational regulation. Proteomics showed that HCF244, a translation factor which associates with and is stabilized by OHP2 in land plants, still partly accumulates in the Chlamydomonas ohp2 mutant, explaining the persistence of psbA translation. Several Chl biosynthesis enzymes overaccumulate in the mutant membranes. Partial inactivation of a D1-degrading protease restored a low level of PSII activity in an ohp2 background, but not photoautotrophy. Taken together, our data suggest that OHP2 is not required for psbA translation in Chlamydomonas, but is necessary for D1 stabilization.

Introduction

Oxygenic photosynthesis, carried out by cyanobacteria and photosynthetic eukaryotes, is based on the light-driven electron transfer from water to NADPH, involving three complexes located in the thylakoid membrane and operating in series: photosystem II (PSII), cytochrome b₆f (Cytb₆f), and photosystem I (PSI). PSII is a multi-subunit complex, whose core complex is made up of 20–23 subunits: the reaction center proteins D1 (PsbA), D2 (PsbD), Cytb₅₅₉ (subunits PsbE and PsbF), and PsbI are encoded in the chloroplast genome by the psbA, psbD, psbE/F, and psbI genes, as are the core antenna CP47 (PsbB) and CP43 (PsbC) and the additional subunits PsbH, PsbJ, PsbK, PsbL, PsbM, PsbT, and PsbZ (reviewed in Gao et al. (2018)). In eukaryotes, nucleus-encoded subunits complete the complex and build its peripheral chlorophyll (Chl) a/b containing light-harvesting antenna complexes (LHCs). Chl a and its derivative pheophytin a are essential components of the intra-PSII electron transfer chain that allows the generation of a stable charge separated state after light-capture. The pigments and redox-active cofactors are scaffolded mainly by the homologous D1 and D2 polypeptides, which also provide Tyr residues involved in re-reduction of P680⁺ and residues liganding the manganese (Mn)-cluster where water-splitting occurs.

Given the tight folding of the enzyme and the high chemical reactivity of Chl cation radicals that can form from the light-excited states, it has long been hypothesized that specific mechanisms must coordinate the synthesis of the reaction center proteins, their intra-membrane insertion, and their association with pigments. Over the years, the biogenesis of PSII has been dissected into a series of discrete assembly steps (reviewed in Nickelsen and Rengstl (2013)). Precursor D1 (pD1), co-translationally inserted into the membrane, binds to PsbI and specific assembly factors, forming a D1 module that will associate with a D2 module formed between D2, Cytb₅₅₉, and other factors (Komenda et al., 2012; Knoppová et al., 2022; Maeda et al., 2022). The reaction center (RC) subcomplex will first integrate CP47 (forming RC47), then CP43 associated with other small subunits, forming the monomeric PSII core. Dimerization ensues, as well as association with LHCs integrated into the membrane by the chloroplast signal recognition particle pathway. On the lumenal surface of the enzyme, cleavage of the C-terminal extension of pD1 and light-driven assembly of the Mn-containing cluster allow formation of the water-splitting PSII enzyme.

Numerous proteins have been found to catalyze various steps of this pathway, both during de novo assembly and during repair after photoinhibition (Heinz et al., 2016; Lu, 2016). Some act by regulating gene expression, in particular translation of psbA which codes for the rapidly turned over D1. Others act as a chaperone, binding an assembly intermediate until the next step can be completed. In this category, particular attention has been paid to proteins that could mediate the assembly of the cofactors. It is believed that Chl molecules are presented to the acceptor PSII subunits by specific carrier proteins that precisely mediate their insertion at the proper position, at the proper stage of assembly.

In Arabidopsis (Arabidopsis thaliana), a complex consisting of two conserved one-helix proteins (OHP1 and OHP2) and a protein related to short-chain dehydrogenase/reductases, high chlorophyll fluorescence 244 (HCF244), has been found to be essential for Chl integration into PSII, or for protection of the newly synthesized Chl-associated D1 during formation of the RC complex (Hey and Grimm, 2018; Myouga et al., 2018; Li et al., 2019; Maeda et al., 2022). In combination with OHP1, OHP2 is able to bind carotenoids and Chl, the latter via specific residues of a Chl-binding motif (Hey and Grimm, 2020), and the reconstituted heterodimer has unique photoprotective properties (Psencik et al., 2020). The OHP1/OHP2/HCF244 Complex (which we will hereafter call “OHC”) is homologous to a complex of similar function described in cyanobacteria, where the relative stability of the assembly intermediates in core subunits mutants has allowed a fine dissection of the pathway (Komenda et al., 2012). OHP1 and OHP2 resemble cyanobacterial high light-inducible proteins (HLIPs) encoded by the hliA-D genes (Komenda and Sobotka, 2016), in that they all present a single transmembrane domain showing the key residues for binding Chl (Engelken et al., 2010). While OHP1 clearly derives from HLIPs, the origin of OHP2 is less clear (Engelken et al., 2010). HliC and HliD have been found, together with the HCF244 homolog hypothetical open reading frame 39 (Ycf39), to associate with the D1 module and then to the RC subcomplex after binding of the D2 module (Chidgey et al., 2014; Knoppová et al., 2014). HliD can bind Chl and β-carotene (Staleva et al., 2015). The Chl synthase ChlG co-immunoprecipitates with HliD/Ycf39 and it is believed that the complex can deliver newly synthesized Chl to D1 and/or D2 during or early after their translation (Chidgey et al., 2014; Proctor et al., 2020). For this process, Chl could also be scavenged from other Chl-binding proteins upon their degradation.

The phenotype of land plant mutants has revealed an additional function of the OHC, namely to regulate psbA translation. Arabidopsis ohp1, ohp2, and hcf244 mutants show strongly reduced PSII levels, while only weak or indirect effects are observed on PSI accumulation (Link et al., 2012; Beck et al., 2017; Myouga et al., 2018; Li et al., 2019). In the three mutants, the other two partners of the OHC are undetectable or dramatically reduced, suggesting that formation of the complex is required for their stabilization (Beck et al., 2017; Li et al., 2019; Hey and Grimm, 2020). Interestingly, ohp1 and ohp2 mutants from Arabidopsis as well as an hcf244 mutant from maize (Zea mays) show a strong reduction (resp. 7-fold, 12-fold and 11-fold) of the ribosome footprint (RF)-Seq signal on psbA, indicating near complete inhibition of translation (Chotewutmontri et al., 2020). Chotewutmontri and Barkan (2020) proposed that the OHC-bound D1 negatively regulates psbA translation by preventing the stroma-exposed HCF244 and/or OHP2 N-terminal domain from activating translation initiation. psbA translation also requires high-chlorophyll fluorescence 173 (HCF173), another short-chain dehydrogenase/reductase, found to bind the psbA 5′ UTR, while S1 RNA-binding ribosomal protein 1 (SRRP1), an S1-domain protein that also binds psbA, would repress translation (Schult et al., 2007; Link et al., 2012; McDermott et al., 2019; Watkins et al., 2020). During biogenesis or photoinhibition, nascent D1 is inserted into the nascent/repairing RC, and its release from the OHC would trigger light-induced psbA translation. High chlorophyll fluorescence 136 (HCF136), homologous to hypothetical open reading frame 48 (Ycf48) in cyanobacteria, is a lumenal protein found associated with the RC, the RC-OHC, with tagged OHP1 and necessary for assembly of pD1 into the RC (Plücken et al., 2002; Komenda et al., 2008; Myouga et al., 2018). Consistent with an inability to insert pD1 into the OHC, hcf136 mutants exhibit constitutively high psbA ribosome occupancy in light and dark (Chotewutmontri and Barkan, 2020). Such a tight coupling between assembly and translation is reminiscent of a process described as control by epistasy of synthesis (CES) in Chlamydomonas (Wollman et al., 1999; Minai et al., 2006). In the PSII CES cascade, unassembled D1, as produced for example in the absence of D2, represses its own translation initiation. It was thus of particular interest to determine whether such an OHC-dependent translational regulation exists in Chlamydomonas.

Many other proteins of the PSII assembly pathway are conserved between cyanobacteria, land plants, and Chlamydomonas, but their detailed working mechanism has only rarely been investigated in the alga (Spaniol et al., 2021). ALBINO 3 (ALB3), which belongs to a conserved family of protein integrases also found in bacteria and mitochondria, plays a major role in the insertion of integral thylakoid membrane proteins in general, and serves as a hub for many other factors namely HCF136, but also low PSII accumulation 1 (LPA1) and probably LPA2 and LPA3 (reviewed in Plöchinger et al. (2016)). The protein Rubredoxin 1 (RBD1) has been proposed to participate in the protection of PSII intermediate complexes from photo-oxidative damage during de novo assembly and repair, to promote proper folding of D1, possibly via delivery or reduction of the non-heme iron, but also to activate the translation of the psbA mRNA (García-Cerdán et al., 2019; Calderon et al., 2022; Che et al., 2022). The cyanobacterial homolog rubredoxin A (RubA), like Ycf48, has been shown to be a component of the initial D1 assembly module (Kiss et al., 2019).

Chl insertion into RC subunits is not only required during PSII biogenesis, but also during repair after photoinhibition. The D1 protein is the primary target of photodamage, which includes a reversible component, repaired in the absence of protein translation, and an irreversible component that requires degradation of D1 and de novo synthesis (for a review see, e.g. Theis and Schroda, 2016). FtsH (referring to the Escherichia coli enzyme Filamentous temperature sensitive H) is a multi-subunit ATP-dependent thylakoid membrane metalloprotease, combining chaperone and peptidase domains. It has been shown to be a major player in the degradation of D1 and is thus essential for repair after photoinhibition (Lindahl et al., 2000; Silva et al., 2003; van Wijk, 2015). In Chlamydomonas, ftsH1 mutants show light-sensitivity, an accumulation of damaged D1 and its degradation products, and defects in PSII repair (Malnoë et al., 2014).

In this study, we show that a null mutant of OHP2 in Chlamydomonas is entirely devoid of PSII. The primary defect is a lack of D1 accumulation which can be fully restored by complementation of the mutant with Arabidopsis OHP2. Ribosome profiling and pulse labeling experiments show that this defect is not caused by an arrest of psbA translation, but by a reduced stability of the nascent D1 protein. Targeted proteomics show that HCF244 accumulates to ∼25% of the wild-type (WT) level in the ohp2 mutant, explaining this partial uncoupling between the association of D1 with Chl and the initiation of translation on psbA. In the mutant, D1 degradation appears in part mediated by the FtsH protease. An intriguing specificity of Chlamydomonas OHP2 is the high-frequency suppression of the PSII-less phenotype in the mutant.

