Pyrenoid proteomics reveals independent evolution of the CO₂-concentrating organelle in chlorarachniophytes

Significance

Approximately one-third of the global carbon assimilation process is estimated to occur within pyrenoids, which are widely present in eukaryotic algae. Pyrenoids are involved in CO₂-concentrating mechanisms (CCMs) supplying CO₂ to the carbon-fixing enzyme Rubisco. Although Rubisco is the predominant component of pyrenoids, other pyrenoid-associated proteins remain to be elucidated in most organisms. Our proteomic approach revealed the core components of pyrenoids in a marine chlorarachniophyte, many of which were encoded by lineage-specific genes. This study provides insights into the convergent evolution of CO₂-concentrating organelles at the molecular level. Furthermore, pyrenoid-associated proteins reported here may serve for engineering a pyrenoid-based CCM into land plants.

Abstract

Pyrenoids are microcompartments that are universally found in the photosynthetic plastids of various eukaryotic algae. They contain ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) and play a pivotal role in facilitating CO₂ assimilation via CO₂-concentrating mechanisms (CCMs). Recent investigations involving model algae have revealed that pyrenoid-associated proteins participate in pyrenoid biogenesis and CCMs. However, these organisms represent only a small part of algal lineages, which limits our comprehensive understanding of the diversity and evolution of pyrenoid-based CCMs. Here we report a pyrenoid proteome of the chlorarachniophyte alga Amorphochlora amoebiformis, which possesses complex plastids acquired through secondary endosymbiosis with green algae. Proteomic analysis using mass spectrometry resulted in the identification of 154 potential pyrenoid components. Subsequent localization experiments demonstrated the specific targeting of eight proteins to pyrenoids. These included a putative Rubisco-binding linker, carbonic anhydrase, membrane transporter, and uncharacterized GTPase proteins. Notably, most of these proteins were unique to this algal lineage. We suggest a plausible scenario in which pyrenoids in chlorarachniophytes have evolved independently, as their components are not inherited from green algal pyrenoids.

Plants and eukaryotic algae perform photosynthetic carbon fixation, which is the conversion of carbon dioxide (CO₂) into organic carbon, in their plastids. This biological process is catalyzed by ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco), which is the most abundant protein on Earth (1). Although Rubisco proteins are dispersed throughout the plastid stroma in land plants, except for hornworts (2), many eukaryotic algae have a microcompartment containing densely packed Rubisco, the so-called pyrenoid (3). This compartment is not enclosed by a membrane and is believed to be formed by liquid–liquid phase separation of proteins (4–6). Because Rubisco has low affinity and specificity for the substrate CO₂ (7), photosynthetic organisms must supply CO₂ to this enzyme in sufficient concentration. To facilitate carbon fixation, eukaryotic algae have evolved biophysical CO₂-concentrating mechanisms (CCMs) that deliver concentrated CO₂ to a pyrenoid (8, 9). Recent studies on the model green alga Chlamydomonas reinhardtii have revealed key components involved in pyrenoid-based CCMs (10).

The pyrenoid of C. reinhardtii is located at the center of the plastid stroma, surrounded by a starch sheath, and the thylakoid tubules penetrate the pyrenoid matrix. Thus far, a certain number of pyrenoid-associated proteins have been identified by proteomic approaches (11, 12) and fluorescent protein-tagging experiments (13, 14). C. reinhardtii possesses a Rubisco-binding linker protein named essential pyrenoid component 1 (EPYC1) as the pyrenoid core component (15). It is an intrinsically disordered protein carrying repeated motif sequences and involved in Rubisco condensation via liquid–liquid phase separation (16). Similar motif sequences have been found in several pyrenoid-associated proteins that probably play important roles in pyrenoid formation (17, 18). The pyrenoid matrix also contains the AAA+ (ATPases associated with various cellular activities) protein Rubisco activase 1 (19). As other important players in the biophysical CCM, carbonic anhydrases and bicarbonate transporters function to supply CO₂ to Rubisco. Bestrophin-like transporters have been proposed to deliver bicarbonate to the thylakoid lumen (20), where α-carbonic anhydrase converts bicarbonate to CO₂ (21). The low luminal pH assists in this conversion (22), and the pyrenoid-penetrating thylakoid tubes facilitate CO₂ supply. Besides the model organisms, these aspects in many other algal groups remain to be explored, and comprehensive understanding of pyrenoid-based CCMs is lacking. To elucidate the diversity and evolution of pyrenoids, investigating these aspects in non-model organisms is necessary.

Photosynthetic organisms have evolved through endosymbiotic events in the eukaryotic tree of life (23, 24). The ancestral plastid of Archaeplastida (streptophytes, chlorophytes, rhodophytes, and glaucophytes) originated from the primary endosymbiosis of a cyanobacterium (25). Other algal groups obtained plastids through secondary or subsequent endosymbioses with plastid-bearing algae (26, 27). Chlorarachniophytes are marine unicellular algae containing secondary plastids derived from a green algal endosymbiont (28, 29), which can be proved by the presence of residual endosymbiont nucleus, the so-called nucleomorph, in the periplastidal compartment between the outer two and inner two plastid membranes (30). This algal group is a favorable model for studying endosymbiotic plastid evolution owing to its capacity for comparative analysis with green algae. Chlorarachniophyte pyrenoids usually protrude from the plastid periphery and are mostly, but not completely, encircled by four plastid membranes (29). The pyrenoid matrix is penetrated by the inner pair of plastid membranes, which are probably constricted by the FtsZ-dependent machinery (31). Chlorarachniophytes synthesize β-1,3 glucans rather than starch as carbohydrate storage in the cytoplasmic vesicles, which cover the projecting pyrenoids (32). Therefore, the ultrastructure of chlorarachniophyte pyrenoids is distinct from that of green algal pyrenoids.

