Charting the native architecture of Chlamydomonas thylakoid membranes with single-molecule precision

Essential revisions:

1) The reviewers agree that the structures mapped to PSII-LHCII (5XNL of C2S2M2 with CP24 in pea as described in Su et al., 2017) and PSI-LHCI (5ZJI of PSI-LHCI-(phospho)LHCII supercomplex from maize state II in Pan et al., 2018) were surprising choices. Given that three cryo-EM structures of PSII-LHCII supercomplexes were recently reported in C. reinhardtii (Burton-Smith et al., 2019, Shen et al., 2019 and Sheng et al., 2019), and several PSI-LHCI models from C. reinhardtii are also available (Su et al., 2019; Kubota-Kawai (2019) JBC; Suga et al. (2019) Nat. Plants), the reviewers would like to see these more appropriate structures mapped to the tomograms using the membranorama software.

We thank the reviewers for the suggestion. The reason that the initial draft of the manuscript used higher plant complexes was simply timing: the analysis in our study was finished before the 2019 Chlamydomonas structures were available. We agree that it is very appropriate to use the new Chlamydomonas structures in the final manuscript, so we have made new figures accordingly. Given the limited resolution that we used for mapped-in structures (25 Å), swapping the PSII and PSI core complexes made very little noticeable difference (e.g., see the new membrane models in Figure 2D). However, mapping in the newly-published structure of the C₂S₂M₂L₂ PSII-LHCII supercomplex (Sheng et al., 2019) produced a clear difference in the membranogram overlays (Figure 4C-D), and resulted in minor changes in the calculated membrane occupancy and inter-membrane overlap of the supercomplex-associated LHCII antennas. Please note that while C₂S₂M₂L₂ supercomplexes were mostly accommodated within the membranes, some in-plane overlap was introduced, unlike for C₂S₂M₂. We discuss this in the text:

"Then we generated membrane models by using the positions and rotational orientations of PSII luminal densities seen in the membranograms to place structures of C₂S₂M₂L₂-type PSII-LHCII supercomplexes (Burton-Smith et al., 2019; Shen et al., 2019; Sheng et al., 2019), the largest supercomplexes that have been isolated from Chlamydomonas (Figure 4C-D). Supercomplex models of different sizes occupied 47.3% ± 6.0% (C₂S₂M₂L₂), 40.7 ± 5.2% (C₂S₂M₂), and 29.1% ± 3.7% (C₂S₂) of the membrane surface area (Figure 4E), with cytb₆f occupying an additional 5.8 ± 1.6%."

Although we now base our primary analysis on C₂S₂M₂L₂ supercomplexes, we felt it was informative to provide the membrane occupancy for a variety of PSII models in our revised Figure 4E, including C₂ (which is similar to the EM density overlap), C₂S₂, C₂S₂M₂, and C₂S₂M₂L₂.

2) Remapping different structures to the membranograms would also mean revisiting the statistical analysis of distribution and overlap between these complexes across membranes.

After mapping in the new structures, we repeated the analysis of membrane occupancy and overlap of complexes between membranes (Figure 4E-F). As mentioned above, this resulted in minor changes in the numbers, with increased occupancy and LHCII overlap for C₂S₂M₂L₂ supercomplexes. However, there was no effect on the conclusion that PSII-LHCII supercomplexes are randomly arranged across stacked membranes (see Figure 4F).