Results

Identification of an ohp2 mutant that reveals a complete loss of PSII activity and a reduced Chl content

By screening a collection of insertional mutants of Chlamydomonas (Houille-Vernes et al., 2011), we obtained a PSII-deficient mutant carrying an insertion of a transposon of Chlamydomonas 1 (TOC1) transposon (Day et al., 1988) within exon 3 of the OHP2 gene (Cre06.g251150, Figure 1A, see Supplemental Figures 1–3 for details). Accordingly, Southern blot analysis of the ohp2 mutant showed integration of a large DNA fragment of ∼6 kb into OHP2 (Figure 1B).

Figure 1.

Identification and complementation of the mutation in OHP2. A, The genomic region of OHP2 (Cre06.g251150) on chromosome 6. Grey boxes indicate exons, smaller grey arrow and box 5′ and 3′ UTRs, respectively, and light grey boxes introns. For simplicity, upstream and downstream loci are not displayed. The probe used in (B) is shown as black bar above the gene model. The insertion site of the TOC1 transposon in OHP2 in the mutant is indicated by an open triangle, while the filled triangle marks the position of a structural variant already in the recipient strain. Note that the position of HindIII (H) and PstI (P) restriction sites is based on the mt + genome assembly and could differ in the investigated mt- strains. B, Southern blot analysis of genomic DNA from the Jex4 recipient strain (WT) and the ohp2 mutant, using enzymes HindIII or PstI. C, Photosynthetic properties of the recipient (WT), ohp2 and ohp2 complemented strain. Fluorescence induction kinetics (upper panel), measured under illumination at 135 µE m⁻² s⁻¹, followed by a saturating pulse (arrow) and dark relaxation. Fluorescence intensity is normalized to the F_m value. Electrochromic shift at 520 nm (lower panel), measured in the absence or presence of DCMU and hydroxylamine. Values are normalized to the signal after the saturating flash. D, Growth test. Cells were spotted onto acetate-containing (TAP) or MIN and grown for 6 days under higher light (HL) at 100 µE m⁻² s⁻¹ or low light (LL) at 30 µE m⁻² s⁻¹.

The OHP2 gene of Chlamydomonas encodes ONE-HELIX PROTEIN 2 (OHP2) which belongs to the Chl a/b binding protein superfamily (Engelken et al., 2010). OHP2 consists of 144 amino acids with a predicted N-terminal chloroplast transit peptide (cTP) of 26 amino acids, resulting in a mature protein of 13.4 kDa. Transformation of the strain UVM4 with a construct where the OHP2 cTP is fused to green fluorescent protein (GFP) (construct 2 in Supplemental Figure 4; Neupert et al., 2009) confirmed localization to the chloroplast (Supplemental Figure 5).

As summarized in Supplemental Table 1, the mutant completely lacks PSII, while PSI and Cytb₆f activities appear normal. In fluorescence induction experiments PSII quantum yield (F_v/F_m) was negative because of the small initial drop in fluorescence (Figure 1C, upper panel). This effect, typical of mutants completely lacking PSII, is due to chemical quenching by plastoquinone brought about by the oxidation of plastoquinol by Cytb₆f and PSI. Similarly, the PSII/PSI ratio, measured from the amplitude of the initial electrochromic shift (ECS) signal after a flash in the absence and presence of the PSII inhibitors 3-(3,4-dichlorophenyl)-1,1-dimethylurea (DCMU) and hydroxylamine, indicated loss of PSII charge separation (Figure 1C, lower panel). The small effect of the inhibitor treatment on the ohp2 mutant was ascribed to hydroxylamine and was also observed in PSII-null mutants such as the psbA deletion-mutant strain FuD7 (Bennoun et al., 1986). As expected from these experiments, the mutant was not able to grow phototrophically (Figure 1D). Together, these results revealed that PSII activity is completely abolished in the mutant.

When cultivated as liquid culture under mixotrophic conditions, the ohp2 mutant displayed a lighter-green coloration than the WT. Determination of Chl a and b content showed that, on a per-cell basis, the ohp2 strain accumulated only about 58% of the total Chl present in the WT. The Chl a content decreased more than the Chl b level, leading to a decreased Chl a/Chl b ratio in ohp2 (Supplemental Table 1), typical of photosynthetic reaction center mutants.

Complementation studies of the ohp2 mutant with Chlamydomonas and Arabidopsis OHP2

Complementation studies were carried out using a plasmid harboring a paromomycin (Pm) resistance cassette along with the OHP2 sequence. First, we used the Chlamydomonas full length OHP2 CDS 3′-terminally fused to a hemagglutinin (HA)-tag encoding sequence, placed under control of the strong nuclear PSAD promoter (construct 3 in Supplemental Figure 4). Transformants first selected for Pm resistance were tested for photoautotrophic growth on minimum medium under high or low light. As can be expected from non-homologous recombination events, not all but approximately three quarters of the analyzed clones exhibited restored photoautotrophic growth shown exemplarily for one clone (ohp2:OHP2-HA in Figure 1D). Full restoration of photosynthetic capacities was accompanied by a largely restored Chl accumulation (Figure 1C; Supplemental Table 1). These results confirm that the PSII deficient phenotype of the mutant is due to inactivation of OHP2.

Mature OHP2 contains a single transmembrane helix including highly conserved residues required for Chl binding (Figure 2A). Based on topology experiments of Li and colleagues (2019), the N-terminus faces the chloroplast stroma, while the C-terminus is located on the lumenal side. Multiple sequence alignment and hydropathy analysis indicate a well-conserved hydrophobic stretch (HS) of nine residues near the C-terminus which is too short to span the lipid bilayer and seems to be missing in cyanobacterial HLIPs like HliD or HliC (compare Figure 2A; Supplemental Figure 6). This HS might be an evolutionary leftover from a second helix of the ancestral two-helix stress-enhanced proteins (SEP) from which OHP2 potentially originates (Andersson et al., 2003; Heddad et al., 2012; Beck et al., 2017). To test the functional role of this C-terminal hydrophobic region we transformed the mutant with a construct encoding a truncated OHP2 protein (pBC1-CrOHP2-ΔHS-HA, construct 4 in Supplemental Figure 4) where the last 14 amino acids are replaced by the HA-tag. None of the 30 Pm-resistant transformants tested showed photoautotrophic growth on minimal medium, and none showed accumulation of the truncated OHP2-HA by immunoblotting (Figure 2B). These results suggest that even though cyanobacterial orthologs do not have the HS, the last 14 amino acids are essential for OHP2 stability in Chlamydomonas.

Figure 2.

Protein sequence alignment and complementation analysis. A, Sequence alignment of predicted mature one-helix proteins and HLIPs. The multiple sequence alignment from ClustalW (Thompson et al., 2002) was manually edited and displayed with Genedoc (Nicholas et al., 1997). Asterisks mark two residues described to be important for Chl binding (Kühlbrandt et al., 1994), the red triangle the position of the TOC1 insertion in the Chlamydomonas mutant. The N-terminal chloroplast transit peptides predicted by TargetP-2.0 (Nielsen et al., 1997; Emanuelsson et al., 2000) for all eukaryotic proteins shown, are not included in the alignment. B, The C-terminal HS is required for restoration of photoautotrophy. Top: growth tests on TAP, TAP + Pm and MIN plates of transformants carrying the pBC1-CrOHP2-ΔHS-HA construct, shown next to the recipient, ohp2 and complemented strains. Bottom: immunoblot analysis with antibodies to the HA tag and to RbcL as a loading control. Throughout the paper, samples run on the same gel but not in adjacent lanes are separated by a vertical black line. C, Arabidopsis OHP2 complements the Chlamydomonas mutant. Four representative transformants out of 20 are shown. F_v/F_m values are indicated below the panel. Immunoblot analysis (bottom panel) with antibodies to the HA tag and to D1.

The N-terminal region of the mature OHP2 proteins is poorly conserved and similarity between the Chlamydomonas and Arabidopsis proteins starts only at residue 65 of the mature Chlamydomonas OHP2, with an overall identity of 58% (compare Figure 2A). To test whether the Arabidopsis protein could complement the Chlamydomonas mutant, we used a construct (pBC1-TP_CrOHP2-AtOHP2-HA, construct 5 in Supplemental Figure 4) fusing the predicted cTP of OHP2 from Chlamydomonas with a codon-adapted sequence encoding the mature part of AtOHP2 (Supplemental Figure 7). After transformation of the Chlamydomonas ohp2 mutant and selection for Pm resistance, 17 of the 20 clones (85%) showed restoration of photoautotrophy and PSII activity (Figure 2C). For all of the immunologically analyzed clones, a protein with the expected size of ∼15 kDa was detected using the α-HA-tag antibody, and D1 proteins accumulated to WT level (Figure 2C).

Reduced PSII subunit accumulation and synthesis in the ohp2 mutant

As judged from the analysis of photosynthetic properties of the ohp2 mutant, OHP2 appears exclusively involved in PSII function. Accordingly, immunoblot analyses revealed no change in the accumulation of the large subunit of Rubisco (RbcL), the PSI reaction center protein PsaA or Cytf of the Cytb₆f complex (Figure 3A). In contrast, the PSII core proteins D1 and D2 were below the detection limit. Typical for mutants lacking D1 or D2, the core antenna protein CP43 accumulated to some extent; to about 12% of the level in the WT (Figure 3A). Complementation of ohp2 with the OHP2 cDNA fully restored protein accumulation in the strain ohp2:OHP2-HA (Figure 3A).

Figure 3.