In this study, we developed a method for isolating intact pyrenoids from the chlorarachniophyte Amorphochlora amoebiformis and performed proteomic analysis using mass spectrometry, resulting in the identification of 154 potential pyrenoid-associated proteins. Subcellular localization experiments revealed eight pyrenoid-specific proteins, most of which are encoded by lineage-specific genes. Our findings suggest that chlorarachniophyte pyrenoids have evolved independently, because their components are not derived from the algal endosymbiont.

Results and Discussion

Proteomics of Isolated Pyrenoids.

To investigate the conditions for isolating pyrenoids from A. amoebiformis cells, we used the transformed strain AaGPY1 (33) expressing a green fluorescent protein (GFP) fused with the Rubisco small subunit (RbcS), which allowed us to visualize the pyrenoids under a fluorescence microscope (SI Appendix, Fig. S1). Short-term sonication (20 kHz, 10 to 20 s) could disrupt the cytoplasmic membranes and detach projecting pyrenoids from plastids. Pyrenoids were collected as pellets by centrifugation through a 40% single-layered Percoll cushion (Fig. 1A). The pellets contained 1 to 2 µm granules with GFP fluorescence, while chlorophyll autofluorescence was hardly detected (Fig. 1B and SI Appendix, Fig. S1), suggesting that it was rich in pure pyrenoids. Moreover, electron microscopy indicated that each granule was a pyrenoid surrounded by plastid membranes and covered by a capping vesicle consisting of polysaccharides (Fig. 1B). Total proteins extracted from the pyrenoid pellets were separated by polyacrylamide gel electrophoresis, which showed several visible bands that probably corresponded to the Rubisco large and small subunits (RbcL and RbcS), as well as RbcS-GFP fusion (Fig. 1C).

Fig. 1.

Purification and proteomic analysis of chlorarachniophyte pyrenoids using Amorphochlora amoebiformis. (A) Pyrenoids were collected from AaGPY1 cells expressing the RbcS-GFP fusion protein by centrifugation through a 40% Percoll cushion. (B) Images of isolated pyrenoids by differential interference contrast (DIC) microcopy, fluorescence microscopy (GFP), and transmission electron microscopy (TEM). Py, pyrenoid; CV, capping vesicle. (C) Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of the total proteins in isolated pyrenoids. (D and E) Pie charts represent the gene location and predicted function of 154 proteins in the pyrenoid proteome.

To determine the protein composition of A. amoebiformis pyrenoids, proteomic analyses were conducted using mass spectrometry. We considered three biological replicates and identified potential proteins with more than three different peptides in all three experiments (SI Appendix, Tables S1–S4). This pyrenoid proteome consisted of 154 proteins, which were encoded by 134 nuclear genes, 10 nucleomorph genes, and 10 plastid genes (Fig. 1D). Based on BLAST searches against the NCBI database, 46 proteins were found to be of unknown function, of which 35 were considered as chlorarachniophyte-specific proteins because their orthologs could not be detected in other organisms (Fig. 1E). The others were annotated as proteins related to membrane transport, protein processing, carbon fixation, translation/transcription, polysaccharide metabolism, glutathione metabolism, oxidoreductase activity, etc. (Fig. 1E). Not all of these proteins exist in the pyrenoid matrix; however, some may be localized in the plastid membranes, intermembrane spaces, and capping vesicle beside the pyrenoid matrix. Notably, seven proteins in the pyrenoid proteome were annotated as components of the thylakoid membrane photosystem (e.g., PsaA, PsaB, PsbB, and PsbC), suggesting contamination of the thylakoid membranes in the pyrenoid-enriched samples. Therefore, further screening was required to elucidate the pyrenoid components of A. amoebiformis.

Subcellular Localization of Pyrenoid-Associated Proteins.

Proteins of the pyrenoid proteome were ranked by the abundance of peptide-spectrum match (PSM), named PPAP1 to PPAP154 (Putative Pyrenoid-Associated Proteins). The major components of the pyrenoid were included in the higher PSM ranks because the top two proteins, PPAP1 and PPAP2, were RbcL and RbcS, respectively. Indeed, the contaminated photosystem proteins had a lower rank than PPAP65. Nucleus-encoded plastid proteins generally have an N-terminal presequence comprising a signal peptide and a transit peptide in chlorarachniophytes, because they are targeted to the plastids via the secretory pathway (34). Based on in silico predictions, an N-terminal signal peptide was predicted in 72 of the 134 nucleus-encoded proteins in the pyrenoid proteome (SI Appendix, Table S1). To identify pyrenoid-localized proteins, GFP tagging experiments were performed on presequence-carrying proteins, mainly up to PPAP30. Several proteins showed pyrenoid-associated localization (Fig. 2 and SI Appendix, Fig. S2). Five proteins (PPAP3, PPAP6, PPAP8, PPAP10, and PPAP28) and RbcS (PPAP2) were localized to the pyrenoid matrix (Fig. 2A). Two membrane proteins (PPAP4 and PPAP9) were detected at the pyrenoid periphery (Fig. 2B). PPAP12 localized at the pyrenoid center corresponding to the pyrenoid-penetrating periplastidal compartment (Fig. 2C). We further generated polyclonal antibodies against PPAP3, PPAP4, PPAP8, PPAP12, and PPAP28 using their recombinant proteins, and performed immunoelectron microscopy (Fig. 3). Immunogold localization mostly supported the GFP tagging results, except for the anti-PPAP28 antibody, which exhibited a non-specific gold label (SI Appendix, Fig. S3). These pyrenoid-specific proteins are explained in detail below.

