Localization of Rcf-proteins

4.2.1 Rcf-proteins as substoichiometric interactors of COX Different cryo-EM structures did not resolve the Rcf proteins as part of the respiratory supercomplexes (Mileykovskaya et al., 2012; Hartley et al., 2019; Rathore et al., 2019). Rcf1 and Rcf2 were found, however, as part of the corresponding mass spectrometry data (Hartley et al., 2019, 2020). Garlich et al. (2017) characterized Rcf1 as it transiently associates to complex IV and does not remain at the complex. In addition, different populations of complex IV can be detected (Vukotic et al., 2012; Garlich et al., 2017). A considerable amount of Rcf-proteins does

not reside at the supercomplexes or mature complex IV (Römpler et al., 2016; Dawitz et al., 2020). Co-immunoprecipitation experiments indicate that the Rcf-proteins interact among each other as well (Römpler et al., 2016). Consequently, they were classified as substoichiometric interactors of complex IV and within the supercomplexes III2IV(2).

A recently published cryo-EM structure of the yeast respiratory supercomplex, remarkably resolved Rcf2^107-205 as a structural subunit of the hypoxic complex III2IV (Hartley et al., 2020).

In a previous approach of the same group, a similar density could be identified within the same pocket, formed by Cox3, Cox12 and Cox13 in complex III2IV2, but not assigned with certainty (Hartley et al., 2019). They argue that isolated from Cox13 as a late-stage assembled protein, completely matured supercomplexes can be isolated efficiently. Additionally, different experiences proved that the association of Rcf-proteins as peripheral interactors is sensitive to detergent treatment (Vukotic et al., 2012; Römpler et al., 2016). This might be a reason why the proteins were missed in previous structures. Our analyses rather corroborate previous data with the Rcf-proteins, including Rcf2, as substoichiometric interactors of the supercomplex. The endogenous proteins were detected in a wild-type situation to localize at the supercomplexes, yet, the majority of the proteins migrate in smaller complexes (section 3.5.1). These findings deliver circumstantial evidence for a population of supercomplexes being present without Rcf-proteins.

In order to address this question, one could ideally isolate from proteins assembling at a different stage or even the Rcf-proteins themselves and subject the sample to BN-PAGE. The proteins would have to be tagged that native application is possible and isolation efficiency remains comparable. When working on this approach we repeatedly faced the problem that the proteins were not fully functional and caused problems in vivo in biogenesis of the respiratory chain. Consequently, some adjustments would be necessary and functionality can be controlled with the respective RCF mutants as they display impaired growth on non-fermentable media (section 3.1.2).

4.2.2 The interaction network of Rcf2 and Rcf3

Chemical crosslinking with subsequent mass-spectrometry analysis revealed Rcf2, Rcf3 and a so far uncharacterized protein Min8 (to be discussed in section 4.1) crosslinked to the cytochrome c oxidase (Linden et al., 2020). We could prove these crosslinks to be specific in our biochemical analyses (section 3.2.1 and 3.2.2). It places all three proteins at the periphery of complex IV in contact with Cox12 and, in the case of Rcf2, with Cox13 as well. When evaluating

our biochemical analysis of Rcf2 crosslinks we could also detect the C-terminal fragment of Rcf2. Our data indicate that the fragment remains at the same localization as full-length Rcf2 as similar specific crosslinks occurred but shifted by the size of cleaved Rcf2^N (section 3.2.1). This finding furthermore matches the resolved C-terminus of Rcf2 in the cryo-structure of Hartley et al. (2020). The reason that the N-terminus is not resolved could be indicative for partial processing of Rcf2 leading to a heterogeneity in supercomplex populations. In contrast to Rcf2 and Rcf3, crosslinks with Rcf1 could not be detected in the approach we used with crosslinking whole mitochondria (Linden et al., 2020). Due to the high amounts of mitochondrial protein crosslinks, low abundant proteins with transient interactions are hard to detect. Additionally, the topology and transmembrane spans are playing a role since the residues embedded into the membrane are barely accessible for the crosslinker (Linden et al., 2020). An adaption of the analysis as described below could still resolve the interaction with Rcf1 and complex IV by a crosslinking-mass spectrometry approach.

The crosslinking results themselves cannot conclude any timeframe the interaction takes place, being obtained from whole mitochondria. However, it is likely that the crosslink occurs at the supercomplex level, since Rcf2, Rcf3 and Min8 were found in previous mass-spectrometry data collected from supercomplex isolation via Cor1^TAP (Vukotic et al., 2012). Rcf1 and Rcf2 were also found in mass-spectrometry data of supercomplex isolation via Cox13^His. Therefore, it would be interesting if Rcf3 and Min8 were present as well (Hartley et al., 2019, 2020).

Unfortunately, the data was only partly published in this case.

In order to get a better understanding of the involvement of the Rcf-proteins and Min8 it could also be a feasible approach to use the Cor1^TAP isolated supercomplex for a more structural analysis. Since the proteins can be co-isolated, a subsequent crosslinking and mass-spectrometry analysis could identify the interaction on the level of the supercomplex. It could also give more information about the stoichiometry of the proteins. By this, however, one would have to rely on a stable interaction for the isolation beforehand. The advantage of our approach (Linden et al. 2020) is that the crosslinking procedure with whole mitochondria is performed without the effects of detergents. Subsequently, harsher detergents like SDS can be used for solubilization without disrupting the chemical connection the proteins established via the crosslinker. Thus, one could use this idea the other way around: First, whole mitochondria are subjected to chemical crosslinking and afterwards isolated via Cor1^TAP to run the mass-spectrometry analysis.

Even transient interactions with the supercomplexes would be enriched and a close to in vivo situation could be evaluated.