Drastically diminished D1 protein accumulation and synthesis in the ohp2 mutant. A, Immunoblot analysis of photosynthesis-related chloroplast proteins. Total proteins (30 µg) from indicated strains were separated by 12% (α-D1, α-PsaA, α-Cytf, α-CP43, α-D2) or 15% (α-HA, α-RbcL) denaturing polyacrylamide gels and analyzed by the antibodies indicated. B, Northern blot analysis of photosynthesis-related chloroplast transcripts. Membranes were hybridized with probes specific for psbA and psbD, and for rbcL as a loading control. nac2 and FuD7 mutants were employed as negative controls for psbD or psbA mRNA accumulation, respectively. C, Synthesis of chloroplast-encoded PSII subunits. For ohp2, two independently grown cultures (#1, #2) were analyzed. D, 2D BN-PAGE analysis of photosynthetic protein complexes. Photosynthetic complexes were detected using the indicated antibodies. The positions of major PSI and PSII complexes are designated. sc: supercomplex.

Northern blot analysis showed no substantial alteration of psbA and psbD transcript accumulation in ohp2 compared to in the WT (Figure 3B). To examine the synthesis of chloroplast-encoded PSII subunits, ¹⁴C pulse chase labeling experiments were carried out in the presence of cycloheximide (Figure 3C). Two independently grown ohp2 cultures revealed a reduction of radiolabeled CP47, and to a lower extent of CP43 and D2 proteins, in comparison to the WT. However, the most substantial effect was observed for the D1 protein for which no incorporation of the radiolabel was detected, neither during the pulse nor after the chase. This suggests that the primary defect in the ohp2 mutant is either a deficiency in D1 synthesis or a very fast degradation of newly synthesized D1. As discussed below, the effects on other PSII subunits may be ascribed to assembly-dependent translational control (CES; Choquet and Wollman, 2009) or reduced stability of the unassembled subunits.

Our efforts to raise a functional antibody against recombinant OHP2 having failed, we used the ohp2:OHP2-HA strain, along with the WT and ohp2, to examine the assembly of PSII and the location of OHP2. Crude membrane fractions were solubilized and subjected to 2D-BN/SDS-PAGE (Figure 3D). All PSI-related complexes, including PSI core and supercomplexes, were assembled in the ohp2 mutant. The slightly higher ratio of PSI supercomplexes to PSI core complexes observed in the ohp2 mutant could be due to the lower Chl content leading to less harsh solubilization. All PSII complexes formed in the WT were below the detection limit in the mutant, but restored in the complemented ohp2:OHP2-HA strain. Interestingly, different from Arabidopsis, where AtOHP2 was detected as a distinct signal at the size of a ∼150 kDa PSII RC-like complex (Li et al., 2019), the HA-tagged OHP2 protein in our complemented strain was mainly detected in the low molecular range, reaching up as a weak smear to the first monomeric CP43-less PSII RC47 subcomplex (Figure 3D). This suggests that in Chlamydomonas the putative OHC does not withstand the solubilization conditions used for BN-PAGE. Accordingly, we were able to pull-down OHP2, using the HA-epitope as a bait, but not any reliable interactants.

Ribosome profiling reveals that D1 is still translated in the absence of OHP2

To obtain a quantitative image of chloroplast translation in the ohp2 mutant, we employed a previously established targeted chloroplast ribosome profiling approach (Trösch et al., 2018). Ribosome footprints (RF) from WT and mutant were extracted and analyzed by hybridization to highly tiled microarrays covering all open reading frames of the chloroplast genome in a 10-codon resolution. In parallel, total RNA was isolated, fragmented and detected by the same approach. Direct correlation between the three biological replicates showed a high reproducibility both for determining RNA accumulation (r > 0.96) and translation output (r > 0.93), calculated for each CDS by averaging all probe intensities for RNA and RFs, respectively (Supplemental Figure 8A; Supplemental Dataset 1). Direct plotting of RNA intensities showed highly comparable abundances between the WT and the mutant (r = 0.98, Figure 4A). This suggests that the lack of OHP2 is not causing any obvious transcription or mRNA stabilization defects. In contrast, translational output revealed clear differences between mutant and WT cells and hence, reduced correlation (r = 0.9, Figure 4A).

Figure 4.

Ribosome profiling of chloroplast translation reveals protein synthesis defects in the ohp2 mutant. A, The average mRNA (blue) and RF (dark grey) abundances are plotted in log₁₀ scale for ohp2 mutant versus WT. PSII transcripts displaying altered translation in the mutant are indicated. Pearsons’s r-value and P-value are given in nEm non-superscript format for n•10 m. B, Relative average transcript abundances (RNA), translation output, and translation efficiency plotted as heat map (ohp2 versus WT) in log₂ scale. PSII subunits that are further discussed are highlighted with an arrow. C, Normalized RF intensities plotted as mean log₂ ratios between ohp2 mutant and wild type (WT) along the psbA and psbB/T/N/H CDS. Error bars denote standard deviation between the three biological replicates. The trans-membrane segments (TMS) are indicated in green. D, The CES cascade of PSII. When newly synthesized CES polypeptides, like D1 and CP47, cannot assemble, they repress the translation initiation of their encoding mRNA as described before by Minai et al. (2006). Reduced translation of psbH may indicate that it represents a further component of the CES cascade.

Unexpectedly, averaged RF intensities of psbA showed only a mild reduction in the mutant (Supplemental Dataset 1). RF abundance was reduced by 1.6-fold (±0.16), which combined with a slight reduction in mRNA abundance, led to a practically unchanged translation efficiency (Figure 4B). For a more detailed view, we also plotted ribosome occupancy (log₂ of the ohp2/WT ratio) over the psbA ORF (Figure 4C, left panel). A more pronounced reduction (up to 4-fold) in ribosome occupancy in ohp2 can be observed over the first half of the ORF (∼160 codons, up to the third transmembrane segment of nascent D1). In the second half, only mild reduction or even a slight increase in ribosome occupancy could be detected. For other PSII subunits, the most pronounced effect was the reduced translation of CP47 (PsbB), PsbH, and PsbT, with up to 3-fold lower translation output for psbB (Figure 4B and Supplemental Figure 8B). This observation agrees with the hierarchical CES cascade contributing to the biogenesis of PSII in Chlamydomonas, where the presence of D2 is required for high-level translation of D1, which in turn is a prerequisite for efficient translation of CP47 (Figure 4D; reviewed in Choquet and Wollman, 2009). When assembly of D1 and CP47 subunits is compromised, the unassembled proteins repress the translation initiation of their encoding mRNA. In accordance with previous investigations of PSII mutants (de Vitry et al., 1989), we observed no effect on translation of the second inner antenna protein CP43, encoded by the psbC mRNA, confirming that its rate of synthesis is not dependent on the assembly with other PSII subunits.

Interestingly, also PsbH, which only recently has been hypothesized to be part of the CES cascade downstream of D1 (Trösch et al., 2018), exhibited a clearly reduced translation output in ohp2, thus further supporting its role as a CES subunit (Figure 4, A, B, and D). A comparable effect was seen for the translation of psbT. Remarkably, the three affected genes are encoded in the psbB-T-H operon. While the mRNAs were only marginally reduced, a clear and relatively even reduction of ribosome occupancy was seen for all ORFs (Figure 4C, right panel). It is noteworthy that also psbN, which is positioned between psbT and psbH in antisense orientation, exhibits a reduced translational efficiency while the mRNA levels slightly increased. PsbN acts as an assembly factor for PSII but is not part of the final complex (Knoppová et al., 2022).

In contrast, the psbD-encoded D2, while maintaining constant mRNA abundance, showed increased ribosome occupancy over the entire ORF in ohp2 (Figurs 4, A and B; Supplemental Figure 9). As D2 is the dominant CES subunit (Figure 4D), this likely points to a compensatory effect caused by the lack of functional PSII. Similarly, other translational alterations in ohp2, like the reduction of ribosome RFs for the large Rubisco subunit RbcL or an upregulation of all chloroplast-encoded ATP synthase subunits (Supplemental Figure 8), may rather be secondary effects caused by a diminished photosynthetic electron flow and/or energy limitation.

Proteomic analysis reveals the presence of HCF244 in the ohp2 mutant

For a quantitative and comprehensive view on the proteome composition in the ohp2 mutant, proteomic shot-gun analysis was conducted on whole cell fractions (Figure 5A, Supplemental Dataset 2). The analysis additionally included the M-Su1 strain, an ohp2 derivative in which the PSII phenotype was genetically suppressed. Such photoautotrophic suppressor strains occurred at a high frequency (approximate rate: 5.5 × 10⁻⁶) under conditions selecting for photoautotrophic growth. Illumina sequencing and Southern blot analysis of photoautotrophic ohp2 derivatives showed that they still contained the TOC1 transposon in OHP2 (Supplemental Figures 2 and 10A), indicating that the PSII phenotype was suppressed by a second site mutation. The suppressor strains showed partially restored photosynthetic parameters (F_v/F_m ratios of 0.46–0.64), a re-accumulation of PSII subunits indicated by restored D2 levels, and the ability to grow photoautotrophically in liquid medium under strong illumination (Supplemental Figure 10, B–D). The M-Su1 strain thus allows us to distinguish the direct effects of the lack of OHP2 from those linked to PSII loss. In addition, LC–MS analysis was carried out on membrane fractions of WT, ohp2 and the suppressed strain (Figure 5B; Supplemental Dataset 3). In these two analyses, reliable quantification was achieved for 3,065 and 1,127 proteins, respectively, with very high reproducibility between the biological replicates (Supplemental Figure 11A). The complemented strain ohp2:OHP2-HA was included as a control, and showed that most of the alterations observed in mutant cells were reverted by complementation (Supplemental Figure 11B). The most obvious defect observed in the ohp2 strain, apart from the expected absence of OHP2 itself, was a severe depletion of the majority of PSII proteins. The most dramatic effect was on D1 showing a ∼200/640-fold reduction in whole cell and membrane fractions, respectively. However, the detection in the ohp2 mutant of five D1 peptides arising from the C-terminal third of the protein supports the notion, based on the results of the ribosome footprinting, that the psbA CDS is translated over its entire length. For the other large PSII subunits D2 and CP47, the depletion was less pronounced. The least affected subunit was CP43, in accordance with results obtained with other PSII mutants (de Vitry et al., 1989). The small molecular weight subunits of PSII were difficult to assess in these experiments, but we noticed a massive depletion of PsbH in the mutant (Supplemental Datasets 2 and 3), in agreement with the ribosome profiling results. Other PSII subunits were severely depleted, at least at the membrane level, including the recently discovered algae-specific subunit PBAS1 (= PBA1, Putatively Photosystem B Associated 1; Spaniol et al., 2021). Among the extrinsic lumenal oxygen-evolution enhancer (OEE) subunits, no significant change was observed in whole cells, but OEE1 was depleted in membranes, in line with previous reports (de Vitry et al., 1989).