Fig. 2.

Subcellular localization of putative pyrenoid-associated proteins (PPAPs). (A–F) Confocal images of fluorescent-tagged fusion proteins. Fluorescent signals of GFP, CFP, YFP, and chlorophyll autofluorescence (Chl) are represented by green, cyan, yellow, and magenta, respectively. White and black scale bars are 5 µm and 2 µm, respectively. BF, bright field. (G) Schematic illustration of the A. amoebiformis plastid. A nucleomorph is present in the periplastidal compartment (PPC) between the inner two and the outer two plastid membranes beside the projecting pyrenoid. The pyrenoid is penetrated by the inner pair of plastid membranes and covered by the capping vesicle.

Fig. 3.

Immunogold localization of five pyrenoid-specific proteins. Immunoblot analyses were performed for each antibody against total proteins of A. amoebiformis, and ultra-thin sections of the cells were labeled by each antibody with a gold-conjugated secondary antibody. (A–C) Immunogold particles of PPAP3 and PPAP8 were observed in the pyrenoid matrix similar to the case of RbcL (PPAP1). (D) Immunogold particles of PPAP4 localized at the pyrenoid periphery (arrowheads). (E) Immunogold particles of PPAP12 were detected in the pyrenoid-penetrating periplastidal compartment (arrowheads). PPC, periplastidal compartment; MT, mitochondrion; CV, capping vesicle.

Several other proteins were localized adjacent to pyrenoids. Briefly, the GFP signals of PPAP15 and PPAP21 involved in polysaccharide metabolism were observed in the cytoplasm and on the outside of the pyrenoids, suggesting the localization of these proteins to the capping vesicles (Fig. 2D and SI Appendix, Fig. S2D). A membrane protein, PPAP30 (bestrophin-like protein) was observed around chlorophyll autofluorescence, probably localized along the plastid surface (Fig. 2E and SI Appendix, Fig. S2E). The signal for PPAP13 (stromal processing peptidase) was detected in the vicinity of pyrenoid, but not overlapping with that of RbcS (Fig. 2F). PPAP20 (chlorarachniophyte-specific protein), PPAP27 (chlorarachniophyte-specific protein), and PPAP29 (chalcone isomerase-like protein) showed a similar localization pattern as PPAP13 (SI Appendix, Fig. S2F). These proteins may be localized in the periplastidal compartment (PPC) where a nucleomorph is present, as a similar localization pattern was observed in PPC-targeted proteins (35). PPAP11 (ClpB protein) localized as small spots outside of the pyrenoid matrix (Fig. 2F) and PPAP14 (chlorarachniophyte-specific protein) and PPAP50 (glutaredoxin) also showed a similar localization pattern (SI Appendix, Fig. S2F). Several others, PPAP16, PPAP18, PPAP23, PPAP24, and PPAP47, were observed in both the pyrenoid matrix and plastid stroma (SI Appendix, Fig. S2G).

Unique GTPases in Pyrenoid Matrix.

PPAP3, PPAP6, and PPAP8 were localized in the pyrenoid matrix and were paralogs of each other (Fig. 4A). These proteins had a conserved G-domain in their C termini, which shows partial sequence homology to immunity-related GTPases implicated in the resistance to intracellular pathogens in vertebrates (36). Using the recombinant PPAP8 protein, we demonstrated the enzymatic activity of the GTPase (Fig. 4B). In contrast to C termini, the N-terminal regions were diverse in sequence and exhibited no obvious similarity with any known proteins in the NCBI database. Based on the deletion analyses, we proposed that the N-terminal uncharacterized regions were involved in the pyrenoid-specific localization of these proteins (SI Appendix, Fig. S4). In addition to the three GTPase paralogs, the pyrenoid proteome included another paralog, PPAP93, with a low PSM rank. BLAST searches against available transcriptomes revealed that orthologs of unique GTPases were conserved in other chlorarachniophyte species (Fig. 4C), including the pyrenoid-lacking species Partenskyella glossopodia (37). This finding was unexpected as it led to the hypothesis that unique GTPases are not involved in pyrenoid biogenesis, at least in P. glossopodia. In connection with GTPases, another pyrenoid matrix protein, PPAP10, carried leucine-rich repeats that were conserved among GTPase-activating proteins. Therefore, the pyrenoid matrix of A. amoebiformis contained four GTPase-related proteins (PPAP3, PPAP6, PPAP8, and PPAP10) among the top 10 proteins listed in the pyrenoid proteome. However, no such proteins have been found in the C. reinhardtii pyrenoid. Although the relationship between Rubisco and GTPases remains unclear, they are assumed to play a pivotal role in pyrenoids due to their abundance. For further functional analyses of these proteins, advanced molecular tools (such as gene manipulation techniques) should be developed in this organism.

Fig. 4.

