Rcf2 crosslinks specifically to Cox12 and Cox13

To be able to uncover specific crosslinks for the different proteins biochemically, co-immunoprecipitation with antisera against Rcf2, Cox13 and Cox12 was carried out. To this end, the experiment was executed in the same way as before, except mitochondria were solubilized with low concentrated Triton X-100 and SDS. Any native interaction should be disrupted under these conditions and only crosslinked protein complexes were co-purified, still providing appropriate conditions for efficient antigen-antibody reactions on coated beads.

Figure 3-6: Rcf2 specifically crosslinks with Cox12.

Wild-type mitochondria were subjected to crosslinking with 50 µM BS³ for 1 h, reaction was quenched with TRIS, mitochondria were lysed with 0.5 % Triton and 0.1 % SDS, and applied to co-immunoprecipitation with Rcf2, Cox12, control beads. Totals and glycine eluates were used for SDS PAGE (10-16 %) and western blotting. Signals were detected with Rcf2 and Cox12 antibodies, respectively. Control samples without added crosslinker went through the same procedure in parallel. x1, x2 and x3

mark specific crosslinks between Rcf2 and Cox12.

The respective antibody against the opposite, crosslinked protein was used for detection and vice versa, in order to identify the definite association. In fact, different crosslinks or crosslink-regions could be purified and detected by this approach (Figure 3-6).

A crosslinking ladder at ~30-40 kDa (x1), another one right beneath at ~21-28 kDa (x2) and one specific crosslink at ~16 kDa (x3) were isolated with both, Cox12 and Rcf2 co-immunoprecipitation, and detected with the particular other antibody. This indicates a specific appearance, although full-length Rcf2 (25 kDa) and Cox12 (10 kDa) can only be responsible for the crosslinks x1, according to their size. X2 might occur due to a processed version of Rcf2, which is detectable in enriched quantity after Rcf2 immunoprecipitation (Figure 3-6, lane 3).

An internal processing event of Rcf2 was studied previously by Römpler et al. (2016) resulting in a stable Rcf2^C fragment (~21 kDa) (Römpler et al., 2016). This corresponds to the fragment detected at ~18 kDa in this case. This minor discrepancy can be justified with the usage of different gel systems and standard reference markers. It has to be noted, however, that another fragment can be enriched as well at ~16 kDa, which is not represented in previous published data (Römpler et al., 2016). In contrast to the slightly larger fragment, that fragment is not detectable in the total and could be an artefact caused by the procedure itself.

Upon addition of the molecular weight of Rcf2^C and Cox12 to ~28 kDa, it can be conceived that those partners generate the isolated crosslink x2. The obtained crosslink x3 however, is only

~7 kDa larger than Cox12 and the interacting partner remains unknown.

In order to identify the crosslinks of Rcf2 and Cox13, we used the same Rcf2 elution fraction for a second SDS-PAGE, loaded next to the Cox13 bound fraction (Figure 3-7). By this, the detection can occur from the same gel or membrane and mutual crosslinks can be specified with higher certainty. At first sight, the isolation of crosslinks between these two proteins appears to be less efficient than before with Cox12. Although general co-immunoprecipitation capacity seems to be at a satisfying level (Figure 3-7, lanes 3 and 13), the detection with the respective other antibody was weak (Figure 3-7, lanes 6 and 12). Paying attention to longer exposures of the corresponding lanes (5, 6, 11 and 12), we could obtain two crosslinks, corresponding the two proteins by size. Full-length Rcf2 (25 kDa) and Cox13 (15 kDa) can be added up to 40 kDa, which is roughly where x1 is detected. Considering again the Rcf2^C fragment at ~18 kDa, the other crosslink x2 finds an explanation. However, there is a third crosslink x3 of unknown nature isolated, as well ~7 kDa larger than the protein of interest, Cox13. Once more, a smaller fragment of Rcf2 could be detected, also present in the Cox13 elution. One has to note, that an

obvious amount of unlinked Cox13 can be determined upon Rcf2 co-immunoprecipitation (lane 12) but only under crosslinking conditions. Arguing for a tight interaction of the two proteins, this should have been detectable in the sample without crosslinker as well. It could also indicate a strong association of Cox13 to another Rcf2 crosslinked protein which was not identified by mass spectrometry.

Figure 3-7: Rcf2 specifically crosslinks with Cox13.

Wild-type mitochondria were subjected to crosslinking with 50 µM BS³, reaction was quenched with TRIS, mitochondria were lysed with 0.5 % Triton and 0.1 % SDS and applied to co-immunoprecipitation with Rcf2, Cox13, control beads. Totals and glycine eluates were used for SDS PAGE (10-16 %) and western blotting. Signals were detected with Rcf2 and Cox13 antibodies, respectively. Control samples without added crosslinker went through the same procedure in parallel. x1, x2 and x3 mark specific crosslinks between Rcf2 and Cox13.

In the end, those identified crosslinks are consistent with the idea of Rcf2 as interacting with the peripheral part of the oxidase. The recently published cryo-structure of complex III2IV resolved Rcf2^107-205 in association with the respirasome where it sits between Cox13 and Cox3 clamped right below Cox12 (Figure 3-8A) (Hartley et al., 2020). This structure not only helped to confirm the proposed crosslinks by analyzing the distances but also indicates in which state the crosslinks occurred. In fact, the distance between Cox12-K41 and Rcf2-K203 measures 11.6 Å, which corresponds well to the linker length of BS³ of 11.4 Å (Figure 3-8B). This implies that the crosslinked Rcf2 is positioned at the supercomplex not only the cytochrome c oxidase itself. It is also interesting to note, that exclusively the C-terminus is resolved in this cryo-structure and it appears that also the Rcf2^C fragment itself can be found crosslinked to Cox12. Since processed Rcf2 is not part of a protein database for mass-spectrometry calculation,

the only evidence is our biochemical analysis. Yet, it indicates that not only full-length Rcf2 but also Rcf2^C is still in proximity to Cox12 and Cox13.

Analyzing the crosslink observed between Rcf2-K203 and Cox13-K85 with the help of the available cryo-structure however, resulted in a distance of 60.5 Å (Figure 3-8C). It is known that the theoretical crosslink linker length can differ from the actual distance. It was shown that BS³ crosslinks up to 30 Å distant lysine chains due to dynamic reasons (Merkley et al., 2014), but the distance between Rcf2 and Cox13 exceeds this threshold twice as much. This suggests that the crosslink between those chains does not take place at the endpoint represented by cryo-EM and might rather be a result of a dynamic interaction of Rcf2 with the cytochrome c oxidase.

Figure 3-8: Structural analysis reveals crosslink of Rcf2 and Cox12 likely to happen at the supercomplex.

A) Localization of Rcf2 at cytochrome c oxidase extracted from cryo-EM structure of Hartley et al. (2020) (PDB ID 6T15).

Dark red: Rcf2, green: Cox13, yellow: Cox12, grey surface: remaining COX subunits B) and C) Crosslink between lysine sidechain K203 of Rcf2 (dark red) and lysine sidechain K41 of Cox12 (yellow) or lysine sidechain K85 of Cox13 (green), respectively. The dashed line indicates the suggested crosslink. Analysis and editing were performed with Pymol software.