Figure 5.

Altered proteome composition in the ohp2 mutant and suppressor strain. Volcano plots representing the relative proteome changes of ohp2 mutant versus WT (left) and suppressor versus WT (right) in (A) whole-cell lysates and (B) membrane fractions. Mean fold change of LFQ values (in log₂) is plotted on the x-axis, P-values (in −log₁₀) are plotted on the y-axis. Dark grey dots show proteins that are significantly different with FDR < 0.05 and S₀ = 1. Proteins of PSI, PSII, the chloroplast ATP synthase, and proteins involved in Chl biogenesis are marked in color as depicted in the legend. Large colored dots are significantly different. C, Abundance of selected proteins in membrane fractions as determined by targeted mass spectrometry in different strains. WT value is set to 1. Error bars indicate Sd over three biological replicates except when peptides were not detected or below baseline (approximately 5% of WT, as judged from dotp values and fragment coelution profiles).

Of particular interest for us was the presence in the mutant and suppressor strains of OHP1 and HCF244, the partners of OHP2 in the plant OHC. Probably due to its small size and hydrophobicity, OHP1 was detected neither in the whole cell extracts nor in the membrane fractions, even in the WT. To further refine our analysis, the same membrane fractions were also subjected to a targeted proteomics approach, whereby specific peptides were monitored at high resolution. By reducing interferences from the MS1 level, this allows a more reliable quantitation. Five proteins were selected: OHP2, OHP1, HCF244, HCF136, and HCF173, using a total of 25 peptides (Figure 5C; Supplemental Dataset 4). This targeted analysis used two peptides for OHP1 and showed a low signal in membrane fractions of the ohp2 mutant (18% of WT). This suggests that in Chlamydomonas, OHP1 is not completely dependent on OHP2 for its stabilization. The OHP1 signal was not detected in the suppressed strain.

In whole cell extracts, HCF244 accumulated to ∼25% of the WT level in ohp2 and in the suppressor strain (Supplemental Dataset 2). Like OHP2 accumulation, this was largely complemented by expression of the OHP2-HA transgene. Untargeted analysis of the membrane fractions revealed the presence of HCF244 in the WT but not in ohp2 or the suppressed strain. However, the more sensitive targeted analysis, based on eight peptides, revealed the presence of HCF244 in the membrane fractions of the mutant and suppressor strains, with an abundance of 22% and 13% relative to the WT levels, respectively (Supplemental Dataset 4; Figure 5C). Together, these results indicate, that in contrast to Arabidopsis, the presence of OHP2 is not absolutely necessary for HCF244 accumulation, nor for its association with the membrane.

Proteomic analysis of other chloroplast proteins

We also examined the accumulation of other known PSII biogenesis factors, in search of further effects of the lack of OHP2 (Figure 5; top lines in Supplemental Datasets 2–4). Several key players in psbA translation were affected. At the whole cell level, the abundance of HCF173 was not altered in the mutant or suppressor, but targeted analysis of the membrane fractions revealed a ∼3-fold higher level in ohp2, compared to in the WT or the suppressed strains. This suggests a regulation by PSII assembly rather than by the OHC. Interestingly, the abundance of SRRP1 which inhibits psbA translation in the dark, significantly increased in the ohp2 mutant, while the Chlamydomonas-specific psbA translation activator TBA1 (Somanchi et al., 2005) overaccumulated. The abundance of HCF136 in whole cells was not significantly altered, but both the untargeted and targeted analyses of the membrane fractions revealed a significantly higher membrane association specifically in the suppressor strain (∼3-fold when compared to the WT; Figure 5; Supplemental Datasets 2–4). This may reveal a more prolonged association of HCF136 with the RC when the absence of OHP2 slows down the biogenesis process.

The mutant also overaccumulated acclimation of photosynthesis to the environment 1 (APE1), initially described as involved in the adaptation to high light (Walters et al., 2003) and later found as a specific interactant of Arabidopsis OHP1 in pull-down assays (Myouga et al., 2018; Maeda et al., 2022). It was found associated with the RCIIa complex isolated from a Synechocystis sp. PCC 6803 strain lacking Ycf39/HCF244 (Knoppová et al., 2022). Overaccumulation was also observed for the homolog of slr1470, another component of the RCIIa complex (Knoppová et al., 2022). This suggests that absence of OHP2 (and most likely of the entire OHC) in Chlamydomonas leads to prolonged association (and stabilization) of those assembly factors whose interaction with the RC is antagonized by that of the OHC.

No significant effect was observed on the accumulation of PSI RC subunits, but all the LHCI subunits showed increased accumulation in the ohp2 mutant as well as in the suppressor strain. This suggests a direct effect of the lack of OHP2. While subunits of the Cytb₆f complex appeared unaffected in the ohp2 mutant, we noticed an overaccumulation of those of the ATP synthase (Figure 5A, left panel). This agrees with the increased translational output for the ATP synthase ORFs seen in our ribosome profiling analysis (Supplemental Figure 8B). Overaccumulation of the ATP synthase was fully reversed in the complemented strain (Supplemental Figure 11B). We note, however, that the suppressor strain, although photoautotrophic, still showed some overaccumulation of the ATP synthase in its membrane fraction (Figure 5B). The Rubisco subunits RbcL and RBCS2, in contrast, appeared to accumulate at lower levels in the mutant in whole cell fractions (Supplemental Dataset 2). The meaning of this observation is unclear, as their accumulation was not restored to normal in the suppressor strain.

Several enzymes of the porphyrin biosynthesis pathway showed changes in their accumulation or association with the membrane (Supplemental Datasets 2 and 3; Figure 5, A and B). The Chl synthase ChlG, whose cyanobacterial homolog interacts with HliD/Ycf39 (Chidgey et al., 2014; Proctor et al., 2020), remained unaltered in the Chlamydomonas mutant, but other Chl synthesis enzymes showed increased abundance in the membrane fractions. This includes the two subunits of the Mg-chelatase (CHLH, CHLI), the MG protoporphyrin IX methyltransferase (CHLM), the cyclases copper response defect 1 (CRD1) and copper target homolog 1 (CTH1) and the protochlorophyllide a oxidoreductase 1 (POR1). Except for POR1, this increase was not observed at the whole cell level, suggesting that the cell responds to the absence of the OHP2 by stabilizing the interaction between these Chl pathway enzymes and the thylakoid membrane. This effect was largely or completely reversed in the suppressor, pointing to a regulatory mechanism compensating for the impaired accumulation of PSII. In the upstream part of the pathway, shared with heme biosynthesis, several enzymes overaccumulated in whole cells of the mutant (Figure 5A): Delta-aminolevulinic acid dehydratase 1 (ALAD1), uroporphyrinogen III synthase 1 (UPS1), one of the three uroporphyrinogen III decarboxylases (UPD1), and protoporphyrinogen oxidase 1 (PPX1). PPX1 abundance also increased in the membrane fraction. The protein fluorescent (FLU), which is involved in the regulation of the whole pathway (Falciatore et al., 2005) and which was severely reduced in a virus-induced OHP2 gene silenced Arabidopsis line (Hey and Grimm, 2018), was not significantly affected in the Chlamydomonas mutants.

Slowing down D1 degradation in the ohp2 mutant can partially restore light-sensitive PSII activity, but not photoautotrophy

The repair of photodamaged PSII involves the selective proteolytic degradation and replacement of the damaged D1 polypeptide with a newly synthesized one (reviewed in Kato and Sakamoto, 2009; Nixon et al., 2010). The thylakoid protease FtsH plays a major role in the D1 degradation process. In Chlamydomonas, FtsH is composed of two subunits FtsH1 and FtsH2, and the ftsh1-1 mutation strongly compromises oligomerization and proteolytic activity (Malnoë et al., 2014). To address the question whether the FtsH protease is involved in the immediate post-translational degradation of D1 in the ohp2 mutant, we crossed the mutant strain with a strain carrying the ftsH1-1 mutation. In the progeny, the WT, ohp2, ohp2, ftsH1-1, and ftsH-1 genotypes showed distinct fluorescence and growth patterns (Figure 6). When strains were analyzed by spot tests on agar plates, the ftsH1-1 strains showed, as expected, retarded growth and decreased F_v/F_m with increasing light intensity under photoautotrophic as well as under mixotrophic conditions (Figure 6A). The ohp2 progeny did not grow photoautotrophically, except for occasional suppressor clones. Under mixotrophic conditions, they showed no deleterious effect of high light for growth, as is typical of PSII mutants, and had practically no F_v/F_m (Figure 6, A and B). The ohp2 ftsH1-1 strains showed a unique phenotype. Like the ohp2 parent, they were unable to grow on minimal medium at any light intensity tested (Figure 6A). However, their fluorescence induction curves were clearly indicative of the presence of a small amount of PSII, especially in dark-grown cells (Figure 6C). On tris-acetate-phosphate medium (TAP), we did not observe any stimulatory effect of low light on growth (Figure 6A). These results suggest that FtsH is involved in the degradation of the PSII units, possibly abnormal, produced in ohp2 mutants. However, the activity of the low amount of PSII stabilized by attenuation of FtsH is either insufficient or too light-sensitive to allow photoautotrophy. Note that the unknown suppressor locus was not genetically linked to FTSH1 in crosses, and the suppressor strains showed no sequence change in any of the FTSH genes, nor in any other known chloroplast protease gene.

Figure 6.

Mutation of the FtsH protease partially restores PSII activity but not photoautotrophy. A, Growth tests of the progeny of a cross ohp2 × ftsH1-1. Cells were grown on TAP or MIN medium at the indicated light intensity. Spots shown are typical of the indicated genotypes. B, F_v/F_m values for the four genotypes, recorded from the plates in (A). Values are average of 8, 7, 11, and 6 strains for the WT, ftsH1, ohp2 and ohp2 ftsH1 genotypes, respectively. Error bars represent Sd. C, Typical fluorescence induction curves for the four genotypes, recorded with the fluorescence camera on the TAP plates described in A. Fluorescence is normalized to the F_m value (arrow indicates saturating flash).