Unique GTPases localized in the pyrenoid matrix. (A) Three GTPase paralogs (PPAP3, PPAP6, and PPAP8) carry an N-terminal plastid targeting signal consisting of a signal peptide (SP) and a transit peptide (TP) and a C-terminal conserved G-domain that is similar to immunity-related GTPases of Homo sapiens (HsIRGC) and Mus musculus (MmIRGC). (B) GTPase activity assay using the recombinant protein of PPAP8. Km values were derived from Lineweaver–Burk plots. Error bars indicate SDs (n = 3). (C) Maximum likelihood phylogenetic tree of unique GTPases from nine species of chlorarachniophytes. The tree was generated under the LG+R5 model implemented in IQ-TREE version 2.2.0. Numbers at major nodes represent bootstrap support values. The immunity-related GTPases of vertebrates were included as an outgroup.

Candidate of Rubisco-Binding Linker Protein.

PPAP28 is a pyrenoid matrix protein of 2,209 amino acids (accession number LC781537). Its homologs have not been detected in other organisms including other chlorarachniophytes, suggesting that it is a species-specific protein. Using IUPred3, PPAP28 was predicted to mainly consist of intrinsically disordered regions (Fig. 5A). The middle region, from 670 to 1,630 amino acids, carried 90 short repeats of [D/E]-X-P-K-S (Fig. 5A). To explore whether PPAP28 interacts with Rubisco in the pyrenoid matrix, we performed co-immunoprecipitation experiments. Using AaGYP1 cells expressing the fusion protein of RbcS and GFP, Rubisco-binding proteins were precipitated with an anti-GFP antibody conjugated with magnetic beads. Immunoblot analyses indicated that both RbcL and PPAP28 were coprecipitated with RbcS, suggesting that PPAP28 interacts with Rubisco either directly or indirectly (Fig. 5B). However, the two pyrenoid-localized GTPases (PPAP3 and PPAP8) described above were not detected by co-immunoprecipitation (Fig. 5B). PPAP28 is the most likely candidate for a Rubisco-binding linker protein; however further investigations are required to reveal whether this protein is involved in the liquid–liquid phase separation of Rubisco.

Fig. 5.

PPAP28 is a potential Rubisco-binding linker protein. (A) In silico prediction of intrinsically disordered regions in PPAP28 that has 90 short repeats of [D/E]-X-P-K-S (shown in red). (B) Co-immunoprecipitation experiments of Rubisco-binding proteins in wild type and AaGPY1 cells. The RbcS-GFP fusion protein was precipitated with an anti-GFP antibody conjugated with magnetic beads. Input, flow-through (FT), elution samples were probed with each antibody shown on the right.

C. reinhardtii possesses the Rubisco linker protein, EPYC1, which is an intrinsically disordered protein containing five tandem repeat sequences (15, 16). EPYC1 and other pyrenoid-localized proteins contain a common motif, [D/N]-W-[R/K]-X-X-[L/I/V/A], which binds to RbcS of C. reinhardtii (18). Recently, an intrinsically disordered repeat protein, PYCO1, was reported in the diatom Phaeodactylum tricornutum as a Rubisco linker protein with the binding motif K-W-S-P-[R/Q] (38). Therefore, unstructured repeat proteins are commonly localized to pyrenoids and associated with Rubisco in distinct algal lineages. As EPYC1, PYCO1, and PPAP28 share no sequence similarity, they would have evolved convergently.

Pyrenoid Peripheral Proteins.

PPAP4 and PPAP9 were localized to the periphery of the pyrenoid (Figs. 2B and 3D). Additional GFP experiments revealed that their N-terminal plastid-targeting signals could deliver the protein to the plastid stroma and the periplastidal compartment, respectively (SI Appendix, Fig. S2B). Both proteins were predicted to have multiple transmembrane domains other than the N-terminal signal peptide, suggesting that PPAP4 and PPAP9 are integrated into the innermost and the second/third plastid membranes, respectively. The C-terminal region of PPAP4 showed sequence homology with tryptophan-rich sensory proteins (TSPOs) of eukaryotic algae and cyanobacteria (SI Appendix, Fig. S5). TSPOs are porphyrin-binding membrane proteins and proposed to be involved in the response to various stresses, including oxidative stress (39, 40). The N-terminal region of PPAP4 was partially similar to the sequence of triose phosphate isomerases (TPIs), which are involved in glycolysis and gluconeogenesis (SI Appendix, Fig. S5). Therefore, PPAP4 probably evolved through gene fusion between TPI and TSPO, and the N-terminal TPI did not appear to exhibit enzymatic activity because of sequence divergence (SI Appendix, Fig. S5). Orthologs for this fusion protein have been found in other chlorarachniophyte species, but not in other organisms. PPAP9 is a homolog of ATP-binding cassette (ABC) transporters of mitochondria, ATMs, belonging to the ABCB subfamily. The yeast ATM exports glutathione polysulfide and has been implicated in iron-sulfur cluster biogenesis in the cytoplasm (41, 42). Notably, a recent study proposed that iron-sulfur cluster metabolism is a secondary function of the C. reinhardtii pyrenoid, as the rhodanese domain-containing proteins, STR16 and STR18, are found in the pyrenoid matrix (12). The pyrenoid-localized ATM homolog suggested that the chlorarachniophyte pyrenoid is also associated with iron-sulfur cluster metabolism.

Carbonic Anhydrase for Pyrenoid-Based CCM.

