The list of identified Doa10 substrates that are membrane proteins has been grow-ing in the last years. Such membrane-spanngrow-ing Doa10 substrates have expanded our knowledge of the function of Doa10-mediated ERAD. A defect in biosynthesis per se is hereby not the only signal for degradation. Instead, Doa10-mediated ERAD also occurs in a regulated manner as for squalene monooxygenase which is regulated by intermediates of the sterol synthesis pathway (Foresti et al., 2013). The tail-anchored
protein Sbh2 is degraded upon deletion of its interaction partner, the translocon sub-unit Ssh1 (Habeck et al., 2015). For Sbh2 it has been shown that its TM anchor is required for its degradation. The components required for recognition, ubiquitination as well as retrotranslocation of a substrate containing such an intramembrane degron could differ from substrates containing a soluble degron such as Deg1. As screens have been only carried out for such soluble substrates, we performed a screen for components required for degradation of a protein containing a TM domain. We chose the substrate Sbh2 for our screen as it is a well-established substrate of Doa10.
7.5.1 Tandem fluorescent timer screens are a useful screening tool
Different screening approaches exist to identify components required for degradation of a protein of interest. These approaches are often based on fusing a reporter pro-tein to a propro-tein of interest. Many reporter propro-teins are enzymes whose activity is used to measure the stability of the fusion protein. One example for a reporter pro-tein is Ura3, an enzyme involved in uracil biosynthesis. Stabilized substrate allows growth of an auxotrophic strain on medium lacking uracil. Another example is beta-galactosidase that cleaves the added chromogenic compound X-Gal. Both of these ap-proaches have been used to characterize the degradation machinery of Deg1-containing fusion proteins (Johnson et al., 1998; Swanson et al., 2001). Recently, a quite sensitive screening method has been developed by the Knop lab termed tandem fluorescent timer screen (Khmelinskii et al., 2012). In this method, two fluorescent proteins with different maturation kinetics are fused to the substrate and thus a readout for protein abundance as well as stability can be provided. Moreover, the fluorescent labeling allows for mea-suring of not only whole colonies but also single cells. A microscopy-based approach is hereby useful, as the cellular localization of the protein in every sample can be analyzed as well. Moreover, by using the intensity of the fast maturing fluorescent protein, the intensity of the slowly maturing fluorescent protein can be normalized. This eliminates effects on protein abundance during protein synthesis and allows to specifically charac-terize protein degradation kinetics. Tandem fluorescent timer screens have been used to identify components of the N-end rule pathway as well as components required for degradation of a mislocalized tail-anchored protein (Khmelinskii et al., 2012; Dederer et al., 2019).
7.5.2 Screening for components required for degradation of Sbh2
Using this screen, we have identified known components of the ERAD machinery (Fig-ure 6.3). We did not identify any other gene deletion as a similarly strong hit, suggesting
that all components required for ERAD of Sbh2 might be identified. This is in agree-ment with a previous study by Carvalho et al. (2006) that has identified components of the Doa10 complex by pulldown of Doa10 and subsequent mass-spectrometric anal-ysis. The results from the screen suggest that all components required for ERAD of a protein containing an intramembrane degron are identified. Moreover, by establishing a reconstituted system, I characterized a minimal machinery for ubiquitination of Sbh2 (see section 7.5.4).
7.5.3 Potential role of the deubiquitinase Ubp3
Interestingly, the deubiquitinase Ubp3 appeared as a potential hit in our screen. I verified these results by creating a UBP3 deletion. I was able to confirm the results from the screen using a tFT-construct of Sbh2. However, when I measured the degradation kinetics of Sbh2 using a CHX-chase, I did not observe any strong impairment of Sbh2 turnover (Figure 6.4). This could be a result of tagging, as the tFT-construct contains a GFP-mCherry tag, whereas the CHX-chase construct contains only an HA-tag. The GFP-mCherry tag could create an artefact in ∆ubp3 cells. However, an alternative explanation is that the CHX-chase is not suitable to report on Sbh2 turnover, as the UBP3 deletion strain grew very slowly and this might have affected the experiment.
Whereas strains from an overnight culture were used for measuring the tFT-construct, exponentially growing cells were used in the CHX-chase. Moreover, due to the fast degradation kinetics of Sbh2, the timepoints chosen were not useful to characterize degradation kinetics. Thus, further experiments are required to determine if Ubp3 is involved in ERAD.
What would be a potential role of Ubp3 in ERAD? Deletion of UBP3 stabilizes Sbh2 indicating that Ubp3 promotes ERAD and does not compete with it. It would therefore not be involved in processes such as substrate deubiquitination. This deu-biquitinase could either act on Sbh2 downstream of Cdc48 by for example promoting release of substrate from Cdc48. It has been shown that deubiquitination occurs af-ter Cdc48 action and is required for substrate release from Cdc48 (Stein et al., 2014;
Bodnar and Rapoport, 2017b). Whereas the deubiquitinase Otu1 can carry out this function in a reconstituted system, its deletion does not impact ERAD inS. cerevisiae, suggesting the presence of multiple deubiquitinases (Stein et al., 2014). Interestingly, Ubp3 interacts with Cdc48 (Ossareh-Nazari et al., 2010a) and deletion of UBP3 sup-presses the temperature-sensitive lethality of the npl4-1 mutant allele (Auld et al., 2006), suggesting that Ubp3 is involved in Cdc48-associated processes.