Discussion

In Chlamydomonas, OHP2 is not necessary for psbA translation

The molecular analysis of an ohp2 knockout mutant from Chlamydomonas revealed a major defect in PSII biogenesis, as indicated by its inability to grow photoautotrophically, the complete loss of PSII activity and the absence of the major PSII subunits, in particular D1 (Figures 1 and 3). No effect was observed on the PSI RC (Figures 1C, 3A, 3D, and 5), as reported for cyanobacterial HLIPs (Komenda and Sobotka, 2016), but at slight variance with land plants, where reductions in PSI subunits and antenna proteins have been reported (Beck et al., 2017; Myouga et al., 2018; Li et al., 2019). In Chlamydomonas, we observed an increased accumulation of LHCI antenna proteins in the mutant (Supplemental Dataset 2). Whether this is related to the observed changes in expression of Chl pathway genes remains to be investigated. In the Arabidopsis ohp1 and ohp2 mutants, the level of PSI RC decreases more than that of LHCI, but this appears to be true for other PSII mutants as well (Li et al., 2019).

In other organisms, a dimer of one-helix proteins (HliC/D in cyanobacteria, OHP1/2 in land plants) has been proposed to mediate the early association of Chl a to the nascent D1 polypeptide. In Chlamydomonas, the expression patterns of OHP2, OHP1, and HCF244 in a variety of conditions are highly similar (Supplemental Figure 12) and on the Phytozome website (https://phytozome-next.jgi.doe.gov/) the latter two appear in each other's lists of best correlated genes, pointing to the existence of a similar complex in the alga. We will, therefore, assume that the OHC exists in Chlamydomonas and is missing in the ohp2 mutant, which prevents normal cofactor insertion into the RC and destabilizes PSII. The main result of our study, compared with those in vascular plants, is that in Chlamydomonas the absence of OHP2 only marginally affects translation of the psbA mRNA. While our ¹⁴C pulse labeling experiments do point to the primary phenotype being a defect in D1 production, they can be interpreted either as a deficiency in psbA translation, or as a very fast degradation of the newly synthesized D1 polypeptide (Figure 3C). Our ribosome profiling experiments strongly support the latter hypothesis, as they show an almost normal abundance of RFs over the CDS (1.6-fold reduction of the signal). Continued translation of the psbA mRNA is observed over its whole length (Figure 4C), further supported by the fact that the proteomic analysis identified several peptides in ohp2 that lie in the C-terminal part of D1.

Note that the two techniques we have used have their limitations. Usually, the absence of a signal in ¹⁴C-pulse labeling in Chlamydomonas is interpreted as loss of translation (e.g. de Vitry et al., 1989). But the ohp2 mutant must be analyzed differently, as the mutation is supposed to affect a co-translational or early post-translational step. In addition, the sensitivity of this approach is limited by the broadness of the D1 band (D originally stands for “diffuse”) and the unavoidable presence of background radioactivity. Ribosome footprinting is more sensitive, even though it is also an indirect proxy for translation initiation. The abundance of RFs over a transcript may also be affected by the dynamics of translation, for example the stalling of ribosomes will increase the probability to generate RFs at this position. It is interesting to note that the reduction in ribosome occupancy in the ohp2 mutant is more pronounced over the first half of the mRNA. Chl attachment to D1 has been proposed in barley (Hordeum vulgare) to be associated with ribosome pausing (Kim et al., 1991, 1994a, 1994b) even though RF experiments in maize have failed to identify such Chl supply dependent pauses (Zoschke and Barkan, 2015; Zoschke et al., 2017). If interaction of the nascent chain with the Chl attachment machinery slows down elongation, then its absence in ohp2 may reduce the density of RFs. Conservatively, we propose that in the absence of OHP2, psbA translation is maintained at a substantial rate, over the whole CDS. However, the degradation of apo-D1 is so fast that full-length D1 remains below detection in pulse labeling experiments, as it is in Western blots (Figure 3, A and C). This statement is backed by the observation that mutations in the FtsH protease and at an unknown suppressor locus can partially or fully restore PSII accumulation, which would be difficult to achieve if OHP2 was required for translation.

The situation is different in land plants, where OHP1, OHP2, and HCF244 are required for the recruitment of ribosomes to the psbA mRNA (Chotewutmontri et al., 2020). The abundance of RFs in the Arabidopsis ohp2-1 knockout mutant was ∼12 times lower than in the WT, a figure probably underestimated by normalization since the psbA CDS itself contributes to a large fraction of the RFs in the WT. Chotewutmontri et al. (2020) convincingly explained the discrepancy with a previous study using polysome analysis (Li et al., 2019), by the difficulty in analyzing psbA polysomes profiles due to the high abundance of the mRNA. They found that the Arabidopsis ohp1-1 and maize hcf244-1/-3 null mutants also fail to translate D1 (6-fold and 10-fold reduction, respectively). No change was observed in the pattern of RFs over psbA in any of the mutants, in line with a defect in translation initiation rather than elongation. All this was in accordance with the role of HCF244 as an essential translation initiation factor for psbA and its total absence in mutants lacking OHP1 or OHP2 (Li et al., 2019). Chotewutmontri and coworkers proposed a model for regulation of psbA translation by light, whereby the presence of D1 in the OHC inhibits the ability of the stroma-exposed components (HCF244 and/or OHP2's stromal tail) to initiate psbA translation (Chotewutmontri and Barkan, 2018, 2020; Chotewutmontri et al., 2020). It must be noted that Chotewutmontri et al. (2020) quantified RFs by Illumina read counting, while we have used hybridization to oligonucleotide probes. But both methods are fully validated, and the results are so different that we cannot invoke a technical bias. Therefore, we conclude that Chlamydomonas lacks the OHP2-dependent control mechanism for psbA translation described in land plants.

We used proteomics analyses to explain this apparent discrepancy in the context of the high conservation of the proteins between algae and land plants (as exemplified by our observation that Arabidopsis OHP2 can complement the Chlamydomonas mutant; Figure 2C). Quantification of OHP1 was difficult because of its small size, but the results of our targeted analysis point to the presence of traces of this protein in the ohp2 mutant (Figure 5C). This, however, should have no effect on the insertion of cofactors: an ohp2 mutation in Arabidopsis is not complemented by overexpression of OHP1, and OHP1 alone does not bind pigments in vitro (Beck et al., 2017; Hey and Grimm, 2020). Studies in land plants have found that while some OHP2 can accumulate in ohp1 null mutants, OHP1 is usually undetectable in ohp2 mutants (Beck et al., 2017; Li et al., 2019). Some variability is seen in these studies, possibly due to the fact that the T-DNA insertion in OHP2 lies within an intron (Beck et al., 2017; Myouga et al., 2018; Li et al., 2019). Similarly, the level of HCF244 reported in Arabidopsis ohp2 mutants varied from nil to very low (Li et al., 2019) and the accumulation of HCF244, even when overexpressed, was found to be limited by that of OHP2 (Li et al., 2019; Hey and Grimm, 2020). Here, our untargeted and targeted LC–MS/MS analyses concur to demonstrate not only that HCF244 accumulates to significant levels in the Chlamydomonas ohp2 mutant (∼25% of WT), but that it can interact with the membrane (Figure 5C). This might be explained by the remaining traces of OHP1 acting as an anchor for HCF244 and stabilizing it, or to an intrinsically higher stability of Chlamydomonas HCF244 when not assembled. We propose that these remaining 25% HCF244 in the Chlamydomonas ohp2 mutant are responsible for the maintenance of psbA translation at an appreciable rate. This does not exclude that an autoregulatory circuit exists in WT Chlamydomonas similar to that described by Chotewutmontri and Barkan (2020) in land plants. In this case, the absence of OHP1/OHP2 in the mutant may prevent the remaining HCF244 from sensing the presence of unassembled D1, leading to enhanced activation of psbA translation. In any event, the strong coupling observed in plants between Chl insertion and psbA translation initiation appears largely broken in Chlamydomonas. A major difference between Arabidopsis and Chlamydomonas is that Chl production depends entirely on light in the former, while the latter can produce Chl in the dark. Thus, the availability of Chl for integration into PSII could be used by Angiosperms to sense light, while algae would have to use other clues. In addition, the translation of psbA appears to mobilize a larger fraction of ribosomes in the plant than in the alga, as judged from RF and pulse-labeling experiments. Arguably, the alga can partly dispense of a regulatory circuit aimed at ensuring that psbA is translated only in the light. Detailed studies in other systems, such as land plants capable of producing Chl in the dark and cyanobacteria, are necessary to decide how strong the coupling needs to be to prevent deleterious effects of uncoordinated D1 synthesis and assembly, in particular during high light stress when D1 needs to be repaired at high rate.

The absence of OHP2 affects translation of several chloroplast transcripts encoding PSII subunits

Interestingly, the ohp2 mutant revealed large changes in the ribosome profiling pattern of PSII transcripts other than psbA. Most striking was the reduced translational output for the CES subunit CP47, encoded by psbB (Figure 4). This observation is in line with previous knowledge on translational regulation by assembly (the CES process) in Chlamydomonas (Minai et al., 2006; reviewed in Choquet and Wollman, 2009). Genetic studies indeed have demonstrated that in the absence of D1, the unassembled CP47 subunit feeds back onto the translation initiation of its own mRNA. Here, assuming that no assembly of CP47 and D1 can occur in the ohp2 mutant, we can quantify this effect. With a 3-fold reduction of RFs over the psbB mRNA in the mutant (log₂ohp2/WT = −1.58), the CES effect appears to be rather dramatic (Figure 4B). Remarkably, the translation of psbH is also reduced in the mutant, in line with a complete lack of detection of PsbH peptides by proteomic analyses (Supplemental Datasets 2 and 3). This supports the identification of PsbH as a component of the PSII CES cascade, as proposed recently by Trösch et al. (2018) (Figure 4D). Early reports in Chlamydomonas describe a role of PsbH in assembly and/or stabilization of PSII (Summer et al., 1997; O'Connor et al., 1998) but the protein's function has not been studied in detail. In Synechocystis, PsbH is associated with CP47 and facilitates D1 processing and incorporation into PSII (Komenda et al., 2005). Assuming a similar position of PsbH in the assembly pathway in Chlamydomonas, its CES-mediated downregulation in the absence of D1 is consistent with that of CP47. In their study, Chotewutmontri et al. (2020) pointed to an increased, rather than decreased, translation of psbB in their ohp1, ohp2, hcf244 and even in hcf173 mutants. The importance of these opposite behaviors between Chlamydomonas and land plants remains to be explored. Noticeably, psbH, compared to other PSII subunits, also appeared as slightly increased in the RF abundancy plots of the ohp2 and hcf244 mutants in Chotewutmontri et al. (2020).

