Fig. 3: SAGA1 and MITH1 localize to membranes, bind each other and perform distinct functions in matrix-traversing membrane biogenesis.

a–g, Localizations of Venus-tagged proteins in various Chlamydomonas mutant backgrounds. Transparent arrowheads point to membranes inside the matrix; filled arrowheads point to membranes at the matrix entry sites. Images are presented at the same scale. h–j, Two-colour immunofluorescence was performed for anti-FLAG and anti-CAH3 (h,i) and anti-FLAG and anti-MITH1 (j). Images are presented at the same scale. k, Cell fractionation and western blot in the absence and presence of 1% DDM detergent; Pel, pellet; Sup, supernatant; Fd, ferredoxin. Representative images from 3 biological repeats are shown. l, Spot test depicting growth at varying levels of CO₂ and 100 µmol photons m⁻² s⁻¹ light levels. m–o, Localization of CYN7-Venus in wild-type and mutant backgrounds. Images are presented at the same scale. p, TEM of wild-type and mith1 mutant pyrenoids. Arrowheads point to matrix surface-associated membranes. q,r, Spectral counts of proteins identified by mass spectrometry after immunoprecipitation of MITH1-Venus-3×FLAG (q) or SAGA1-Venus-3×FLAG (r), plotted against the spectral counts of the same proteins after immunoprecipitation of Venus-3×FLAG as a control for non-specific binding. Non-labelled points are partially transparent to facilitate visualization of overlapping points. Results from 2 replicates were averaged. s–v, Model for how SAGA1 and MITH1 produce matrix-traversing membranes. s, Without either protein, membranes are unable to enter the Rubisco matrix. t, SAGA1 initiates contact between thylakoid membrane sheets and the surface of the pyrenoid by binding to membranes and Rubisco. u, SAGA1 recruits MITH1, which increases the binding force per unit area between membranes and matrix, outcompeting EPYC1–Rubisco interactions and allowing extension of membrane sheets into the matrix. v, In Chlamydomonas, additional factors shape the membrane sheets into cylindrical tubules.