Assembly of the algal CO2-fixing organelle, the pyrenoid, is guided by a Rubisco-binding motif

Abstract

Approximately one-third of the Earth's photosynthetic CO₂ assimilation occurs in a pyrenoid, an organelle containing the CO₂-fixing enzyme Rubisco. How constituent proteins are recruited to the pyrenoid and how the organelle's subcompartments-membrane tubules, a surrounding phase-separated Rubisco matrix, and a peripheral starch sheath-are held together is unknown. Using the model alga Chlamydomonas reinhardtii, we found that pyrenoid proteins share a sequence motif. We show that the motif is necessary and sufficient to target proteins to the pyrenoid and that the motif binds to Rubisco, suggesting a mechanism for targeting. The presence of the Rubisco-binding motif on proteins that localize to the tubules and on proteins that localize to the matrix-starch sheath interface suggests that the motif holds the pyrenoid's three subcompartments together. Our findings advance our understanding of pyrenoid biogenesis and illustrate how a single protein motif can underlie the architecture of a complex multilayered phase-separated organelle.

Fig. 1. A polyclonal antibody raised against the pyrenoid protein SAGA1 interacts with at least five other pyrenoid proteins.

(A) Electron micrograph of a median plane section through an air-acclimated Chlamydomonas cell. N, nucleus; C, chloroplast; P, pyrenoid; M, Rubisco matrix; T, tubules; S, starch sheath. Scale bar, 1 μm. (B) Two proteins, the Rubisco linker EPYC1 and the starch sheath–binding protein SAGA1, have been previously characterized and localized to the pyrenoid. (C) An anti-SAGA1 antibody was incubated with cell lysate in an effort to coimmunoprecipitate proteins that bind to SAGA1. (D) Coomassie-stained SDS-PAGE of proteins immunoprecipitated by the anti-SAGA1 antibody from wild-type (WT) and saga1 mutant lysates. Immunoprecipitated proteins were eluted from anti-SAGA1 antibodies on beads by boiling; beads not incubated with lysate were also boiled for reference. Asterisks show heavy and light immunoglobulin chains. (E) Proteins immunoprecipitated by the SAGA1 antibody from wild type and saga1 were identified by mass spectrometry. Raw spectral counts are plotted on a log scale. (F) Anti-SAGA1 Western blot on denatured protein extracted from wild type, saga1, and epyc1.

Fig. 2. A motif found on pyrenoid proteins is necessary and sufficient for targeting proteins to the pyrenoid.

(A) The location of motifs along the primary sequence of each protein is shown (not to scale). The SAGA proteins each have a predicted starch-binding domain. The RBMP proteins each have predicted transmembrane domains (see fig. S2). (B) Sequence alignment of protein regions containing the pyrenoid motif. Motif residues are colored by physicochemical properties. Intensity of coloring is proportional to frequency at a given amino acid position. Peptides with the sequences shown in (B) were synthesized, and their binding to Rubisco was measured by surface plasmon resonance and peptide tiling array (see Fig. 3). (C) The localization of poorly characterized protein Cre10.g430350 was determined by tagging with the fluorescent protein Venus. (D) The localization of the same protein was determined after mutation of the central tryptophan-arginine dipeptide of the motif to a double alanine. (E) Localization of Venus-tagged FDX1 protein without and with the C-terminal addition of three copies of the C-terminal SAGA2 motif. Localization in (C) to (E) was determined by transforming the corresponding constructs into wild-type Chlamydomonas. Scale bar, 2 μm.

Fig. 3. The motif binds to Rubisco.

(A) Peptides containing motifs from the indicated proteins were synthesized, and their binding to Rubisco was measured by surface plasmon resonance (SPR). A.U., arbitrary units. Two peptides not containing the motif were included as controls (Ctrl). Each dot shows the binding response of an independent replica. Negative values (when the experimental binding signal was lower than that of the reference cell) are not plotted. The positions of the predicted motifs are indicated below the graph (not to scale). Significance levels of increased binding relative to control peptides were determined using Welch’s t test. *P < 0.05 and **P < 0.01. (B and C) Arrays of 18–amino acid peptides tiling across the sequences of SAGA2 (B) and RBMP2 (C) were synthesized and probed with Rubisco. The binding signal in (B) and (C) is normalized to a control EPYC1 peptide known to bind to Rubisco (one unit of binding) (21). The positions of motifs are indicated below each graph (to scale).

Fig. 4. The motif orchestrates the architecture of the pyrenoid’s three subcompartments.

(A) Representative confocal images of Venus-tagged proteins that have the Rubisco-binding motif and Rubisco small subunit (RBCS1). Chlorophyll autofluorescence delimits the chloroplast. Scale bar, 2 μm. (B) Proposed model for how the motif mediates assembly of the pyrenoid’s three subcompartments in wild type. The motif on tubule-localized transmembrane proteins RBMP1 and RBMP2 mediates Rubisco binding to the tubules [Retic. region is the reticulated region of the tubules (9)]. Multiple copies of the motif on EPYC1 link Rubiscos to form the pyrenoid matrix (21). At the periphery of the matrix, the motif on starch-binding proteins SAGA1 and SAGA2 mediates interactions between the matrix and surrounding starch sheath. (C) The model in (B) explains the matrix-less phenotype observed in EPYC1-less mutants. (D) The model also explains the absence of matrix and starch plates in mutants where Rubisco’s binding site for the motif has been disrupted (21).