The comparable effect that we observed on psbT translation in our experiments (Figure 4) may be construed as evidence that this small polypeptide also belongs to the PSII CES cascade. It should be noted that all three CES cistrons belong to the conserved psbB-T-H operon and that the tetratricopeptide repeat protein MBB1 is necessary for stabilization of the 5′ end of the psbB-psbT mRNA, as well as the processing and translation of the psbH mRNA (Vaistij et al., 2000; Loizeau et al., 2014). This suggests a role for Chlamydomonas MBB1 in the CES regulation. An unexpected result for the ohp2 mutant in the RF experiment was the 2.8-fold increase in RFs over psbD, encoding D2. As all other D1 mutants studied thus far, like FuD7 or F35, completely lacked D1 translation (Bennoun et al., 1986; Girard-Bascou et al., 1992; Yohn et al., 1996), a stimulation of D2 translation by non-productive psbA translation (as in ohp2) could be envisioned.

OHP2 is not completely essential for PSII biogenesis

Another intriguing characteristic of the Chlamydomonas ohp2 mutant is the high frequency at which the non-photoautotrophic phenotype can be suppressed. A mere plating on mineral medium, without any mutagenic treatment, was enough to generate dozens of extragenic suppressors. Genetic analysis of three of them indicated that the suppressor mutations were tightly linked, but unlinked to the OHP2 or FTSH1 loci. To identify the underlaying mutation, we performed whole genome sequencing as well as direct sequencing of obvious candidate genes, like OHP1, HCF244, LPA2, or psbA. However, in spite of all our efforts, we were unable to identify a plausible causative variant. Also, the proteomic analysis gave no clear hint on the nature of the suppressor mutation. Further supporting the existence of an OHP2-independent pathway for PSII biogenesis, we found that combination of ohp2 with the ftsH1-1 mutation allowed partial recovery of stable PSII charge separation (Figure 6C). This indicates that whatever the exact action of OHP2 on D1 is, it is not completely essential for biogenesis of a functional PSII. A by-pass reaction must be possible, that leads to partial restoration of PSII biogenesis. This pathway would be activated by the suppressor mutation, but it probably already exists in the ohp2 mutant: by slowing down D1 degradation, the ftsH1-1 mutation may allow a normally unstable form of D1 to proceed through this alternative pathway. This pathway may also operate in cyanobacteria, where deletion of all HLIP genes does not prevent PSII biogenesis (Xu et al., 2004).

It is probable that the PSII produced by the OHP2-independent pathway is not fully functional, because the suppressed strains showed some light-sensitivity, and the ohp2 ftsH1-1 double mutant was not photoautotrophic (Figure 6A). On the other hand, the alternative pathway may be inefficient for repair in high light. The ftsH1-1 mutation by itself renders the cell highly sensitive to photoinhibition. Detailed studies comparing PSII purified from these strains may be necessary to answer this question.

Consequences of the absence of OHP2 on other factors involved in D1 biogenesis

Our proteomics data showed changes in the accumulation or membrane association of many known PSII assembly factors. In our working model for the role of the OHC and associated proteins in D1 synthesis (Figure 7), HCF173 binds the 5′ UTR of psbA mRNA to promote its translation in an HCF244-dependent manner (Schult et al., 2007; Link et al., 2012; McDermott et al., 2019; Williams-Carrier et al., 2019). Nascent pD1 protein is co-translationally inserted into the thylakoid membrane where it is stabilized by membrane-associated lumenal HCF136. The OHC is proposed to further stabilize pD1, allowing pigment association and assembly into the RC complex. A striking result of our targeted proteomics analysis was the increased association of HCF173 to the membranes of the ohp2 mutant (Figure 5C; Supplemental Dataset 4). Based on the model of Chotewutmontri et al. (2020) this probably indicates that a larger fraction of the psbA mRNA is found on membrane-associated polysomes. However, this effect is no longer seen in the suppressor, suggesting that it is not a direct consequence of the absence of OHP2, but rather a compensatory mechanism caused by the PSII deficiency. In the ohp2 mutant, unassembled D1 can no longer regulate the activity of HCF244 because the OHC cannot form (Link et al., 2012; Chotewutmontri et al., 2020). This is expected to lead to enhanced translation initiation on the psbA mRNA driven by the residual but fully active HCF244 (Figure 7, middle panel), unless a counteracting mechanism signals that PSII biogenesis proceeds, as in the suppressor. SRRP1, another known interactant of HCF173 and of the psbA mRNA, is believed to act as a repressor of translation (Watkins et al., 2020). It could not be quantitated in the membranes, but interestingly, while abundant in cells of the mutant and WT it was undetectable in the suppressor, even though the gene appeared unaltered. HCF136/Ycf48 is likely promoting the association of the pD1/PsbI subcomplex with a distinct precomplex consisting of D2 and the heterodimeric Cytb₅₅₉, to form the RC (Zhang et al., 1999; reviewed in Nickelsen and Rengstl, 2013). It is required for the assembly of early PSII intermediates, possibly with the concurrent incorporation of Chl and has been found to bind pD1 (but not mature D1) in a split-ubiquitin assay (Meurer et al., 1998; Plücken et al., 2002; Komenda et al., 2008; Knoppová et al., 2014; Myouga et al., 2018; Li et al., 2019; Hey and Grimm, 2020). It was, therefore, striking to observe that association of HCF136 with the membrane (not its overall abundance) was markedly increased in the suppressor compared to in the WT or mutant (Figure 5; Supplemental Datasets 2–4). While genome sequencing rules out HCF136 as being the suppressor locus, its increased abundance in the membranes of the suppressor may indicate the stabilization of a complex comprising pD1, allowing it to associate with Chl via the alternative pathway proposed above (Figure 7, right panel).

Figure 7.

Schematic model for the role the OHC complex and associated proteins in D1 synthesis and first steps of PSII de novo assembly. In the wild type (WT; left panel), likely triggered by HCF244, HCF173 binds the 5′ UTR of psbA mRNA to promote translation initiation. pD1 is co-translationally inserted into the thylakoid membrane. Membrane associated lumenal HCF136 stabilizes pD1 and is involved in RC assembly. The OHC is proposed to insert Chl into pD1 (or into the nascent RC). Model modified from Chotewutmontri and Barkan (2020). Recently reported negative autoregulatory circuits of psbA translation initiation involving the OHC are not displayed here (Chotewutmontri and Barkan, 2020; Chotewutmontri et al., 2020). In ohp2 (middle panel) and suppressor strains (right panel) residual levels of HCF244 may promote psbA translation via interaction with HCF173. Ongoing psbA translation may be further supported by overaccumulation of HCF173. However, in the ohp2 mutant synthesized D1 is rapidly degraded and does not accumulate in the absence of the OHC complex. In the suppressor strains, stabilization of nascent D1 may be accomplished by increased membrane association of HCF136 or other mechanisms to allow assembly of early PSII intermediates. HCF173, HCF136, and HCF244 protein amounts detected by proteomics are indicated by differently sized ovals. Names of cyanobacterial homologs are given in brackets.

Some assembly factors acting downstream of the formation of the RC are also impacted by the absence of OHP2 or the suppressor mutation (see Supplemental Datasets 2 and 3). This includes PSB28 (Zabret et al., 2021), PSB27 (Avramov et al., 2020; Huang et al., 2021; Zabret et al., 2021), PSB29/THF1 (Huang et al., 2013; Zhan et al., 2016; Bec̆ková et al., 2017), and PSB33/TEF5 (Fristedt et al., 2015; Nilsson et al., 2020). The absence of the OHC thus has a profound impact on all steps of PSII biogenesis.

Other effects of the mutation in OHP2 on chloroplast biogenesis

The chloroplast-encoded subunits of the ATP synthase were found to be more actively translated in the ohp2 mutant (Supplemental Figure 8; Supplemental Dataset 1), leading to an increased abundance of all subunits of the complex, including the nucleus-encoded ATPC/D/G (Figures 5, A and B; Supplemental Datasets 2 and 3). Increased levels of chloroplast ATP synthase was recently also observed in complexome profiling of the PSII-deficient lpa2 mutant from Chlamydomonas (Spaniol et al., 2021). This could be involved in maintaining energy balance in the chloroplast: in the absence of PSII, linear electron flow is abolished, but cyclic electron flow remains possible, fueling the proton motive force that drives ATP synthesis.

Our proteomics analysis showed increased accumulation of LHCI subunits in the ohp2 mutant, as well as in the suppressor strain. This suggests that the lack of OHP2 directly affects the biogenesis of PSI. This may be related to the observation that the PSAN subunit, although not affected in whole cells, was undetectable in the membranes of the mutant. In maize, PSAN lies at the interface between the RC and LHCI (Pan et al., 2018) but this subunit was not found in the structure of the Chlamydomonas PSI–LHCI complex and a role in forming a complex between plastocyanin and PSI was proposed (Suga et al., 2019).

Proteomics also revealed a general stimulation of the heme/Chl biosynthesis pathway in the ohp2 mutant (Figures 5, A and B; Supplemental Datasets 2 and 3). To our knowledge, this is not a general feature of PSII deficient mutants and may be specifically linked to the main purported function of OHP2/OHP1, i.e. integration of Chl into PSII. Accumulation of POR1 was increased both at the cell and membrane level, while for other enzymes like the Mg-chelatase, CHLM and CTH1, the observed effect was an increased association with the membrane. The OHC complex has been linked before with Chl synthesis, albeit in different manners. In cyanobacteria, ChlG (found unaffected here in the Chlamydomonas ohp2 mutant) co-immunoprecipitates with the homologs of OHP2 and HCF244 (Chidgey et al., 2014). In the Arabidopsis ohp2 mutant, Hey and Grimm (2018) reported a decreased accumulation for most of the immunologically analyzed Chl biosynthesis enzymes. The authors proposed a posttranslational destabilization of these proteins in response to OHP2 deficiency, which could be beneficial when D1, a major sink for newly synthesized Chl, is not translated. In Chlamydomonas, the continued translation of D1, when the absence of OHP2 limits its ability to ligate Chl, may instead start a signaling cascade aimed at increasing production of Chl. Alternatively, this could be a response to the marked Chl deficiency in the mutant (Supplemental Table 1). The PSII core contains 35 Chl a molecules, while its pigment bed harbors ∼200 Chl molecules, so loss of the core itself cannot fully account for the 40% decrease in Chl content that we observed. In the Arabidopsis ohp2 mutants, a comparable or even stronger reduction in Chl accumulation was observed, with Chl b reduced to a lesser extent than Chl a (Hey and Grimm, 2018; Myouga et al., 2018). A regulatory role of the OHC on Chl synthesis might be worth further exploration.

