Fig. 4: Matrix-traversing membrane-deficient mutants have multiple pyrenoids and defective CO₂ delivery to the pyrenoid.

a, Max-z projections depicting RBCS1-Venus localization in Chlamydomonas wild-type, mith1 and saga1 cells. The chloroplast is visualized by chlorophyll autofluorescence. b, Graph depicting the largest cross-sectional area observed in any z-slice for each condensate in each cell. Each tick on the x axis corresponds to a single cell; each point on the graph plots the area for a single condensate. The point depicting the largest condensate in each cell is enlarged to aid visualization. Data for 10 cells per strain are shown. c, Model: in the absence of normal matrix-traversing membranes, their resident matrix-binding proteins could mislocalize and nucleate Rubisco puncta elsewhere in the chloroplast. d, Localization of matrix-binding membrane protein RBMP1 in wild-type versus saga1 mutant backgrounds. e, Anti-CAH3 immunofluorescence. Arrowheads point to regions of strong CAH3 signal associated with the pyrenoid. f, Quantification of fluorescence signal in anti-CAH3 immunofluorescence samples. Results depict the fraction of a cell’s total fluorescence signal found in the pyrenoid. Kruskal–Wallis followed by Dunn’s multiple comparisons test were used for statistical analyses. n = 20 cells per strain except for saga1 with n = 16 cells. *P < 0.05, ****P < 0.0001. See Methods for exact P values. Mean ± s.d. g, Anti-CAH3 western blot performed on whole cell lysates. Anti-alpha-tubulin was used as a loading control. Representative results of 3 biological repeats are shown. h, Model: partially localized CAH3 provides enough CO₂ to sustain partial growth in the mith1 mutant.