To supply CO₂ to Rubisco, C. reinhardtii converts bicarbonate to CO₂ by α-carbonic anhydrase (α-CA) localized in the pyrenoid-penetrating thylakoid lumen (21). Similarly, θ-CAs have been found in pyrenoid-penetrating thylakoid lumen of diatoms (43, 44). Unlike the pyrenoids in these algae, the pyrenoid in A. amoebiformis is deeply invaginated by an inner pair of plastid membranes, and the pyrenoid matrix is partitioned by the plate of periplastidal compartment (45). PPAP12 was identified as a homolog of θ-CAs and found to be localized inside the pyrenoid-penetrating periplastidal compartment (Figs. 2C and 3E). Therefore, the conversion of bicarbonate to CO₂ is likely to occur in the pyrenoid-penetrating periplastidal compartment, which is a counterpart of pyrenoid-penetrating thylakoids in green algae and diatoms. An α-CA localized to the lumen of the epiplastid endoplasmic reticulum has been reported to be involved in the CCM of the ochrophyte Nannochloropsis oceanica (46). This suggests that plastid intermembrane spaces can also serve as sites for CO₂ formation. If so, A. amoebiformis should have unknown mechanisms for transporting bicarbonate into the periplastidal compartment and regulating its pH for CA activity. Our pyrenoid proteome provided clues regarding these mechanisms. PPAP30 is a bestrophin-like protein and its homologs have been proposed to facilitate bicarbonate uptake into the thylakoid lumen of C. reinhardtii (20). PPAP47 was localized in the plastid stroma and pyrenoids (SI Appendix, Fig. S2) and was homologous to LETM1 domain-containing proteins known as proton/cation exchangers in mitochondria (47). As shown in these examples, the pyrenoid proteome would have included candidates involved in the pyrenoid-based CCM of A. amoebiformis.

Pyrenoid Evolution through Secondary Endosymbiosis.

We identified 154 proteins in the pyrenoid proteome of A. amoebiformis and confirmed that eight proteins were specifically targeted to pyrenoids (Fig. 6). When these proteins were compared to the 40 pyrenoid-specific proteins of C. reinhardtii reported to date (10, 13, 18), only one shared protein was identified. PPAP12 of A. amoebiformis and LCIB/LCIC of C. reinhardtii are pyrenoid-associated θ-carbonic anhydrases. However, their localization patterns and roles differ. PPAP12 is speculated to convert bicarbonate to CO₂ in the pyrenoid-penetrating periplastidal compartment, whereas LCIB/LCIC has been proposed to recapture CO₂ leaking from the pyrenoid matrix by conversion to bicarbonate based on its localization; LCIB/LCIC is dispersed in the plastid stroma under high-CO₂ conditions and relocalized to the vicinity of pyrenoid under low-CO₂ conditions (48). Recently, an ABC transporter of the ABCF subfamily was found in the pyrenoid matrix of C. reinhardtii (12). However, it is distantly related to PPAP9, which belongs to the ABCB subfamily. Notably, C. reinhardtii has Rubisco activase 1 (RCA1) as an important component of the pyrenoid matrix, while its homolog has never been found in the transcriptome/genome sequences of chlorarachniophytes or in the pyrenoid proteome. Other than pyrenoid-specific proteins, it is worth noting that thylakoid-localized bestrophin-like proteins of C. reinhardtii (20) and the plastid membrane-localized bestrophin-like protein (PPAP30) of A. amoebiformis seem to be involved in CCMs in common. Overall, pyrenoid constituents, except for Rubisco, seem to differ between A. amoebiformis and C. reinhardtii despite the green algal origin of chlorarachniophyte plastids. Considering that most pyrenoid-associated proteins in A. amoebiformis are unique to chlorarachniophytes, it is plausible that chlorarachniophyte pyrenoids underwent remodeling subsequent to secondary endosymbiosis, encompassing not only the morphology, but also the molecular level. Furthermore, the potential Rubisco linker protein PPAP28 is a species-specific protein, as its homolog was not detected, even in closely related species of chlorarachniophytes. This suggested that compositional changes in pyrenoids also occurred after species diversification. Nonetheless, pyrenoid proteomes from other species are required to elucidate the diversity of pyrenoid constituents in chlorarachniophytes.

Fig. 6.

Schematic representation of pyrenoid-based CCM in A. amoebiformis. The pyrenoid matrix is penetrated by the periplastidal compartment (PPC). In this model, Rubisco proteins bind to the intrinsically disordered protein PPAP28. Three unique GTPases (PPAP3, PPAP6, and PPAP8) and a putative GTPase-activating protein (PPAP10) are the core components of the pyrenoid matrix, while their functions are yet unknown. Two membrane proteins, TSPO (PPAP4) and ABC transporter (PPAP9), are integrated into the innermost and the second/third plastid membranes, respectively. A carbonic anhydrase (PPAP12) is involved in the conversion of bicarbonate to CO₂ in the pyrenoid-penetrating PPC.

After the identification of the Rubisco linker protein EPYC1 in C. reinhardtii, introducing an algal CCM to Arabidopsis thaliana has been attempted, to enhance the photosynthetic efficiency (49, 50). As an initial step, pyrenoid-like Rubisco condensation was observed when EYPC1 and the plant-algal hybrid Rubisco (native RbcL and C. reinhardtii RbcS) were expressed in plant plastids. Notably, chlorarachniophyte RbcS proteins have a sequence resemblance to those of streptophytes rather than to those of green algae, which could be a result of lateral gene transfer (51). This could be advantageous in introducing a pyrenoid-based CCM into land plants using the pyrenoid proteome of A. amoebiformis. The pyrenoid-associated proteins reported here may serve as a basis for improving crop productivity in the future.