Materials and methods

Strains and culture conditions

The Chlamydomonas (Chlamydomonas reinhardtii) ohp2 mutant (10.1.a) was generated by nuclear transformation of the mating type minus (mt-) cell-walled recipient strain Jex4 with the plasmid pBC1, as described in Houille-Vernes et al. (2011). For backcrossing of ohp2 strains, the wild type WT-S34 was used. For localization studies, the cell-wall deficient UVM4 strain generated by Neupert et al. (2009) served as recipient strain. Two PSII mutants, nac2-26 which lacks the psbD mRNA stabilization factor NAC2, and the psbA deletion mutant FuD7, served as controls (Bennoun et al., 1986; Boudreau et al., 2000).

Algal strains were grown at 23°C under continuous white light on TAP at ∼10–30 µE m⁻² s⁻¹, or on high salt minimum medium (MIN) agar plates at 100 µE m⁻² s⁻¹ (Harris, 2009). ohp2 mutant strains were kept on plates at very low light (∼5 µE m⁻² s⁻¹) to limit inadvertent selection of suppressor mutants. Before usage of ohp2 in experiments the absence of potential PSII suppressors was confirmed by F_v/F_m measurements. Liquid cultures were grown in TAP medium, supplemented with 1% (w/v) sorbitol (TAPS) for transformation experiments until a cell density of 2–3 × 10⁶ cells/mL was reached.

The reversion rate of ohp2, was determined according to Kuras et al. (1997).

Spectroscopy and fluorescence measurements

In vivo spectroscopy measurement was performed as described before in Jalal et al. (2015), using cells grown under low light conditions. F_v/F_m is presented as the average of four fluorescence induction measurements at 26, 56, 135, and 340 µE m⁻² s⁻¹ followed by a saturating pulse. PSII/PSI ratio was calculated from the difference in ECS signal at 520 nm, measured 160 µs after a saturating single turnover flash, in the absence or presence of DCMU (10 µM) and hydroxylamine (1 mM) to measure (PSI + PSII) or PSI activity, respectively. In the latter condition, the b/a ratio is the ratio of amplitude of the second phase of the ECS signal (phase b, due to proton pumping by Cytb₆f) to the initial phase a (PSI) and is a rough measure of Cytb₆f activity.

Analysis of nucleic acids

For isolation of nucleic acids, cells were harvested by centrifugation at 1,100 × g, 4°C for 6 min. Genomic DNA was extracted using the DNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to manufacturer's protocol or CTAB buffer (2% (w/v) cetyltrimethylammonium bromide, 100 mM Tris–HCl, pH 8) followed by phenol/chloroform/isoamyl alcohol extraction (25:24:1, Carl Roth GmbH, Mannheim, Germany). DNA (10 µg) was digested by restriction enzymes and separated on 0.8% (w/v) agarose gels in TPE-buffer (89 mM Tris–phosphate, 2 mM Na₂EDTA). Total cellular RNA was extracted by using the TRI reagent (Sigma–Aldrich, Saint Louis, USA). RNA (3 µg) was separated on 1% (w/v) denaturing formaldehyde agarose gels. After separation, nucleic acids were transferred to Roti Nylon⁺ membrane (Roth, Karlsruhe, Germany), followed by UV light cross-linking (UV Crosslinker, UVC 500, Hoefer Inc., San Francisco, USA). Dig-labeled probes were synthesized by PCR from total DNA or cloned cDNA (psbA) by using primers denoted in Supplemental Table S2. Hybridizations and detection of dig-labeled probes were performed using standard methods.

Immunoblot and pulse-chase analysis

For isolation of total protein extracts, cells from 20 mL liquid cultures were harvested by centrifugation and cell pellets resuspended in 200 µL 2 × lysis buffer (120 mM KCl, 20 mM tricine pH 7.8, 5 mM β-mercaptoethanol, 0.4 mM EDTA, 0.2% (v/v) Triton ×100) supplemented with protease inhibitors (cOmplete ULTRA Tablets, Mini, Roche, Switzerland) and lysed via sonication on ice. To separate membrane proteins from soluble proteins for cell subfractionation, cells from 50 mL culture were resuspended in 500 µL hypotonic solution (10 mM Tricine/KOH pH 6.8, 10 mM EDTA, 5 mM β-mercaptoethanol, and protease inhibitors). Lysis was performed mechanically by vortexing thoroughly with glass beads (0.5 mm diameter) two times for 1 min. After centrifugation at 15,000 × g for 10 min, the supernatant was considered as total soluble protein extract.

Protein concentration was measured by Bradford Protein Assay (Bradford, 1976). SDS-PAGE, protein gel blotting, and immunodetection were performed as described by Sambrook and Russel (2001). Antibodies were as follows: α-PsaA (Agrisera, #AS06 172), α-D1 (Agrisera, #AS05084), α-CP43 (Agrisera, #AS11 1787) α-Cytf (Agrisera, # AS08 306), α-HA (Sigma-Aldrich, #H6908). The antiserum against the spinach Rubisco holoenzyme used for the detection of RbcL was kindly provided by G. F. Wildner (Ruhr-University Bochum). The antibody against the Chlamydomonas D2 protein was generated by immunization of rabbits using recombinant D2-GST protein (BioGenes GmbH, Berlin, Germany).

¹⁴C labeling of chloroplast-encoded proteins was performed as in Spaniol et al. (2021). Whole cells pulsed for 5 min with ¹⁴C-acetate in the presence of cycloheximide, then chased for 45 min after washing and chloramphenicol addition were loaded onto a 12%–18% (w/v) urea-containing polyacrylamide gel.

Complementation and localization studies

Constructs generated for the complementation of the ohp2 mutant and OHP2 localization studies are shown in Supplemental Figure 4. All constructs were created by the insertion of a PCR-amplified sequence of interest into the pBC1-CrGFP vector (= pJR38, Neupert et al., 2009; construct 1, Supplemental Figure 4) via NdeI or NdeI/EcoRI restriction sites. Primers used for cloning are given in Supplemental Table 2. For cloning details of the synthetic Arabidopsis (A. thaliana) gene (construct 5, Supplemental Figure 4), see Supplemental Figure 7. For localization studies, the strain UVM4 was transformed by the glass-bead method (Kindle, 1990; Neupert et al., 2009) and positive transformants were selected by growth on TAP plates supplemented with 10 µg/mL Pm. For complementation, constructs were integrated into the genome of the ohp2 mutant strain by the electroporation method (Shimogawara et al., 1998), selected for Pm resistance and subsequently screened for photoautotrophic growth on HSM plates.

2D. blue native (BN) page

Liquid cultures (500 mL) of Chlamydomonas strains were harvested at 1,000 × g for 10 min at 4°C and resuspended in 1 mL TMK buffer (10 mM Tris–HCl, pH 6.8, 10 mM MgCl₂, 20 mM KCl) and protease inhibitors (cOmplete ULTRA Tablets, Mini, Roche, Switzerland). Resuspended cells were lysed by sonication (ultrasound pulses of 10 s for three times). Cells were centrifuged (1,000 × g, 1 min, 4°C) to remove cell debris and unlysed cells. The supernatant was centrifuged for 10 min at 20,000×g at 4°C. The resulting pellet was washed twice with TMK buffer and finally resuspended in 500 µL TMK buffer. Chl concentration was determined by adding 20 µL of the sample to 980 µL methanol, 5 min incubation at RT, and a 1 min centrifugation step to remove the starch. OD652 of the supernatant was measured and Chl concentration calculated using the formula: Chl a (mg/mL) = A652 nm ×1.45. Aliquots containing 25 μg of Chl were again centrifuged for 10 min, 20.000 × g at 4°C and resuspended in 51 µL ACA buffer (750 mM ε-aminocaproic acid, 50 mM Bis–Tris pH 7.0, 5 mM pH 7.0 EDTA, 50 mM NaCl) and solubilized for 10 min on ice after addition of n-dodecyl-ß-D-maltoside (ß-DM) to a final concentration of 1.5% (w/v). Solubilized proteins were separated from insoluble material by centrifugation for 20 min, 16,000 × g at 4°C and mixed with 1/10 volume of BN loading dye (750 mM ε-aminocaproic acid, 5% (w/v) Coomassie G250). First dimension native electrophoresis was carried out according to Schägger and Jagow (1991) on a 5%–12% linear gradient. Protein complexes were further subjected to 2D 12% SDS-PAGE and analyzed by immunoblotting.

Ribosome profiling and data analysis

For each replicate, 500 mL culture (TAP, 30 µE m⁻² s⁻¹) was supplemented with 100 µg/mL chloramphenicol and 100 µg/mL cycloheximide, followed by rapid cooling using plastic ice cubes. The cold (<8°C) culture was then centrifuged at 3,000 × g for 2 min to pellet the cells. The pellet was washed once with ice-cold polysome buffer (20 mM Tris pH 8.0, 25 mM KCl, 25 mM MgCl₂, 1 mM dithiothreitol, 100 µg/ml chloramphenicol, 100 µg/ml cycloheximide) and then flash-frozen in liquid nitrogen. The pellet was resuspended in 8 mL of polysome buffer containing 1× Protease inhibitor Cocktail (Roche) and 1 mM PMSF, and the cells were lysed using a French press (Avestin) at 2 bar. Polysome isolation from the lysate, RNA digestion, and RF isolation have been performed as described in Trösch et al. (2018). For the total RNA, a separate 8 mL of the culture was centrifuged at 3,000 × g for 2 min, and 750 µL of Trizol was added to the pellet directly. Total RNA extraction as well as RNA labeling and microarray analysis was performed as described in Trösch et al. (2018).