Alternatively, Ubp3 acts on Doa10 itself. An interesting hypothesis hereby is that
Ubp3 deubiquitinates autoubiquitinated Doa10 and thus ensures that Doa10 is not subjected to proteasomal degradation. Autoubiquitination of E3-ligases is often used as a readout for E2 activity in vitro. However the relevance of E3 autoubiquitination in vivois for most E3 ligases not clear. Some examples exist where autoubiquitination leads to either proteasomal degradation or catalytic activation of E3 ligases (reviewed by Deshaies and Joazeiro (2009)). Recent studies indicate that autoubiquitination of Hrd1 has a role in ERAD. Although Hrd1 is stable in wildtype cells, some level of autoubiquitination of Hrd1 can be detected in intact cells (Baldridge and Rapoport, 2016). Upon mutation of certain lysine residues in Hrd1, substrate degradation is com-promised, indicating that autoubiquitination is important for Hrd1 function (Baldridge and Rapoport, 2016).
In contrast to Hrd1, it is unclear if autoubiquitination plays a role for Doa10 func-tion. InS. cerevisiae, Doa10 is stable as analyzed by pulse-chase experiments (Zattas et al., 2016). However in the mammalian system, the Doa10 homolog MARCH6 is unstable and its degradation depends on its own activitiy indicating that autoubiquiti-nation of MARCH6 leads to its degradation (Hassink et al., 2005; Zattas et al., 2016).
Our reconstituted system shows that Doa10 is autoubiquitinated in the presence of Ubc6 and Ubc7/Cue1 (Figure A2F and A2G). When Doa10 is autoubiquitinated, it gets extracted by the Cdc48-complex to some extent (Figure A3J). Interestingly, the autoubiquitination activity of Doa10 seems to be not required for polyubiquitination or extraction of Ubc6, as Ub-Ubc6C87A is ubiquitinated and extracted, in the absence of Doa10 autoubiquitination due to the absence of Ubc6 activity (Figure 3.3F and 3.3G). Moreover, Doa10-mediated retrotranslocation of Ubc6 occurs in the absence of ubiquitinaton machinery and thus also does not require Doa10 autoubiquitination.
Concluding, autoubiquitination of Doa10 might be a side-product of its activity. To avoid proteasomal degradation in vivo, deubiquitination of Doa10, and thus Ubp3 might play a role.
As we did not identify any new component strongly required for degradation of Sbh2, I set out to determine the minimal machinery for ERAD of Sbh2. Therefore, I established a reconstituted system using purified components reconstituted into lipo-somes.
7.5.4 Minimal machinery for ubiquitination of Sbh2
In a reconstituted system, Uba1, Ubc6, Ubc7/Cue1 and Doa10 are sufficient to cat-alyze polyubiquitination of Sbh2 (Figure 6.8). My results further indicate that both Ubc6 and Ubc7 are involved in ubiquitination of Sbh2, as efficient polyubiquitination
only occurs in the presence of both E2 enzymes. However, these ubiquitination ex-periments were done with a soluble fragment of Cue1 which reduces the efficiency of polyubiquitination (data not shown), and thus the difference might be less strong when full-length Cue1 is used. Nevertheless, both Ubc6 and Ubc7 seem to have a role in ubiquitination of Sbh2 in vitro, in agreement with stabilization of Sbh2 when UBC6 or UBC7 is deleted in vivo (Habeck et al., 2015).
7.5.5 Ubc6 and Ubc7 have different functions in Doa10-mediated ERAD Multiple lines of evidence indicate that Ubc6 and Ubc7 have different functions in Doa10-mediated ubiquitination. Weber et al. reconstituted autoubiquitination of the Doa10 RING domain by Uba1, Ubc7 and soluble fragments of Ubc6 and Cue1 (Weber et al., 2016). The authors show that Ubc6 initiates a ubiquitin chain by attaching a single ubiquitin and Ubc7 is involved in forming a polyubiquitin chain. Also in our system, I observe that Ubc6 catalyzes monoubiquitination of itself, Doa10 and proba-bly Sbh2 (Figures 3.1, A2F, A2G and 6.8). This monoubiquitination can happen on multiple residues, as I observe no difference of the ubiquitination pattern of Ubc6 when a ubiquitin variant is used in which all lysines are mutated to arginines (Figure A1L).
Only in the presence of Ubc7/Cue1, polyubiquitin chains are formed on Ubc6, Doa10 and probably Sbh2. This task division of E2 enzymes is a common feature and probably serves efficient ubiquitination (Deshaies and Joazeiro, 2009). Sequential ubiquitination has also been observed for the E2 enzymes Ubc1 and Ubc4 that function with APC/C.
Whereas Ubc4 initiates a ubiquitin chain, Ubc1 elongates this chain (Rodrigo-Brenni and Morgan, 2007). The human E2 enzyme Ubc13 (heterodimer with Msm2) only catalyzes polyubiquitination of its E3 ligase Brca1, when the mono-ubiquitinating E2 Ube2w is present (Christensen et al., 2007). Thus, E2 enzymes can be specific for chain initiation and elongation. Such a task divison also occurs for the Ubc6/Ubc7 pair.