Materials and Methods

Pyrenoid Isolation.

Amorphochlora amoebiformis (formerly Lotharella amoebiformis) cells of CCMP2058 (wild type) and AaGPY1 (expressing GFP fused with the Rubisco small subunit) (33) were grown in Erlenmeyer flasks with 200 mL ESM medium at 22 °C under white illumination (60 to 80 µmol photons m⁻² s⁻¹) on a 12:12-h light:dark cycle. Approximately 1 × 10⁷ cells were harvested by gentle centrifugation from 1-wk-old cultures and resuspended in 500 µL of isolation buffer (600 mM sorbitol, 5 mM EDTA, 1 mM MgCl₂, 10 mM KCl, 1 mM MnCl₂, and 50 mM HEPES-KOH; pH 7.6) by pipetting. The cells were disrupted by ultrasonication using a Q125 sonicator with a standard probe #4422 (QSONICA, Newtown, CT, USA) at 20% of the maximum amplitude for 10 s on ice. The cell lysate was layered on a 40% (v/v) single-layered Percoll cushion in isolation buffer and centrifuged for 1 min at 20,000 g using a fixed-angle rotor (AF-5004CH; KUBOTA). After removing the supernatant by pipetting, the white pellet containing the pyrenoids was washed with isolation buffer and collected by centrifugation.

Proteomic Analysis of Pyrenoid Proteins.

Pyrenoids isolated from 5 × 10⁷ cells were lysed with 25 µL of 2 × Laemmli sample buffer (Bio-Rad) and 5% (v/v) 2-mercaptoethanol. The protein sample was boiled at 95 °C for 5 min and loaded onto an Any kD Mini-PROTEAN TGX gel (Bio-Rad). Proteins were concentrated at the interface between the stacking and resolving gels by electrophoresis, and the protein bands were excised from the gel and subjected to in-gel digestion as described previously (52). Briefly, the gel pieces were reduced with 10 mM dithiothreitol in 50 mM ammonium bicarbonate at 56 °C for 45 min, and then alkylated in 55 mM iodoacetamide and 50 mM ammonium bicarbonate at 22 °C for 30 min. After dehydration with 100% acetonitrile, the gel piece was incubated with 20 µg/mL of Trypsin Gold (Promega) in 25 mM ammonium bicarbonate at 37 °C for 16 h. Digested peptides were eluted from the gel by 50% (v/v) acetonitrile with 0.1% (v/v) formic acid. Mass spectra of the peptide fragments were obtained using an EASY-nLC 1000 liquid chromatograph coupled with an Orbitrap Elite mass spectrometer (Thermo Fisher Scientific). Proteins were identified using the SEQUEST algorithm within Proteome Discoverer 1.4 software (Thermo Fisher Scientific) against protein sequences inferred from the MMETSP0042 transcriptome (53), nucleomorph genome (AB996602, AB996603, and AB996604 in GenBank) (54), plastid genome (LC781622), and mitochondrial genome (LC781623) of A. amoebiformis. The mass spectrometry analysis was performed at the Global Facility Center of the Creative Research Institution in Hokkaido University. This experiment was independently repeated three times to obtain reliable candidates of the pyrenoid proteome.

In Silico Analyses.

To verify the accuracy of the nucleus-encoded protein sequences of A. amoebiformis, we compared the originally assembled transcriptome data obtained from iMicrobe (https://www.imicrobe.us/) to the re-assembled data deposited in the Zenodo repository (https://doi.org/10.5281/zenodo.1212585), and manually selected a longer sequence. After removing truncated sequences lacking N-terminal methionine, targeting signals were predicted using four programs: PredSL (55), TargetP-2.0 (56), Predotar (57), and SignalP-3.0 (58), as plastid-targeted proteins of chlorarachniophytes generally possess an N-terminal signal peptide (34). Functional annotation of the pyrenoid-associated proteins was conducted based on BLASTp searches against the non-redundant protein sequences in the NCBI database. If the searches yielded no hits with an e-value cutoff of 1E−20, such proteins were annotated as chlorarachniophyte-specific unknown proteins. Intrinsically disordered protein regions and transmembrane regions were predicted using IUPred3 (59) and DeepTMHMM (60), respectively.

Phylogenetic Analysis.

To construct phylogenetic trees, homologous sequences of unique GTPases and TPI-TSPO were collected from the NCBI database and the MMETSP transcriptomes of chlorarachniophytes (53). Sequence alignments were generated using the L-INS-i method in the MAFFT package (61) and poorly aligned positions were removed using trimAl with the automated1 option (62). Phylogenetic analyses were performed using IQ-TREE2 with the best-fit models selected using the MFP option (63). Branch support was evaluated with 100 standard non-parametric bootstrap replicates using the same substitution models. Phylogenetic trees were visualized using FIGTREE v.1.4.4.

Plasmid Construction for GFP Imaging.

