ROMO1 and YME1L seem to be functionally linked

4.3.1 ROMO1 as a new YME1L substrate

An experiment using the cytosolic translational inhibitor emetine showed that ROMO1 has a fast turnover in HEK293T cells. This was in contrast to other translocase components, as well as the yeast Mgr2 (Figure 3.12 and 3.20). This unusual finding prompted further investigation and the data presented here show that ROMO1 is specifically degraded by YME1L. The levels of ROMO1 increased dramatically upon transient knockdown of YME1L, along with the known substrate TIM17A (Rainbolt et al., 2013). Furthermore, the sequence of ROMO1 displays an F-G-T-F motif in each of its transmembrane domains. It has been postulated that human YME1L recognizes its substrates by the specific recognition signal F-h-h-F, where h is a hydrophobic amino acid (Shi et al., 2016). While threonine is not strictly classified as a hydrophobic amino acid, the presence of these very similar degron sequences further indicates that ROMO1 is indeed degraded by YME1L. Interestingly, regarding other known YME1L substrates, only PRELID contains an F-A-A-F sequence, while both TIM23 and TIM17A do not have a similar motif in their sequence.

One can only speculate about the mechanism or underlying cause for the observed degradation of ROMO1 by YME1L. Overexpression of ROMO1 has been correlated with increased ROS production and severe malignancies. One study even named ROMO1 an oncomarker (Shyamsunder et al., 2015). It is therefore possible that the fast turnover of ROMO1 acts as a protective mechanism to prevent the accumulation of ROMO1. This would also allow the cell to downregulate YME1L and regulate OPA1 processing in response to environmental stress.

However, ROMO1 is a deeply membrane-embedded protein. It would therefore be interesting to address the mechanism of how ROMO1 is extracted from the membrane and then degraded.

4.3.2 Microdomains might connect different processes within the inner mitochondrial membrane

This study showed that YME1L import is dependent on ROMO1 but that it also degrades ROMO1. This hints at a self-regulatory or feedback mechanism with yet to be addressed implications. It seems reasonable that a tight control of both proteins is necessary since their effects on ROS production or mitochondrial morphology can have a huge impact on the whole cell or organism.

There is evidence which suggests that Yme1 in yeast is in close proximity to the TIM23 complex and possibly even actively plays a role in the translocation of certain proteins (Rainey et al., 2006). Furthermore, there is also data that YME1L is associated with the human translocase (SILAC mass spectrometry data to be published) and therefore in vicinity of ROMO1. However, it still remains to be determined if and how ROMO1, or the translocase in general, interact with YME1L. For this purpose, immunoprecipitation approaches will have to be optimized, since the tagging of both proteins remains difficult and the interaction could be rather transient.

A recent idea in the field concerns the presence of microdomains within the protein-rich inner membrane. This seems conceivable since the clustering of proteins and complexes that carry out related functions makes these processes more efficient. Known examples include the organization of respiratory chain supercomplexes (Letts & Sazanov, 2017), or the MICOS complex (Wollweber et al., 2017). Furthermore, the idea of a “proteolysis hub” spatially organizing mitochondrial proteases has also been suggested (Wai et al., 2016). This data indicates that within the SPY complex (STOML2, PARL, YME1L), the stomatin-like protein, STOML2, spatially organizes YME1L and therefore might contribute to YME1L processing during high substrate load. Along these lines, it could be possible that YME1L specific translocases, containing ROMO1, are localized in close vicinity to the SPY complex, forming a microdomain within the inner membrane. This would make the regulation of mitochondrial protein import and proteolysis more efficient and easily adaptable during stress and environmental influences. Furthermore, OPA1 needs to be processed right after import. It is

therefore plausible that YME1L is found in close proximity to the translocase so that it can process OPA1 right away.

It has been postulated for some time that the mitochondrial import machinery is integrated into multiple processes, as well as integrated into a dynamic network within mitochondria (i.e.

(Harbauer et al., 2014; Wiedemann & Pfanner, 2017). This study brings further evidence that protein biogenesis is linked to quality control and membrane morphology. The fact that the translocase and YME1L are functionally linked is present in multiple kingdoms; human YME1L degrades TIM23, TIM17A and ROMO1 (Rainbolt et al., 2013; Wai et al., 2016; this study); yeast Yme1 degrades the small Tims, Tim9 and Tim10 (Baker et al., 2012) and Tom22 (Wu et al., 2018); and in plants, the i-AAA protease, FTSH4, degrades essential Tim17-2 (Opalińska et al., 2018).

In general, it seems feasible that many processes are not separate and independent but rather connected functionally, as well as physically. Partly, due to its high abundance (Morgenstern et al., 2017), the TIM23 complex can form a hub and it has already been shown to not only interact with components of other import pathways (Albrecht et al., 2006; Chacinska et al., 2010; Gold et al., 2014; Waegemann et al., 2015), but also the respiratory chain (van der Laan et al., 2006;

Wiedemann et al., 2007) and the ADP/ATP carrier (Dienhart & Stuart, 2008; Mehnert et al., 2014). These interactions make the import process more efficient and link protein import to bioenergetics. In human, the presequence translocase is also connected to respiratory chain assembly and mitochondrial translation by a complex called MITRAC (Dennerlein et al., 2015;

Mick et al., 2012; Richter-Dennerlein et al., 2016).

In this thesis, novel evidence is provided that the TIM23 complex is also connected to protein processing and quality control in the inner membrane. These networks between different mitochondrial processes might be especially valuable to mediate a quick response to stress or pathological conditions. The exact implications will have to be addressed by further studies.