Data processing and analysis was conducted according to previous studies (Zoschke et al., 2013; Trösch et al., 2018). Local background was subtracted from single channels (F635-B635 and F532-B532, respectively) and probes located within annotated and confirmed chloroplast reading frames were normalized to the average signal of the compared datasets including all replicates of RFs and total mRNA to remove overall differences introduced by technical variations. Probes covering a respective ORF were averaged, and relative abundance of RFs and total mRNA was calculated for each ORF by normalizing each ORF value to the average of all ORF values. By this, expression of the individual ORF is considered in relation to mean values of all plastid-encoded genes. Relative translation efficiency is determined by comparing to values of average RF intensities relative to the average RNA intensities, for each ORF. All average values and standard deviations are based on three independent biological replicates. Significant differences in gene-specific RNA and RF accumulation and translation efficiencies between WT and ohp2 mutant data were determined with a Welch’s t-test and corrected for multiple testing according to Storey's q-value method. Genes were marked as significant for q-value of <0.05 and with expression changes more than 2-fold (Supplemental Dataset 1).

Mass spectrometry

Whole cell proteomics

Proteins (50 µg) were separated by SDS-PAGE. After a short migration (<0.5 cm) and Coomassie blue staining, gel bands containing proteins were excised and destained. Gel pieces were subjected to a 30 min reduction at 56°C and a 1 h cysteine alkylation at room temperature using 10 mM dithiothreitol and 50 mM iodoacetamide in 50 mM ammonium bicarbonate, respectively. Proteins were digested overnight at 37°C using 500 ng of trypsin (Trypsin Gold, Promega). Supernatants were kept and peptides remaining in gel pieces were further extracted with 1% (v/v) trifluoroacetic acid. Corresponding supernatants were pooled and dried. Peptide mixtures were subsequently reconstituted in 200 µL of solvent A (0.1% (v/v) formic acid in 3% (v/v) acetonitrile). Five microliters of peptide mixtures were analyzed in duplicate on a Q-Exactive Plus hybrid quadripole-orbitrap mass spectrometer (Thermo Fisher) as described in Pérez-Pérez et al. (2017) except that peptides were separated on a PepMap RSLC C18 Easy-Spray column (75 µm × 50 cm, 2 µm, 100 Å; Thermo Scientific) with a 90 min gradient (0% to 20% B solvent (0.1% (v/v) formic acid in acetonitrile) in 70 min and 20% to 37% B solvent in 20 min).

Membrane proteomics

Cells were collected by centrifugation, resuspended in 25 mM phosphate buffer at a density of approximately 0.5 µg/µL total Chl and disrupted by three consecutive freeze and thaw cycles. Soluble proteins were separated by centrifugation at 4°C, 25,000 × g for 15 min. Total protein (20 µg) from the pellet fraction was precipitated in 80% (v/v) acetone, tryptically digested and desalted as described (Hammel et al., 2018). Peptides were resuspended in a solution of 2% (v/v) acetonitrile, 1% (v/v) formic acid just before the LC–MS/MS run. The LC–MS/MS system (Eksigent nanoLC 425 coupled to a TripleTOF 6600, ABSciex) was operated basically as described for data dependent acquisition (Hammel et al., 2018).

For targeted MRM-HR data acquisition, a list of 39 precursor m/z signals were chosen for targeting the five different proteins. The precursors were selected from the results of the data dependent acquisitions and good responding peptides as predicted by a deep learning algorithm for peptide detectability prediction, d::pPop (Zimmer et al., 2018). Collision energies were calculated from the standard CE parameters of the instrument, dwell time was set to 60 ms, fragments were acquired from 110 m/z to 1600 m/z, resulting in a cycle time of 2.6 s.

Data processing and label-free quantification

Shotgun proteomics raw data were processed using the MaxQuant software package as described in Martins et al. (2020) with slight modifications. For protein identification and target decoy searches were performed using a home-made Chlamydomonas protein database consisting in the JGI Phytozome nuclear-encoded proteins database (v.5.6) concatenated with chloroplast- and mitochondria-encoded proteins in combination with the Maxquant contaminants.

For the whole-cell proteomics, the mass tolerance in MS and MS/MS was set to 10 ppm and 20 mDa, respectively and proteins were validated if at least two unique peptides having a protein FDR < 0.01 were identified. For quantification, unique and razor peptides with a minimum ratio count ≥2 unique peptides were used and protein intensities were calculated by Delayed Normalization and Maximal Peptide Ratio Extraction (MaxLFQ) according to Cox et al. (2014). For membrane protein data dependent runs MaxQuant software (v1.6.0.1) (Cox and Mann, 2008; Tyanova et al., 2016) was used. Peptide identification, protein group assignment, and quantification were done with standard settings for ABSciex Q-TOF data except that three miss-cleavages were allowed, minimum peptide length was set to six amino acids, maximum peptide mass 6600 Da.

Shotgun data analysis

For statistical analysis, LFQ intensities obtained from MaxQuant were further analyzed with the Perseus software package version 1.6.15.0 (Tyanova et al., 2016). In case of the whole-cell data, the two measurement replicates were averaged first. For both datasets the biological replicates were grouped and the LFQ intensities were Log₂ transformed. The data were filtered to contain at least three valid values (out of four biological replicates) or two valid values (out of three biological replicates) in case of the whole-cell or membrane fractions, respectively. Significant changes in LFQ intensities compared to the WT were identified by a modified two sample t-test of the Perseus software (permutation-based FDR = 5%, artificial within group variance S0 = 1). The ohp2 mutant, the suppressor mutant M-su1 and in case of the whole-cell samples the complemented line ohp2:OHP2-HA, were compared to the WT independently. Data visualizations shown in Figure 5 and Supplemental Figure 11 were done using R (R Core Team, 2018). The log₂ transformed LFQ intensities and t-test results of the whole-cell samples and membrane fractions are shown in Supplemental Dataset 2 and Supplemental Dataset 3, respectively.

Targeted mass spectrometry data analysis was done using Skyline software (v20.1.0.155) (MacLean et al., 2010; Pino et al., 2020). The spectral library was built by prediction using Prosit (Gessulat et al., 2019), retention times were taken, where possible, from identification results of the respective peptides using MaxQuant. A minimum of two precursors per protein and four transitions per precursor was used for quantification. The integration borders were manually adjusted where necessary and dotp values for the coeluting fragments as reported from Skyline was taken as a quality measure for correct assignment (Supplemental Dataset 4). Retention time varied over all runs by less than 0.5 min, dotp values were usually >0.85 across the replicates where a correct assignment was assumed. Summed fragment peak areas were normalized for unequal sample loading to the total intensities of proteins as identified by MaxQuant in data dependent runs.

The untargeted mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2021) partner repository with the dataset identifier PXD031558. The targeted mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the Panorama Public (Sharma et al., 2018) partner repository with the dataset identifier PXD031631.

Accession numbers

Most sequence data for the protein alignment shown in Figure 2 has been retrieved from the Phytozome database (https://phytozome-next.jgi.doe.gov/): OHP2 and OHP1 protein sequences from C.r. (C. reinhardtii, v5.5, Cre06.g251150 and Cre02.g109950), A.t. (Arabidopsis thaliana, TAIR10, AT1G34000.1 and AT5G02120), P.p. (Physcomitrium patens, v3.3, Pp3c2_26700V3.1), and Z.m. (Zea mays, PH207 v1.1, Zm00008a032025_T01).

Amino acid sequences of the High light-induced proteins HliC (ssl1633) and HliD (ssr1789) from Synechocystis sp. PCC6803 were taken from CyanoBase (http://www.kazusa.or.jp/cyano/).

Supplemental data

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

Supplemental Figure 1 . Genetic analysis of the ohp2 mutation.

Supplemental Figure 2 . Identification of the ohp2 mutation by genome sequencing.

Supplemental Figure 3 . Flanking sequence tags in the ohp2 mutant obtained by Illumina sequencing.

Supplemental Figure 4 . DNA constructs used for complementation and localization studies.

Supplemental Figure 5 . OHP2 is localized to the chloroplast.

Supplemental Figure 6 . Hydrophobicity prediction for Chlamydomonas and Arabidopsis OHP2 amino acid sequences.

Supplemental Figure 7 . Synthetic Arabidopsis OHP2 nucleotide sequences and derived protein sequence used to complement the Chlamydomonas ohp2 mutant strain.

Supplemental Figure 8 . Reproducibility of RF experiments and results for all chloroplast encoding genes.

Supplemental Figure 9 . Targeted ribosome profiling of chloroplast translation reveals enhanced psbD translation in the ohp2 mutant.

Supplemental Figure 10 . Re-accumulation of D2 and restoration of photoautotrophy in suppressor mutants.

Supplemental Figure 11 . Heat maps representing the reproducibility of LC–MS experiments and complementation of ohp2.

Supplemental Figure 12 . Transcript accumulation level of OHP2, OHP1, and HCF244 in Chlamydomonas under various growth conditions.

Supplemental Table 1 . Photosynthetic parameters and Chl composition of Chlamydomonas WT Jex4 (WT), the ohp2 mutant and complemented strain.

Supplemental Table 2 . Primers used in this study.

Supplemental Dataset 1 . Array-based ribosome profiling experiments.

Supplemental Dataset 2 . Whole-cell proteomics experiments.

Supplemental Dataset 3 . Membrane proteomics experiments.

Supplemental Dataset 4 . Targeted mass spectrometry experiments.

Funding

We gratefully acknowledge financial support from the Deutsche Forschungsgemeinschaft to J.N. (Grants Ni390/7-1 and TRR175-A06), A.V.B. (Grant BO 4686/1-1), F.Wi. (TRR 175-A05), M.S. (TRR 175-C02) and from the CNRS (UMR7141) LABEX DYNAMO (ANR-11-LABX-0011-01) and ANR (ANR-08-BIOE-002) to O.V. and X.J. Further support to the Proteomic Platform of IBPC (PPI) was provided by EQUIPEX (CACSICE ANR-11-EQPX-0008). F.Wa. was supported by the Chinese Scholarship Council (CSC).