Total RNA was extracted from A. amoebiformis cells using the TRIzol reagent (Thermo Fisher Scientific), and cDNA was synthesized using Superscript IV reverse transcriptase (Thermo Fisher Scientific) with oligo (dT)₂₀ primers. To construct plasmids for the expression of GFP fusion proteins, cDNA fragments encoding potential pyrenoid proteins were amplified by PCR with KOD One PCR Master Mix (TOYOBO) and the specific primers listed in SI Appendix, Table S5. Each fragment was inserted into an expression vector (pLaRGfp+mc or pLaRGfp+linker), which encodes enhanced GFP controlled by a promoter region from the Rubisco small subunit (RbcS1) of A. amoebiformis (64). The pLaRGfp+mc plasmid was digested with HindIII and NcoI at sites between the promoter and GFP coding region. The pLaRGfp+linker plasmid was linearized by PCR using the primer set (GGATCCGCTGGCTCCGCTGCT and ATTTACGTATCTGTCTGCTTGG). A linker peptide “GSAGSAAGSG” was inserted between a target protein and GFP in pLaRGfp+linker. DNA fragments were assembled using GeneArt Seamless Cloning and Assembly Enzyme Mix (Thermo Fisher Scientific) and transformed into Escherichia coli DH5α cells. To introduce deletion mutations into the coding regions of PPAP3, PPAP4, PPAP6, PPAP8, and PPAP9, the plasmids were PCR-amplified using the outward-facing primers listed in SI Appendix, Table S5, and the products were self-assembled using GeneArt Seamless Cloning and Assembly Enzyme Mix (Thermo Fisher Scientific). For co-localization experiments, the GFP-coding region of pLaRbcS182-GFP (65) was replaced by cyan fluorescent protein (CFP) gene. Each fragment of PPAP11 and PPAP13 was inserted between the beta-tubulin promoter and yellow fluorescent protein (YFP) gene of the pAaTu-YFP vector (SI Appendix, Fig. S6). Fragments between the SP6 and T7 promoter regions of the resulting plasmids were amplified by PCR and subcloned into pLaRbcS182-CFP at the end of the terminator. All the inserted fragments were verified using Sanger sequencing.

Transient Transformation for GFP Imaging.

Plasmid DNA was extracted from E. coli cells using a QIAGEN Plasmid Mini/Midi Kit (QIAGEN, Hilden, Germany). A. amoebiformis cells were transfected with plasmids using a Gene Pulser Xcell electroporation system (Bio-Rad) as described previously (33). Briefly, 20 to 50 μg of plasmid DNA were dissolved in 100 µL of Gene Pulser Electroporation Buffer (Bio-Rad) after ethanol precipitation. Approximately 5 × 10⁶ to 1 × 10⁷ cells were resuspended in the DNA solution and pulsed at 110 or 120 V for 25 ms in a 0.2 cm cuvette. Transfected cells were immediately transferred to glass-bottomed dishes (Iwaki) containing fresh ESM medium, and incubated under the culture conditions described above. After 24 h of incubation, GFP-expressing cells were observed under an Olympus IX71 inverted fluorescence microscope (Olympus); GFP signals were usually observed for 5 d after electroporation. Confocal images were obtained using an inverted Zeiss LSM 510 laser scanning microscope (Carl Zeiss) and analyzed using ZEN 2012 software (Carl Zeiss).

Antibody Preparation and Immunoblotting.

To generate antibodies, we constructed plasmids to express antigenic proteins in E. coli. Fragments encoding PPAP3 (80 to 587 amino acids), PPAP4 (70 to 293 amino acids), PPAP8 (90 to 574 amino acids), PPAP12 (47 to 319 amino acids), and PPAP28 (94 to 418 amino acids) were amplified by PCR using the primers listed in SI Appendix, Table S5. Each fragment was inserted into pET28a (Merck, Darmstadt, Germany) digested with NdeI and EcoRI and cloned into E. coli DH5α cells. All the inserted fragments were verified by Sanger sequencing. The resulting plasmids were transformed into E. coli Rosetta 2 (DE3) cells (Merck). To express recombinant proteins, E. coli cells were grown in 200 mL of LB medium at 37 °C; subsequently, isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 1 mM at OD₆₀₀ between 0.5 and 0.7. After 4 h incubation at 37 °C, the cells were harvested and lysed in 10 mL BugBuster Protein Extraction Reagent (Merck) with Benzonase Nuclease (Merck), rLysozyme Solution (Merck) and cOmplete Protease Inhibitor Cocktail (Merck). The recombinant proteins were mainly expressed in insoluble forms (inclusion bodies), which were collected by centrifugation and purified using His GraviTrap columns (Cytiva, Buckinghamshire, UK) under denaturing conditions. Briefly, the inclusion bodies were solubilized in binding buffer (8 M Urea, 500 mM NaCl, 5 mM imidazole, and 20 mM Tris-HCl; pH 8.0), and the bound proteins were eluted with elution buffer (8 M Urea, 500 mM NaCl, 500 mM imidazole, and 20 mM Tris-HCl; pH 8.0). Subsequently, the purified proteins were dialyzed against 20 mM Tris-HCl (pH 8.0) with 100 mM NaCl in Slide-A-Lyzer Dialysis Cassettes (7 K or 10 K MWCO) (Thermo Fisher Scientific), which were used for immunizing rabbits in Kiwa Laboratory Animals. Co., Ltd.

The specificity of the antibodies was tested using western blot analysis. Approximately 1 × 10⁷ A. amoebiformis cells were lysed with 200 µL of 2× Laemmli sample buffer (Bio-Rad) including 5% (v/v) 2-mercaptoethanol and boiled at 95 °C for 5 min to solubilize the proteins. The protein sample (10 µL in each lane) was electrophoresed on an Any kD/10% Mini-PROTEAN TGX gel (Bio-Rad) and blotted to a polyvinylidene difluoride membrane using a Trans-Blot Turbo Transfer System (Bio-Rad). Immunoblotting was performed using an iBind Western system (Thermo Fisher Scientific) with rabbit antisera against PPAP3, PPAP4, PPAP8, PPAP12, and PPAP28 at 1:250 dilution and an anti-rabbit IgG horseradish peroxidase-linked secondary antibody (Cytiva Cat# NA934VS) at a dilution of 1:10,000. Signals were detected using ECL Prime western blotting Detection Reagent (Cytiva) and a ChemiDoc MP System (Bio-Rad) (SI Appendix, Fig. S7).

Electron Microscopy.

Isolated pyrenoids were fixed with 2.5% (v/v) glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) at 4 °C for 3 h. After washing with the buffer, the pyrenoids were postfixed in 1% osmium tetroxide for 1 h. The fixed pyrenoids were rinsed three times with distilled water and dehydrated through an ethanol series (30, 50, 75, 90, 95, and 100%), followed by 100% acetone. The dehydrated pyrenoids were infiltrated with 1:1 acetone and Agar low viscosity resin 1078 (Agar Scientific) for 30 min, followed by three changes of 100% resin and incubation at 22 °C for 12 h. The resin was polymerized at 60 °C for 24 h. Ultrathin sections were cut using a Leica EM UC6 ultramicrotome (Leica) and collected onto Formvar-coated copper grids. The grids were observed under a Hitachi H7650 transmission electron microscope (Hitachi) at 80 kV.

For immunogold localization, the specimens were prepared by freeze substitution. A. amoebiformis wild type cells on copper grids (with 0.1 mm hole) were rapidly frozen in liquid propane and immediately transferred into liquid nitrogen. The frozen cells were incubated in 100% ethanol at −80 °C for 48 h, −20 °C for 2 h, followed by 4 °C for 2 h using a Leica EM AFS2. The cryofixed cells were infiltrated and embedded in LR White resin (Agar Scientific) as described previously (65). Ultrathin sections were labeled with rabbit antisera against PPAP3, PPAP4, PPAP8, PPAP12, and PPAP28, and an anti-RbcL antibody (Agrisera Cat# AS03 037, RRID:AB_2175406) at a dilution of 1:25, followed by a gold-conjugated secondary antibody (Sigma-Aldrich Cat# G7402, RRID:AB_259953) at 1:25 dilution. Immunolabeled sections were stained with 2% uranyl acetate for 10 min and observed under a transmission electron microscope.

Co-Immunoprecipitation.

To investigate the interaction between Rubisco and other pyrenoid matrix proteins, we performed co-immunoprecipitation with an anti-GFP antibody using AaGYP1 cells expressing the RbcS-GFP fusion protein. Isolated pyrenoids from 1 × 10⁸ cells were resuspended by 150 µL lysis buffer (10 mM Tris-HCl, 150 mM NaCl, 0.5 mM EDTA, and 0.4% TritonX-100; pH 8.0), which was diluted with 350 µL wash buffer (10 mM Tris-HCl, 150 mM NaCl, and 0.5 mM EDTA). The pyrenoid lysate was mixed with 25 μL of GFP-Trap Magnetic Agarose beads (Proteintech) and incubated for 2 h at 4 °C with gentle rotation. After washing thrice, the beads were resuspended in 80 µL of 2× Laemmli sample buffer (Bio-Rad) including 5% (v/v) 2-mercaptoethanol and boiled at 95 °C for 5 min to dissociate the proteins from the beads. As previously described, western blot analysis was performed using antibodies against PPAP3, PPAP8, and PPAP28 at 1:250 dilution. An anti-RbcL antibody (RRID:AB_2175406) and anti-GFP antibody (Takara Bio Cat# 632380, RRID:AB_10013427) were used at 1:2,500 dilution, and anti-mouse IgG secondary antibody (NA931; Cytiva) was used at a dilution of 1:10,000.

GTPase Activity Assay.

The GTPase activity of PPAP8 was measured using an ATPase/GTPase Activity Assay Kit (Merck). To express the recombinant protein of PPAP8 in soluble form, the E. coli culture was incubated at 20 °C for 20 h with 1 mM IPTG. The E. coli cells were lysed in BugBuster Protein Extraction Reagent (Merck), as previously described. The recombinant protein was purified from the cell lysate using a His GraviTrap column (Cytiva) under non-denaturing conditions, according to the manufacturer’s instructions. Subsequently, the buffer was replaced with 20 mM Tris-HCl (pH 8.0) and 100 mM NaCl, using a PD-10 desalting column (Cytiva). Mixtures of 10 µL protein solution (0.5 mg/mL) and 10 µL of various concentrations of GTP (0, 0.1, 0.2, 0.5, 1.0, and 1.5 mM) were incubated at 22 °C for 30 min. As a negative control, recombinant protein was inactivated by boiling for 5 min. Subsequently, 100 µL of malachite green reagent was added to each reaction and incubated for 30 min. The concentration of free phosphate was calculated by measuring the absorbance at 620 nm using a spectrophotometer (DU 730; Beckman Coulter). The experiment was repeated three times.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Data, Materials, and Software Availability

Sequences of A. amoebiformis are deposited to DDBJ/EMBL/GenBank under the accession number LC781622 (66), LC781623 (67), and LC781537 (68). All other data are included in the manuscript and/or SI Appendix.